N.C. State Conference of the Naacp v. McCrory

Full title: NORTH CAROLINA STATE CONFERENCE OF THE NAACP; EMMANUEL BAPTIST CHURCH…

Court: UNITED STATES DISTRICT COURT FOR THE MIDDLE DISTRICT OF NORTH CAROLINA

Date published: Apr 25, 2016

182 F. Supp. 3d 320 (M.D.N.C. 2016)

Opinion

1:13CV6581:13CV6601:13CV861

2016-04-25

NORTH CAROLINA STATE CONFERENCE OF THE NAACP; Emmanuel Baptist Church; Covenant Presbyterian Church; Barbee's Chapel Missionary Baptist Church, Inc.; Rosanell Eaton; Armenta Eaton; Carolyn Coleman; Jocelyn Ferguson-Kelly; Faith Jackson; Mary Perry; and Maria Teresa Unger Palmer, Plaintiffs, v. Patrick Lloyd MCCRORY, in his Official Capacity as Governor of North Carolina; Kim Westbrook Strach, in her Official Capacity as Executive Director of The North Carolina State Board of Elections; Rhonda K. Amoroso, in her Official Capacity as Secretary of The North Carolina State Board of Elections; Joshua D. Malcolm, in his Official Capacity as a Member of The North Carolina State Board of Elections; James Baker, in his Official Capacity as a Member of The North Carolina State Board of Elections; and Maja Kricker, in her Official Capacity as a Member of The North Carolina State Board of Elections, Defendants. League of Women Voters of North Carolina; A. Philip Randolph Institute; Unifour Onestop Collaboarative; Common Cause North Carolina; Goldie Wells; Kay Brandon; Octavia Rainey; Sara Stohler; and Hugh Stohler, Plaintiffs, and Louis M. Duke; Asgod Barrantes; Josue E. Berduo; Charles M. Gray ; Nancy J. Lund; Brian M. Miller; Becky Hurley Mock; Mary-Wren Ritchie; Lynne M. Walter; and Ebony N. West, Plaintiff-intervenors, v. The State of North Carolina; Joshua B. Howard, in his official capacity as a member of the State Board of Elections; Rhonda K. Amoroso, in her official capacity as a member of the State Board of Elections; Joshua D. Malcolm, in his official capacity as a member of the State Board of Elections; Paul J. Foley, in his official capacity as a member of the State Board of Elections; Maja Kricker, in her official capacity as a member of the State Board of Elections; and Patrick L. McCrory, in his official capacity as the Governor of the State of North Carolina, Defendants. United States of America, Plaintiff, v. The State of North Carolina; The North Carolina State Board of Elections; and Kim W. Strach, in her Official Capacity as Executive Director of the North Carolina State Board of Elections, Defendants.

Bridget K. O'Connor, Christopher J. Maner, Daniel T. Donovan, Jodi K. Wu, Kenneth Winn Allen, Michael A. Glick, Ronald K. Anguas, Jr., Susan Marie Davies, Thomas D. Yannucci, Kirkland & Ellis, LLP, Caitlin Swain-McSurely, Denise D. Lieberman, Donita Judge, Jadine C. Johnson, Jasmyn G. Richardson, Penda Denise Hair, Advancement Project, Washington, DC, Irving Joyner, N.C. Central University School of Law, Cary, NC, Jennifer R. Basch, John J. Song, Madelyn A. Morris, Kirkland & Ellis, LLP, New York, NY, Adam Stein, Tin Fulton Walker & Owen, PLLC, Chapel Hill, NC, for Plaintiffs. Karl S. Bowers, Jr., Bowers Law Office LLC, Columbia, SC, Alexander McClure Peters, Katherine A. Murphy, N.C. Department of Justice, Phillip John Strach, Thomas A. Farr, Michael Douglas McKnight, Ogletree Deakins Nash Smoak & Stewart, P.C., Raleigh, NC, Amy M. Pocklington, Ogletree, Deakins, Nash, Smoak & Stewart, P.C., Richmond, VA, Elizabeth R. Dangel, Ogletree, Deakins, Nash, Smoak & Stewart, P.C., Charlotte, NC, for Defendants.

Thomas D. Schroeder, United States District Judge

Bridget K. O'Connor, Christopher J. Maner, Daniel T. Donovan, Jodi K. Wu, Kenneth Winn Allen, Michael A. Glick, Ronald K. Anguas, Jr., Susan Marie Davies, Thomas D. Yannucci, Kirkland & Ellis, LLP, Caitlin Swain-McSurely, Denise D. Lieberman, Donita Judge, Jadine C. Johnson, Jasmyn G. Richardson, Penda Denise Hair, Advancement Project, Washington, DC, Irving Joyner, N.C. Central University School of Law, Cary, NC, Jennifer R. Basch, John J. Song, Madelyn A. Morris, Kirkland & Ellis, LLP, New York, NY, Adam Stein, Tin Fulton Walker & Owen, PLLC, Chapel Hill, NC, for Plaintiffs.

Karl S. Bowers, Jr., Bowers Law Office LLC, Columbia, SC, Alexander McClure Peters, Katherine A. Murphy, N.C. Department of Justice, Phillip John Strach, Thomas A. Farr, Michael Douglas McKnight, Ogletree Deakins Nash Smoak & Stewart, P.C., Raleigh, NC, Amy M. Pocklington, Ogletree, Deakins, Nash, Smoak & Stewart, P.C., Richmond, VA, Elizabeth R. Dangel, Ogletree, Deakins, Nash, Smoak & Stewart, P.C., Charlotte, NC, for Defendants.

MEMORANDUM OPINION AND ORDER

Thomas D. Schroeder, United States District Judge

TABLE OF CONTENTS

I. FINDINGS OF FACT ... 332

A. North Carolina Voting Laws ...332

1. Voter ID ... 332

2. Early Voting ...332

3. Out-of-Precinct Provisional Voting ...334

4. SDR ... 336

5. Pre-registration ... 336

B. Post-2011 Legislation ... 337

1. Introduction of HB 589 ... 337

2. Revision of HB 589 ... 339

3. Enactment of HB 836 ... 344

C. Procedural History ... 348

D. Evidence of Voter Experience Under Current Law ... 351

1. Voter ID ... 351

a. Voter Education about the Voter-ID Requirement Prior to the Reasonable Impediment Exception ... 351

b. Voter Education After Enactment of the Reasonable Impediment Exception ... 353

c. Voters' Experience in Acquiring Qualifying ID ... 358

d. Evidence of North Carolina Voters Without ID ... 364

e. Availability of the Reasonable Impediment Exception ... 369

2. Change in the Early-Voting Schedule ... 382

3. Elimination of SDR ... 393

4. Elimination of OOP Provisional Voting ... 404

5. Elimination of Pre-Registration ... 407

6. Other Challenged Provisions ... 409

7. 2014 Data ... 411

E. Testimony of Other Experts ... 412

II. CONCLUSIONS OF LAW ... 412

A. Section 2 of the VRA ... 412

1. The Law of Vote Denial and Abridgement Claims. ... 412

2. The Totality of the Circumstances & Gingles ... 422

a. The Success of the Prior Practices in Fostering Minority Political Participation ... 422

b. History of Official Discrimination ... 425

c. Racially-Polarized Voting ... 429

d. Enhancing the Opportunity for Discrimination ... 429

e. Candidate Slating Process ... 430

f. Continuing Effects of Discrimination Hindering Participation ... 430

g. Racial Appeals in Campaigning ... 438

h. Minority Electoral Success ... 438

i. Responsiveness of Elected Officials ... 439

j. Tenuousness of the State's Justifications ... 440

i. Voter ID ... 440

ii. Early Voting ... 445

iii. SDR ... 445

iv. OOP Voting ... 447

v. Pre-Registration ... 463

3. Equality of Opportunity and Social and Historical Conditions ... 465

a. Voter ID ... 466

b. Early Voting ... 472

c. SDR ... 473

d. OOP Voting ... 479

e. Pre-registration ... 483

f. Cumulative Effect ... 484

4. Discriminatory Result: Conclusion ... 486

5. Discriminatory Intent ... 488

6. Additional Problems with the § 2 Results Claim ... 503

B. "Traditional" Fourteenth and Fifteenth Amendment Claims ... 509

C. Anderson–Burdick Claim ... 509

1. Voter ID ... 511

2. Early Voting ... 513

3. SDR ... 514

4. OOP ... 516

5. Pre-registration ... 518

6. CBOE Discretion ... 520

7. Poll Observers and Challengers ... 520

8. Cumulative Effect of Provisions ... 521

D. Twenty-Sixth Amendment Claim ... 521

E. Remedy ... 525

III. CONCLUSION ... 526

In these related cases, Plaintiffs seek to permanently enjoin Defendants from implementing various provisions of North Carolina Session Law 2013–381 ("SL 2013–381"), an omnibus election-reform law, as amended by Session Law 2015–103 ("SL 2015–103").¹Plaintiffs are the United States of America (the "United States") in case 1:13CV861, the North Carolina State Conference of the NAACP and several organizations and individual plaintiffs (the "NAACP Plaintiffs") in case 1:13CV658, and the League of Women Voters of North Carolina along with several organizations and individuals (the "League Plaintiffs") in case 1:13CV660. Additionally, the court allowed a group of "young voters" and others (the "Intervenor Plaintiffs") to intervene in case 1:13CV660. (Doc. 62 in case 1:13CV660.) Considered together, Plaintiffs raise claims under the Fourteenth, Fifteenth, and Twenty-Sixth Amendments to the United States Constitution as well as § 2 of the Voting Rights Act of 1965 ("VRA"), 52 U.S.C. § 10301 (formerly 42 U.S.C. § 1973 ). (Doc. 365 in case 1:13CV861; Doc. 384 in case 1:13CV658; Docs. 1 & 63 in case 1:13CV660.) The United States also moves for the appointment of federal observers to monitor future elections in North Carolina pursuant to § 3(a) of the VRA, 52 U.S.C. § 10302(a) (formerly 42 U.S.C. § 1973a(a) ). (Doc. 365 at 33.)² Defendants are the State of North Carolina, Governor Patrick L. McCrory, the State Board of Elections ("SBOE"), and several State officials acting in their official capacities.

1 The parties sometimes refer to the challenged law as "House Bill 589," its original designation by the North Carolina General Assembly. The final product, as a duly-enacted law passed by both chambers of the General Assembly and signed by the governor, will be referred to as Session Law 2013–381. Prior to passage, the bill will be referred to as HB 589.

2 Because of the duplicative nature of the filings in these three cases, the court will refer only to the record in case 1:13CV861 except where necessary to distinguish the cases. Where the court has cited to docket entries, (e.g., Doc. 346 (Plaintiffs' proposed conclusions of law and findings of fact)), pinpoint cites are to the CM/ECF pagination. Where the court cites to exhibits by the parties, pinpoint cites are to the exhibit's internal pagination where possible. This includes exhibits containing deposition designations.

The record is extensive. The court held a four-day evidentiary hearing and argument beginning July 7, 2014, on Plaintiffs' motion for preliminary injunction, which evidence is now part of the trial record. Fed. R. Civ. P. 65(a)(2). Fifteen days of trial on the merits were conducted from July 13 through 31, 2015. An additional six days of trial on the voter photo identification ("ID") provisions of the law were conducted from January 25 through February 1, 2016. The court has considered testimony of twenty-one expert witnesses and 112 fact witnesses. The record consists of more than 11,000 pages from the preliminary injunction phase, in excess of 12,000 pages from the July trial, and over 2,500 additional pages from the January trial.³ As can be seen from the length of this memorandum opinion, merely trying to concisely state the court's findings has presented a monumental challenge.

3 This includes all expert reports, to which the parties waived hearsay objections, but does not include the trial transcript or Plaintiffs' Exhibit 646, which is a database with 39,912 pages.

This case presents important questions as it tests North Carolina's newly-enacted voter photo-ID requirement and the State's modification or elimination of certain voting procedures not contemplated by the State a little more than a decade ago: seventeen days of in-person early voting before Election Day, same-day registration, voting provisionally on Election Day in an unassigned precinct, and pre-registering to vote as early as age sixteen. Under both the Elections Clause of, and the Tenth Amendment to, the United States Constitution, such decisions are traditionally reserved to the States, but they are subject to other constitutional and congressional limitations. The principal question in these cases is whether the North Carolina General Assembly imposed a voter-ID requirement and altered these relatively recently-developed voting procedures—deemed "conveniences" and "fail-safes" by some of Plaintiffs' own experts—based on race or, even if not, in a manner that presents an unlawful discriminatory burden on voters.

After careful consideration of the complete record and pursuant to Rule 52(a) of the Federal Rules of Civil Procedure, the court enters the following findings of fact—based upon an evaluation of the evidence, including the credibility of witnesses, and the inferences that the court has found reasonable to be drawn therefrom—and conclusions of law. To the extent any factual statement is contained in the conclusions of law, it is deemed a finding of fact as well.

I. FINDINGS OF FACT

A. North Carolina Voting Laws

The provisions of North Carolina SL 2013–381 at issue establish a voter-ID requirement and repeal certain voting and registration mechanisms enacted since 1999. An understanding of the purposes and effect of the current regime requires an understanding of the previous laws, including their origin and history. SeeLeague of Women Voters of N.C. v. North Carolina, 769 F.3d 224, 242 (4th Cir.2014) ("League") (noting that "North Carolina's previous voting practices are centrally relevant"). Each modified or removed voting and registration mechanism was enacted while Democrats controlled both houses of North Carolina's General Assembly and its governorship, which they held until 2011. Because North Carolina was a covered jurisdiction under § 5 of the VRA, each change required approval by the United States Department of Justice ("DOJ").

1. Voter ID

Prior to 2016, North Carolina relied on a system of signature attestation to prevent voter fraud. Under this system, poll workers—as the primary gatekeepers to voter fraud—would ask the name and address of voters presenting to vote in person. (Doc. 407 at 43.) If the poll worker was able to locate a registration for the name and address provided, the voter was required to sign an authorization to vote ("ATV") form attesting that he was the person under whose registration he sought to vote and that he currently resided at the address of registration. (Doc. 410 at 83; Plaintiffs' Exhibit ("Pl. Ex.") 1056.) The ATV form warned voters that "fraudulently or falsely completing this form is a Class I Felony." (Pl. Ex. 1056.) Although the SBOE maintained voters' signatures as a result of registration forms, (Pl. Ex. 212A), poll workers did not have access to the signatures, either during early voting or on Election Day, (Doc. 414 at 123). Accordingly, signatures were not verified at the polling place and, unless the poll worker knew the voter, the poll worker had very limited means of determining whether the voter was the same person as the registrant. (See id.)

2. Early Voting

Prior to 1973, North Carolina required all voters to cast their ballot on Election Day or to apply for an absentee ballot. SeeN.C. Gen. Stat. § 163–227 (1972). In 1973, the General Assembly passed legislation that permitted voters to participate in "one-stop" "early voting"⁴ for a period of sixty days before Election Day, but only if they provided a statutorily-acceptable excuse (e.g., absence from the county, sickness, or disability) and obtained their ballot from the county board of election ("CBOE").⁵ 1973 N.C. Sess. Law 536, § 1.

4 "One-stop" refers to the procedure allowing voters to request and cast an absentee ballot at the same time. "Early voting" describes in-person absentee voting at designated locations before Election Day. Absentee mail-in voting, like early in-person voting, is a form of "absentee voting." Even when a voter shows up in person, he is simply applying for and completing an absentee voting application and ballot at the same time. Mail-in voting breaks this into two steps: the voter applies for a no-excuse absentee ballot, and, after the local CBOE mails it to the voter, the voter returns the completed absentee ballot to the CBOE, either by mail or in person.

5 Unlike the form of absentee voting that existed prior to 1973, all actions necessary for the one-stop early-voting ballot had to "be performed in the office of the board of elections." 1973 N.C. Sess. Law 536, § 1.

In 1979, the General Assembly reduced the one-stop early-voting period from sixty days to thirty days. 1979 N.C. Sess. Law 799, § 1. But seeN.C. Gen. Stat. § 163–227.3 (providing that, unless otherwise authorized, a CBOE shall provide absentee ballots for voting by mail "60 days prior to the statewide general election in even-numbered years"). Then, as now, a voter had to be registered at least twenty-five days before the election for which the absentee ballot was being offered. SeeN.C. Gen. Stat. § 163–82.6(c) ; N.C. Gen. Stat. § 163–67 (1979) (making the registration cut-off twenty-one days before Election Day, excluding Saturdays and Sundays). This law provided that a ballot executed at a CBOE be completed in a voting booth or private room. 1979 N.C. Sess. Law 799, § 2.

In 1999 (effective January 1, 2000), on a vote almost entirely along party lines,⁶ the General Assembly removed the excuse requirement for "one-stop" voting in North Carolina's even-year general elections, thus establishing "no-excuse" early voting. 1999 N.C. Sess. Law 455, §§ 1, 6; (Pl. Ex. 46 at 25 (chronicling partisan voting)). It also permitted a CBOE, upon unanimous CBOE vote and the approval of the SBOE, to open additional early-voting sites beyond the one site at the CBOE. 1999 N.C. Sess. Law 455, § 6. Thus, a registered voter could present herself at the CBOE or another designated site in her county of residence "[n]ot earlier than the first business day after the twenty-fifth day before an election ... and not later than 5:00 p.m. on the Friday prior to that election" to cast her ballot. Id. Because the law permitted only weekday operations, see id.(amending N.C. Gen. Stat. § 163–227.2(f) ), this offered registered voters fifteen days of early voting, id.

6 The vote in the Senate was 36–10, with four Republicans voting with the majority, and 60–53 in the House with all Republicans joined by one Democrat in opposition. (Pl. Ex. 46 at 25.) The court may take judicial notice of the legislative history of the laws at issue. SeeFed. R. Evid. 201 ; see e.g., Hall v. Louisiana, No. 12–00657, 2015 WL 1383532, at *3 (M.D.La. Mar. 23, 2015) (citing Territory of Alaska v. Am. Can Co., 358 U.S. 224, 226–27, 79 S.Ct. 274, 3 L.Ed.2d 257 (1959) ) (taking judicial notice of the legislative history of a bill in a voting rights case). Plaintiffs ignore this history of North Carolina's early voting, especially the partisan opposition to changing early voting to "no-excuse." (Cf.Doc. 346 at 22 (noting simply that later amendments to early voting received "substantial bipartisan support")).

In 2000, the General Assembly enacted SL 2000–136, which allowed CBOEs to petition the SBOE for approval when they are unable to reach unanimous agreement as to the location of additional early-voting sites. 2000 N.C. Sess. Law 136, § 2. The law empowered the SBOE, on a simple majority vote, to approve additional sites based on the consideration of the "partisan interests of that county," among other factors. Id.(not requiring SBOE unanimity). Because the governor controls appointments to the SBOE, which in turn appoints the members of the CBOEs, both boards are effectively controlled by the same political party as the governor. SeeN.C. Gen. Stat. § 163–19 (giving the governor power to appoint SBOE members but requiring that "[n]ot more than three members of the [five-member] Board shall be members of the same political party"); N.C. Gen. Stat. § 163–30 (providing appointment power of CBOE members to the SBOE and requiring that "[n]ot more than two members of the [three-member CBOE] shall belong to the same political party").⁷ Thus, this change injected partisan considerations into the location of additional early-voting sites.

7 As detailed infra, the trial evidence demonstrated that this resulted in early-voting sites being situated in areas more favorable to persons who tended to vote for Democratic candidates. (See, e.g., Defendants' Exhibit ("Def. Ex.") 212A at 14–16, 19 (finding that the placement of and hours offered for early-voting sites favor Democrats over Republicans).)

In 2001, the General Assembly expanded no-excuse early voting to all elections and absentee ballots. 2001 N.C. Sess. Law 337, § 1. With votes split largely along party lines in the North Carolina House of Representatives, but with bi-partisan support in the Senate,⁸ the General Assembly also amended the early-voting period so that voters could appear at the CBOE office to vote "[n]ot earlier than the third Thursday before an election ... and not later than 1:00 P.M. on the last Saturday before that election." 2001 N.C. Sess. Law 319, § 5(a). Under this revision, CBOEs were granted discretion to extend the closing time on that final Saturday to 5:00 p.m. and, upon unanimous agreement (or in its absence, upon approval of the SBOE), to maintain early-voting hours during the evening or on weekends throughout the early-voting period.⁹Id.§ 5(b).

8 The final House vote was 60–54, with six Democrats not voting but four Republicans joining the Democratic majority. (Pl. Ex. 47 at 16.) The Senate's final vote was 46–2. (Id.)

9 CBOEs remained free to open additional early-voting sites other than the CBOE office by unanimous vote or, in its absence, upon approval of the SBOE upon petition. N.C. Gen. Stat. § 163–227.2(g) (2001).

In sum, these 2001 changes, effective January 1, 2002, moved the start of early voting three days closer to Election Day, reduced the number of required days of early voting to twelve and one-half days, but permitted an expansion up to seventeen days upon unanimous CBOE agreement. No one criticized or challenged the reduced minimum or other changes.

3. Out-of-Precinct Provisional Voting

The next voting change, chronologically, was the advent of out-of-precinct ("OOP") provisional voting, whose origins in North Carolina can be traced to Congress' passage in 2002 of the Help America Vote Act ("HAVA"), 52 U.S.C. §§ 20901 –21145 (formerly 42 U.S.C. §§ 15301 –15545 ), which in turn was passed in the wake of evidence of irregularities in the 2000 presidential election. HAVA, in part, required States to offer provisional ballots to individuals on Election Day who seek to vote and claim to be registered and eligible to vote for federal office, but who do "not appear on the official list of eligible voters for the polling place or an election official asserts that the individual is not eligible to vote." See52 U.S.C. § 21082(a). However, HAVA only requires such provisional ballots to be counted "in accordance with State law." Id.§ 21082(a)(4). Thus, a provisional ballot must be counted only if State law authorizes it.

In 2003, a bill was introduced in the General Assembly titled, "Help America Vote Act Compliance." H.B. 842, 2003 Gen. Assemb., Reg. Sess. (N.C. 2003). Its stated purpose was "to ensure that the State of North Carolina has a system for all North Carolina elections that complies with the requirements for federal elections set forth in" HAVA. See 2003 N.C. Sess. Law 226, § 1. It was approved unanimously. As to provisional ballots specifically, the law provided that the CBOE shall count a provisional ballot "for all ballot items on which it determines that the individual was eligible under State or federal law to vote."Id.§ 15(5).

Soon after, the SBOE claimed authority to count provisional ballots cast outside the voter's correct precinct, and several affected Republican candidates raised a legal challenge. SeeIn re Election Protest of Fletcher, 175 N.C.App. 755, 756, 625 S.E.2d 564, 565 (2006) (noting the challenger's party affiliation). The North Carolina Supreme Court unanimously held that the counting of such ballots violated State law and SBOE regulations, which required voters to cast ballots in their assigned precinct. James v. Bartlett, 359 N.C. 260, 267–70, 607 S.E.2d 638, 642–44 (2005) ("The plain meaning of [N.C. Gen. Stat. § 163–55 (2003) ]¹⁰ is that voters must cast ballots on election day in their precincts of residence."). In reaching its decision, the North Carolina Supreme Court recognized several "advantages" of the precinct system and in-precinct voting, which it observed were "woven throughout the fabric of [the State's] election laws," id. at 267, 607 S.E.2d at 642 (citing statutes), including that

10 N.C. Gen. Stat. § 163–55 provided (emphasis added):

Every person born in the United States, and every person who has been naturalized, and who shall have residedin the State of North Carolina and in the precinctin which he offers to register and vote for 30 days next preceding the ensuing election, shall, if otherwise qualified as prescribed in this Chapter, be qualified to register and vote in the precinct in which he resides:Provided, that removal from one precinctto another in this State shall not operate to deprive any person of the right to vote in the precinctfrom which he has removed until 30 days after his removal.

it caps the number of voters attempting to vote in the same place on election day; it allows each precinct ballot to list all of the votes a citizen may cast for all pertinent federal, state, and local elections, referenda, initiatives, and levies; it allows each precinct ballot to list only those votes a citizen may cast, making ballots less confusing; it makes it easier for election officials to monitor votes and prevent election fraud; and it generally puts polling places in closer proximity to voter residences.

Id. at 271, 607 S.E.2d at 644–45 (quoting Sandusky Cty. Democratic Party v. Blackwell, 387 F.3d 565, 569 (6th Cir.2004) (per curiam)). The court also noted:

If voters could simply appear at any precinct to cast their ballot, there would be no way under the present system to conduct elections without overwhelming delays, mass confusion, and the potential for fraud that robs the validity and integrity of our elections process.

Id. at 270, 607 S.E.2d at 644. The North Carolina Supreme Court found that "it is but a perfunctory requirement that voters identify their proper precinct and appear within that precinct on election day to cast their ballots." Id. at 271, 607 S.E.2d at 645.

In response to James, the General Assembly—on a purely partisan division—immediately passed SL 2005–2,¹¹ amending N.C. Gen. Stat. § 163–55 to remove the requirement that voters appear in the proper precinct on Election Day in order to vote. 2005 N.C. Sess. Law 2, § 2. The General Assembly went further, however, to require that the law apply retroactively to the 2004 election, thus ensuring electoral victory for the Democratic candidates in the elections challenged in James. Id.§§ 1–14. And as "extra insurance" against judicial intervention, the Democratic majority put in place a procedure mandating that the legislature—and not the courts—would decide contested elections for State-wide offices. (Pl. Ex. 46 at 30.) The General Assembly also placed in the law a finding that it had "take[n] note" that African Americans disproportionately used OOP voting on Election Day in November 2004. 2005 N.C. Sess. Law. 2, § 1.

11 The final votes were 29–21 in the Senate and 61–54 in the House. (Def. Ex. 168.)

4. SDR

The National Voter Registration Act ("NVRA"), 52 U.S.C. § 20507(a)(1) (formerly 42 U.S.C. § 1973gg–6(a)(1) ), permits a State to set a registration cut-off of thirty days before an election. North Carolina extends that deadline by five days such that a person is required to have registered to vote at least twenty-five days before an election in order to cast a ballot. N.C. Gen. Stat. § 163–82.6(c).

In July 2007, the General Assembly—split almost entirely along party lines—passed legislation permitting voters to register and vote at early-voting sites, which Governor Michael Easley signed into law.¹² The law provided that "an individual who is qualified to register to vote may register in person and then vote at [an early-voting] site in the person's county of residence during the period for [early] voting provided under [§] 163–227.2." 2007 N.C. Sess. Law 253, § 1. The law required a prospective voter to complete a voter registration form and produce documentary proof of her current name and address, either through a North Carolina driver's license, a photo ID from a government agency, or a HAVA document.¹³Id. If she elected to vote immediately, the voter could "vote a retrievable absentee ballot as provided in [§] 163–227.2 immediately after registering." Id. Within two business days, the CBOE was required, in conjunction with the SBOE, to verify the voter's driver's license or social security number ("SSN"), update the voting database, proceed to verify the voter's proper address, and count the vote unless the CBOE determined that the applicant was not qualified to vote in accordance with the provisions of that chapter. Id. As will be seen, this meant that, as a practical matter votes were counted even though the voter registered and voted at the CBOE too close to Election Day to permit the CBOE to comply with North Carolina's preexisting mail verification system for voter registration. SeeN.C. Gen. Stat. § 163–82.7.

12 The final votes were 69–47 in the House, with three Republicans joining the Democratic majority, (Def. Ex. 169), and 34–15 in the Senate, with four Republicans joining the Democratic majority, (Def. Ex. 170).

13 The following constitute a valid "HAVA document": "a current utility bill, bank statement, government check, paycheck, or other government document" showing the voter's "name and address." N.C. Gen. Stat. § 163–166.12(a).

5. Pre-registration

Ever since the ratification of the Twenty-Sixth Amendment in 1971, a person who would be eighteen years-old on the next Election Day could register to vote in North Carolina, which included the primary for that election even if he would not be eighteen on the date of the primary. N.C. Gen. Stat. §§ 163–55(a)(1), 163–59. In 1993, a bill to permit sixteen- and seventeen-year-olds to "preregister" was introduced but failed to gain passage. (Pl. Ex. 46 at 23.)

In 2009, a bipartisan General Assembly passed SL 2009–541, which allowed for "pre-registration" of sixteen- and seventeen-year-olds who would not be eighteen before the next general election.¹⁴ 2009 N.C. Sess. Law 541, § 7(a). With pre-registration, "[a] person who is at least 16 years of age but will not be 18 years of age by the date of the next election and who is otherwise qualified to register may preregister to vote and shall be automatically registered upon reaching the age of eligibility following verification of the person's qualifications and address in accordance with [§] 163–82.7." Id. Session Law 2009–541 also mandated that CBOEs conduct pre-registration drives. Id.§ 16(a).

14 The final votes were 32–3 in the Senate and 107–6 in the House. (Pl. Ex. 46 at 33.)

B. Post-2011 Legislation

In 2011, Republicans gained majorities in both houses of the General Assembly, yet the Democrats continued to control the governorship. With that shift, however, efforts to alter several of the recently-enacted election laws began. Those included bills to reduce early voting and end SDR and pre-registration. See e.g., H.B. 658, 2011 Gen. Assemb., Reg. Sess. (N.C. 2011) (reducing early voting by a week);¹⁵ S.B. 714, 2011 Gen. Assemb., Reg. Sess. (N.C. 2011) (requiring all satellite early-voting sites to remain open at least the same number of days per week and the same number of hours per day as the CBOE site); S.B. 657, 2011 Gen. Assemb., Reg. Sess. (N.C. 2011) (eliminating a week of early voting and any Sunday voting); S.B. 47, 2011 Gen. Assemb., Reg. Sess. (N.C. 2011) (eliminating a week of early voting, SDR, and straight-ticket voting); S.B. 657, 2011 Gen. Assemb., Reg. Sess. (N.C. 2011) (eliminating a week of early voting, Sunday voting, SDR, and pre-registration). Most prominent among those proposals was H.B. 351, 2011 Gen. Assemb., Reg. Sess. (N.C. 2011), entitled "Restore Confidence in Government," which introduced a photo-ID requirement. HB 351 was debated in and passed North Carolina's House and Senate, but was vetoed by Governor Beverly Perdue. (Pl. Ex. 46 at 35.) According to a Plaintiffs' expert in this case, "All of the votes were almost pure partisan splits." (Id.)

15 S.B. 658 failed when Democratic SBOE Executive Director Gary Bartlett issued a memorandum on the day of the third reading of the bill, stating that he believed that reducing early voting would actually cost more. (Pl. Ex. 46 at 35.)

In 2012, Republicans gained control of the governorship and, in 2013, control of both houses. After more than a century since controlling all offices, they renewed attempts to change North Carolina's election administration. In 2013, several bills were introduced to reduce the early-voting period, eliminate SDR, and alter other procedures. See, e.g., H.B. 913, 2013 Gen. Assemb., Reg. Sess. (N.C. 2013) (eliminating SDR and enhancing observer rights); S.B. 428, 2013 Gen. Assemb., Reg. Sess. (N.C. 2013) (eliminating a week of early voting and SDR); S.B. 666, 2013 Gen. Assemb., Reg. Sess. (N.C. 2013) (eliminating a week of early voting, weekend voting hours, and SDR); S.B. 721, 2013 Gen. Assemb., Reg. Sess. (N.C. 2013) (implementing voter ID and reducing early voting to six days); H.B. 451, 2013 Gen. Assemb., Reg. Sess. (N.C. 2013) (eliminating a week of early voting, Sunday voting, SDR, and straight-ticket voting).

1. Introduction of HB 589

On March 12, 2013, the legislative process for SL 2013–381 began, with the North Carolina House Committee on Elections, chaired by Republican Representative David R. Lewis, holding public hearings on voter ID. (See Pl. Ex. 127.) Over seventy-five citizens from a wide variety of organizations spoke before the committee. (Id. at 2–5.) The next day, the committee met and considered the testimony of five individuals representing a wide variety of organizations, including the Brennan Center for Justice and the Heritage Foundation. (See Pl. Ex. 128.) One of the speakers was Allison Riggs, counsel of record for the League Plaintiffs in case 1:13CV660, who appeared on behalf of the Southern Coalition for Social Justice. (Id. at 4.) On April 3, the committee heard from Ion Sancho, the Supervisor of Elections for Leon County, Florida, who testified about Florida's experience when it reduced early-voting days in advance of the 2012 general election. (Pl. Ex. 129 at 61–62, 69–70, 78–79.) Those public hearings were not required by the North Carolina House's internal rules. (Defendants' Exhibit (Def. Ex.) 217 at 3); see H.R. 54, 2013 Gen. Assemb., Reg. Sess. (N.C. 2013), http://www.ncleg.net/Sessions/2013/Bills/House/PDF/H54v3.pdf (last visited April 6, 2015) (hereinafter "H.R. 54").¹⁶

16 Fed. R. Evid. 201.

On April 4, HB 589 was introduced in the House. (Pl. Ex. 105.) The bill dealt mostly with the implementation of a voter-ID requirement beginning in 2016 in portions titled the "Voter Information Verification Act." (Id. at 1–6, 11–12.) The remainder of the bill dealt with the procedure for obtaining and voting mail-in absentee ballots. (Id. at 6–11.)

Under House rules, legislation must pass three readings.¹⁷ On April 8, the bill passed "first reading" and was referred to the House Committee on Elections. (Pl. Ex. 121.) The committee subsequently held another public hearing on April 10, during which over seventy-five citizens from across the political spectrum had the opportunity to speak. (Pl. Ex. 130.) That same day, the committee held another hearing during which the bill was discussed and additional public comments were received. (Pl. Ex. 545.) The committee further debated the bill and added amendments at a meeting held on April 17. (Pl. Ex. 546.) The bill was referred to the House Committees on Finance. (Pl. Ex. 121.)

17 House Rule 41(a) states, "Every bill shall receive three readings in the House prior to its passage. The first reading and reference to standing committee of a House bill shall occur on the next legislative day following its introduction." H.R. 54; Fed. R. Evid. 201.

HB 589 advanced, as amended, from the various House committees and was debated on the House floor on April 24. (Id.; Pl. Exs. 547, 548.) Three amendments were adopted, six others were rejected, and the bill passed "second reading" on a roll-call vote of 80–36.¹⁸ (Pl. Ex. 121; Pl. Ex. 548 at 177.) The bill subsequently passed "third reading," on a vote of 81–36, and was passed by the House. (Pl. Ex. 548 at 178.) Five House Democrats joined all present Republicans in voting for the voter-ID bill, (Pl. Ex. 122 (noting roll call vote on April 24 third reading); Pl. Ex. 138 at 67–68, 77, 88), but none of the African American members of the House supported it, (Pl. Exs. 122, 154). Representative Rick Glazier, who strongly opposed the bill, nevertheless acknowledged that "[f]or a large bill," HB 589 received up to this point "the best process possible" in the House, one he characterized as "excellent." (Doc. 165 at 56–57; see also Pl. Ex. 25 at 6.)

18 House Rule 41(b) states: "No bill shall be read more than once on the same day without the concurrence of two-thirds of the members present and voting ...." H.R. 54.

HB 589 was received in the North Carolina Senate the next day, passed first reading, and was assigned to the Senate Rules Committee. (Pl. Ex. 121.) The committee took no immediate action on the bill. The parties do not dispute that the Senate believed at this stage that HB 589 would have to be submitted to the DOJ for "pre-clearance" under § 5 of the VRA, 52 U.S.C. § 10304 (formerly 42 U.S.C. § 1973c ), because many North Carolina counties were "covered jurisdictions" under that section. At that time, however, the United States Supreme Court was considering a challenge to the DOJ's ability to enforce § 5. So, the bill sat.

On June 25, the Supreme Court issued its decision in Shelby County v. Holder, ––– U.S. ––––, 133 S.Ct. 2612, 186 L.Ed.2d 651 (2013), declaring the formula used to determine the § 5 covered jurisdictions, 52 U.S.C. § 10303(b) (formerly 42 U.S.C. § 1973b(b) ), to be unconstitutional. The next day, Senator Thomas Apodaca, Republican Chairman of the Rules Committee, publicly stated, "I think we'll have an omnibus bill coming out" and words to the effect that the Senate would move ahead with the "full bill." (Pl. Exs. 81, 714.) The contents of the "omnibus bill" were not disclosed at the time. HB 589 remained in the Senate Rules Committee without legislative action until late July 2013. (Pl. Ex. 121.)

2. Revision of HB 589

A meeting of the Rules Committee was scheduled for July 23. (See Pl. Exs. 121, 135, 202.) The night before the Rules Committee meeting, the revised version of HB 589, now fifty-seven pages in length, was posted for the members on the Rules Committee website.¹⁹ (Pl. Ex. 18A at 7–8 (declaration of Sen. Josh Stein); Pl. Ex. 107; Doc. 164 at 111–12 (testimony of Sen. Dan Blue); Doc. 335 at 169–72.) The revised bill contained a number of changes and now focused more broadly on election law reform. (See Pl. Ex. 107.) Plaintiffs have characterized the bill as a "monster voter suppression law," focusing on the fact that it emerged at fifty-seven pages. However, in truth, most of HB 589's changes—some forty-two of the fifty-seven pages (74%)—have gone unchallenged in this case. The changes were also highlighted for the convenience of the reader.

19 A version of HB 589 appears to have been distributed to members of the Rules Committee who were present on July 18, 2013. (Doc. 134–4 at 3.) It is not clear whether this version differed from that posted on the website on July 22.

HB 589's various unchallenged revisions, which claimed to "reform" North Carolina's election law, included: (1) standardizing the process for requesting an absentee ballot through an absentee ballot request form created by the SBOE (Part 4); (2) expanding the public agencies offering voter registration to include senior centers and parks and recreation services (Part 5); (3) making it illegal to compensate persons collecting voter registrations based on the number of forms they submitted (Part 14); (4) requiring biannual efforts by the SBOE to remove ineligible voters from North Carolina's voter rolls (Part 18); (5) reducing the number of signatures required to become a candidate in a party primary (Part 22); (6) deleting obsolete provisions about the 2000 census (Part 27); (7) mandating that several matters be referred for further study, including requiring the Joint Legislative Oversight Committee to examine whether to maintain the State's current runoff system in party primaries (Part 28); (8) eliminating the option of straight-ticket voting (voting for an entire party rather than individual candidates) (Part 32); (9) moving the date of the North Carolina presidential primary earlier in the year (Part 35); (10) eliminating taxpayer funding for appellate judicial elections (Part 38); (11) allowing funeral homes to participate in canceling voter registrations of deceased persons (Part 39); and (12) requiring provisional ballots to be marked as such for later identification (Part 52). (Pl. Ex. 107.) The bill also changed the ordering of North Carolina's ballots. Prior to 2013, while the candidates' names were listed in random order in primaries, Democratic candidates were always listed first in the general election ballots. (Doc. 341 at 165.) HB 589 altered the listing for general elections. (Pl. Ex. 107 at 43.)

The provisions challenged in the present lawsuit comprise approximately fifteen of HB 589's fifty-seven pages.²⁰ (See id.) Of those, roughly nine pages contain the voter-ID requirement.²¹ Many of the voter-ID provisions did not differ from those in the old, already debated version of the bill: The new changes principally included the removal of certain government, state university, and community college IDs from the acceptable list. (Compare Pl. Ex. 105 at 2–3 (original bill filed in the House on April 4, 2013), with Pl. Ex. 107 at 2 (version approved by the Senate Rules Committee on July 23, 2013.) The bill proposed that the voter-ID requirement go into effect in 2016 but be implemented through a "soft roll out," whereby voters at the polls in 2014 and 2015 would be advised of the law's requirement that they will need a qualifying photo ID to vote beginning in 2016 and to permit them time to obtain a free ID from the State. (Pl. Ex. 107 at 14.)

20 Several of the challenged parts of SL 2013–381 simply remove references to the old law, such as deleting terms like "pre-registration."

21 It is worth noting that, while this was the first time the Senators had seen this bill, HB 351 had already been passed by the Senate two years earlier. At least two of the Democratic Senators in the Rules Committee (Sens. Martin Nesbitt and Stein) in 2013 were also Senators when HB 351 passed.

So, of the fifty-seven-page bill, nine pages related to the voter-ID requirement, much of which was in the original version of the bill, and approximately six pages contained the other challenged provisions in this case. Those are: (1) the reduction of the period of early voting from seventeen to ten days; (2) the elimination of SDR; (3) the prohibition on the counting of ballots cast outside a voter's correct voting precinct on Election Day ("OOP voting"); (4) the allowance for up to ten at-large poll observers within each county; (5) the ability of any registered voter in the county, as opposed to precinct, to challenge a ballot; (6) the elimination of the discretion of CBOEs to keep the polls open an additional hour on Election Day in "extraordinary circumstances"; and (7) the elimination of "pre-registration" of sixteen- and seventeen-year-olds who will not be eighteen by the next general election.

Several legislators reported they had been caught off guard by these changes. (See, e.g., Pl. Ex. 18A at 8 (Sen. Stein); Pl. Ex. 21 at 7 (Sen. Blue).) In truth, many of these additions to HB 589 were drawn from or patterned after similar bills then pending in the General Assembly. See H.B. 913, 2013 Gen. Assemb., Reg. Sess. (N.C. 2013) (SDR); S.B. 666, 2013 Gen. Assemb., Reg. Sess. (N.C. 2013) (early voting and SDR); S.B. 721, 2013 Gen. Assemb., Reg. Sess. (N.C. 2013) (early voting); H.B. 451, 2013 Gen. Assemb., Reg. Sess. (N.C. 2013) (early voting and SDR). Moreover, and as discussed below, any assertions of surprise are weakened by the fact that Senator Stein appeared the next day with charts and statistics on early voting and SDR that likely could not have been tabulated overnight. (See Pl. Ex. 18A at 18 & Ex. A.)

When the Senate Rules Committee met as scheduled on July 23, Senator Apodaca allowed members of the public in attendance to speak for two minutes.²² (Pl. Ex. 202 at 41–56.) Speakers included the League Plaintiffs' counsel, Ms. Riggs, as well as Jamie Phillips, who represented the North Carolina State Conference of the NAACP. (Id. at 41–43, 53–54.) The majority of comments addressed the voter-ID requirement, although citizens also spoke in opposition to the other challenged provisions, including the changes to SDR, pre-registration, and early voting. Several opponents characterized the bill as an effort at voter suppression. (See, e.g., id. at 41 (Riggs: "voter suppression at its very worst"); id. at 53 (Phillips: "The fewer young people and minorities who vote, the better it seems in your minds. We get it. No one is being fooled.").) Proponents denied the charges. (See id. at 67–69, 74–75.)

22 There is no indication the two-minute time allotment was a deviation from normal rules.

The Senate Rules Committee debated the recent additions to HB 589. Senator Stein argued that a voter-ID requirement, the reduction in early voting, and the removal of SDR were unneeded changes that would burden voters. (Id. at 32, 37–40.) Senator Robert Rucho, a Republican supporter of HB 589, responded. As to early voting, Senator Rucho cited concerns about inconsistency in the administration of early voting, lack of optimal utilization of early voting during the seventeen-day period, and the then-exiting early-voting system's potential for "gamesmanship and partisan advantage." (Id. at 30–32, 74–75.) He also noted the potential for "savings in the sense that by going from seventeen to ten days you actually have more opportunity to open up more sites." (Id. at 30.) Senator Rucho further cited "integrity and honesty" in North Carolina's election administration as well as increased public confidence as reasons for the voter-ID provisions. (Id. at 37, 68.) He noted that other States have an ID-requirement but also expressed a belief that most people had one of the required forms of ID or, in combination with the two-year soft roll out, had ample opportunity to obtain a free photo ID, as provided through the bill. (Id. at 36, 39, 67–68.) Senator Rucho also observed that the bill eliminated college IDs from the list of acceptable IDs because of the inconsistency in the issuance of those IDs across the State. (Id. at 68–69.) The elimination of OOP voting was described—apparently by a legislative staffer introducing the bill—as "mov[ing] the law back to the way it was prior to 2005; conform[ing] to federal law"; no Senator spoke in opposition to its elimination.²³ (Id. at 12.) Senator Rucho defended the removal of SDR as a way to verify voter registrations by "giv[ing] the Board of Elections an opportunity to do their job correctly, [to] validate those individuals." (Id. at 41.) Finally, as to pre-registration, Senator Rucho stated that its elimination was meant to "offer some clarity and some certainty as to when ... that young person is eligible to vote and registers to vote," citing his son's own confusion about when pre-registration authorized him to vote. (Id. at 22.) After debate, the bill passed the committee and proceeded to the floor for second reading. (Id. at 76.)

23 One member of the public noted in passing that OOP voting was being removed. (Pl. Ex. 202 at 54.)

The following afternoon, on July 24, HB 589 was introduced on the floor of the full Senate. (Pl. Ex. 549 at 1–2.) During several hours of debate after the bill's second reading, Democratic senators introduced and discussed several proposed amendments. Plaintiffs argue that amendments "designed to ameliorate burdens on African Americans [proposed during the debate] were defeated with little discussion." (Doc. 346 at 52.) This is simply untrue. First, many such "amendments" were no more than proposals to remove the key provisions at issue. (See Pl. Ex. 549 at 32–33.) Moreover, the Senate did consider and adopt an early-voting aggregate-hours amendment by Senator Stein, which was substantive and significant. This came after Senator Stein argued that the reduction in early voting would disproportionately impact African Americans, and he introduced an amendment to require CBOEs to offer the same number of aggregate hours of early voting as were offered in the last comparable election (whether presidential or off-year). (Id. at 16–18, 43–44; see also Pl. Ex. 115 (text of the amendment).) This could be accomplished, he proposed, by CBOEs offering more hours at present sites, or by opening more sites. (Pl. Ex. 549 at 46.) Senator Stein argued that the amendment would "mitigate" the impact the reduction of early-voting days would have on all voters, including African Americans. (Id. at 28–29.) Responding, Senator Rucho urged the Senate to support Senator Stein's amendment, (id. at 44), and it passed by a vote of forty-seven to one, (id. at 49). In all, ten amendments were raised; two were withdrawn, and three were adopted. (E.g., id. at 49–51; 64–65; Pl. Ex. 121.)

During the more than four hours of debate, the Senators exchanged argument on many of the other challenged provisions, including voter ID, SDR, pre-registration, and the increase in allowable poll observers, as well as several provisions not at issue here (including the elimination of straight-ticket voting and reduction of various campaign-finance restrictions). (See Pl. Ex. 549 at 66–141.) Senator Stein presented charts to support his arguments about HB 589's disparate impact²⁴ with respect to early voting and SDR, (see Pl. Ex. 18A at 18 & Ex. A.), although it is not clear how many senators reviewed them, (Doc. 335 at 195–97). During this hearing, supporters of the bill offered the following reasons in support of its enactment: reestablishing confidence in the electoral process through voter ID, with noted skepticism about the number of voters lacking acceptable identification or the ability to obtain one in North Carolina, (Pl. Ex. 549 at 2–3, 86–88, 90); the public support for voter ID, (id. at 3); concern over voter fraud, (id. at 78, 95); inconsistency in college IDs (which were previously permitted in the House version of the bill), (id. at 91–92); providing CBOEs with flexibility to expand early-voting hours and sites to ensure voter access, (id. at 4–5, 11); allowing for the verification of voters' information, (id. at 5, 78); eliminating confusion stemming from pre-registration, (id. at 6–7); and a desire to align the State with the practices of other States as to SDR, pre-registration, and voter ID, (id. at 37, 76–77). At the end of the debate, Senator Martin Nesbitt (Democrat), although opposing the bill strongly and urging its defeat, described the debate as "heated," "healthy," and "good." (Id. at 136.) He characterized two of the "unintended consequences" of the bill to have been "fixed" through the amendment process. (Id. at 137.) After the bill passed the second reading, Senator Apodaca objected to a third reading that day, which extended its consideration by mandating that the debate of the bill be carried over into the next day. (Id. at 142.)

24 In denying Plaintiffs' motion for a preliminary injunction, this court used the term "disparate impact" as shorthand for the fact that African Americans disproportionately used the removed mechanisms. SeeN.C. State Conference of NAACP v. McCrory, 997 F.Supp.2d 322, 355–56 (M.D.N.C.2014). This court did not view "disparate impact" and "discriminatory burden" as necessarily equivalent terms, and thus by using the term disparate impact never meant to imply that it had found an inequality of opportunity as contemplated by § 2. Cf.League, 769 F.3d at 240, 244–46 (using the term "discriminatory burden" in laying out its two part test, but using the term "disproportionate impact" in applying the first step of that test).

On July 25, the Senate began its session with the third reading of amended HB 589. (Pl. Ex. 550 at 1–2.) Senator Rucho offered a bipartisan amendment to clarify Senator Stein's aggregate-hours amendment to permit a county to obtain a waiver from the aggregate-hours requirement upon unanimous approval of both the CBOE and the SBOE; it passed forty-six to zero. (Id. at 7, 16; see also Pl. Ex. 119 (text of amendment).) Proponents and opponents of the bill debated both its provisions and the merits of various amendments over the next four-plus hours, and the Senate accepted an amendment dealing with electioneering from Senator Dan Blue (Democrat). (Pl. Ex. 550 at 82–83.) Points made in favor of the bill at this time included the increased integrity of elections furthered by requiring voter ID, (id. at 44, 99); public support for a voter ID requirement, (id. at 44, 52, 98, 100); concerns of voter fraud, (id. at 76); increased time for verification of voter registrations (SDR), (id. at 45–46); bolstering public confidence in the election process, (id. at 53); increased early-voting hours for voters who worked full-time jobs, (id. at 55–56); and statewide uniformity in early voting, (id. at 56–57). Several Democratic senators characterized the bill as voter suppression of minorities. (E.g., id. at 26–35 (Sen. Stein), 57–67 (Sen. Blue), 68–74 (Sen. Gladys Robinson).) Others characterized the bill as partisan. (Id. at 42 ("I can't help but wonder if the goal is simply to maintain political power."); id. at 66 (contending that the intent of the law is incumbency protection).) Proponents strongly denied such claims and claimed the bill reversed past practices Democrats passed to favor themselves. See, e.g., id. at 50–53 (Sen. Thom Goolsby (Republican) alleging Democrat-partisan influence in past election administration); id. at 74 (Sen. Andrew Brock (Republican) expressing desire to correct Democrat-influence in the placement of early-voting sites).)

By the close of debate, a total of fourteen amendments had been raised in the Senate, with five being adopted, and the Senate voted in favor of HB 589 along party lines; the bill then returned to the House for concurrence, as amended. (Id. at 100; Pl. Ex. 121; Pl. Ex. 124.) Senator Nesbitt (Democrat), although a vocal opponent of the bill, noted that "we've had a good and thorough debate on this bill over two days" and "reviewed the bill in great detail." (Pl. Ex. 550 at 90–91.)

With the end of the legislative session approaching, the House received the Senate's version of HB 589 that night. (Pl. Ex. 121.) At the beginning of a two-hour floor session starting at 7:45 p.m., Representative Henry M. Michaux, Jr. (African American, Democrat) moved that the House form a Committee of the Whole²⁵ to consider the bill. (Pl. Ex. 138 at 1–3.) Representative Michaux testified at trial that forming a Committee of the Whole was not customary, and he could not recall the House ever before having done so. (Doc. 336 at 38.) Representative Tim Moore (Republican) opposed the motion on the grounds that "it is simply a waste of time" because such a committee "is the same as the full House," which the bill was properly before at the moment. (Pl. Ex. 138 at 5 ("I can't think of the last time the House has met as the Committee of the Whole.").) The motion appears to have been a tactic to slow or stop the bill, and it failed by a vote of forty-one to sixty-nine. (Id. at 6–7.)

25 A Committee of the Whole is a legislative device where the whole membership of a legislative house sits as a committee and operates under informal rules. Webster's Third New International Dictionary458 (1986); see alsoH.R. 54 at 12.

Two amendments offered by opponents (Sen. Blue's amendment of the date for electioneering; amendments by Senators Rucho and Stein altering several items, including "expand[ing] and "better defin[ing]" the type and number of IDs that can be presented for voting, and requiring the same number of hours of early voting) were adopted 109 to 0. (Id. at 7–11.) The provisions of the new full bill were then reviewed. (Id. at 12–27.) Each member of the House Democratic caucus present—including four of the five members who voted for the House version in April—were granted time to speak in opposition to the bill. (Id. at 67–69, 76–79, 88–89; Doc. 165 at 64–65 (testimony of Rep. Glazier).) Some opponents characterized the measure as voter suppression, partisan, and disproportionately affecting African Americans, young voters, and the elderly. (E.g., Pl. Ex. 138 at 57 ("[O]ur anger tonight is palpable. Passage of this bill is a political call to arms."); id. at 59 ("This is the most pointedly, obviously politically partisan bill I've ever seen."); id. at 64 ("voter suppression"). On the Republican side, Representative Lewis, a House supporter of the bill, spoke in support of the amended bill.²⁶ (Id. at 116–20.) He pointed out, among other things, that the bill does not bar Sunday voting, maintains the same overall hours of early voting, provides for free photo ID, and, in his opinion, strengthens the requirements for absentee voting.²⁷ (Id.) Subsequently, the House voted—again along party lines—to concur in the Senate's version of HB 589 at 10:39 p.m. (Id. at 120; Pl. Ex. 122 (noting July 25 roll call vote in House).)

26 Plaintiffs contend that Representative Harry Warren (Republican) "[i]n describing the changes made by the Senate ... misleadingly claimed that the Senate substitute made very few substantive changes to the House version." (Doc. 346 at 54.) This is a mischaracterization. Representative Warren actually stated that the Senate "made very few substantive changes to the VIVA Act," specifically referring to the voter ID portion of HB 589—not the entire bill. (Pl. Ex. 138 at 13, 17 (describing changes to the bill beyond voter ID as "addition[s] to the VIVA bill").) While Plaintiffs may debate the impact of the changes made to the voter-ID portion, Representative Warren's statement was a reasonable description of the number of changes made to that portion of the bill. (Seeid.at 13–15.)

27 Plaintiffs argue that Representatives Warren and Lewis failed to identify the changes made to early voting. (Doc. 346 at 54.) A review of the entire transcript reveals that Representative Warren was not tasked with introducing the early-voting revisions (Part 25 of HB 589); rather, Representative Lewis was. (Pl. Ex. 138 at 20, 22.) And, while Representative Lewis did not state that the bill "eliminated a full week of early voting" in his initial description of Part 25 of the bill (just as he did not mention the ameliorative amendment of the aggregate-hours requirement (id.at 22)), he did in fact address it during the debate, (id.at 116–17).

In total, there are over 430 pages of transcript representing several hours of debate on HB 589 after its amended version was introduced in July 2013. There is no evidence that any House or Senate Rule was disregarded or violated at any time during the bill's legislative process. (Def. Ex. 217 at 3; Doc. 335 at 193.)

The bill was ratified the next day and presented to Governor McCrory on July 29. (Pl. Ex. 121.) The governor signed the bill into law on August 12, 2013, over the recommendation of the Attorney General (an elected Democrat), who nevertheless appears in this lawsuit to defend it. (Id.)

3. Enactment of HB 836

On June 18, 2015, less than a month before trial was set to begin in these cases, the General Assembly passed House Bill 836, and the governor signed it into law as SL 2015–103 on June 22, 2015.²⁸ The law modified the photo-ID scheme created by SL 2013–381 in three primary ways.

28 Session Law 2015–103 will be referred to as HB 836 when discussing legislative history.

First, it expands the category of acceptable photo IDs by permitting driver's licenses, permits, provisional licenses, and non-operator IDs that have been expired for up to four years. N.C. Gen. Stat. § 163–166.13(e)(1)–(2). Moreover, any voter seventy years of age or older is permitted to vote using any of the acceptable identifications that expired at any point after the voter's seventieth birthday. Id.§ 163–166.13(f).

Second, the law requires poll workers to inform those without an acceptable ID that they can complete a written request for an absentee ballot at an early-voting site until 5:00 p.m. on the Tuesday before Election Day (i.e., the deadline for requesting absentee ballots). Id.§§ 163–166.13(c)(3), 163–227.2(b1), 163–230.1.

Third, and most importantly, it creates an additional exception that permits in-person voters who do not have an acceptable photo ID to cast a provisional ballot so long as they complete a declaration stating a reasonable impediment prevented them from acquiring qualifying photo ID. Id.§§ 163–166.13(c)(2), 163–166.15(a) –(b). Such voters must present alternate identification, which can consist of "the voter registration card issued to the voter by the county board of elections" or "a current utility bill, bank statement, government check, paycheck, or other government document"²⁹ that shows the name and address of the voter.³⁰Id.§§ 163–166.15(c), 163–166.12(a)(2). Alternatively, voters may provide their date of birth and the last four digits of their SSN ("SSN4"). Id.§ 163–166.15(c).

29 These are the same methods of identification that were required for SDR when it was in place. 2007 N.C. Sess. Law 253, § 1 (permitting SDR-registrants to use any of the documents listed in N.C. Gen. Stat. § 163–166.12(a)(2) ).

30 A voter who is unable to provide either of these forms of identification may comply with the requirement by returning to their CBOE by noon the day prior to the election canvass and presenting the required alternate ID. N.C. Gen. Stat. §§ 163–166.15(d), 163–182.1B(c).

Session Law 2015–103 expressly addresses the scope of the reasonable impediment exception. At a minimum, all reasonable impediment declaration forms are required to include separate boxes listing the following reasonable impediments to acquiring a photo ID: (1) "Lack of transportation; (2) "Disability or illness"; (3) "Lack of birth certificate or other documents needed to obtain photo identification"; (4) "Work Schedule"; (5) "Family responsibilities"; (6) "Lost or stolen photo identification"; and (7) "Photo identification applied for but not received by the voter voting in person." Id.§ 163–166.15(e). In addition, the form must list a box for "[o]ther reasonable impediment," which the voter can check and provide a "brief written identification of the reasonable impediment."³¹Id.§ 163–166.15(e)(1) h.

31 The voter can also indicate that State or federal law prohibits listing the impediment. N.C. Gen. Stat. § 163–166.15(e)(1) h.

Although a reasonable impediment voter casts a provisional ballot, the ballot must be counted unless one of the following is true: the impediment described in the declaration is "factually false, merely denigrate[s] the photo identification requirement, or [is an] obviously nonsensical statement[ ]"; the voter fails to provide one of the alternate forms of identification discussed above; the CBOE could not confirm the voter's registration using the alternate form of identification provided; or the "voter is disqualified for some other reason provided by law." Id.§ 163–182.1B(a). Significantly, if a voter's reasonable impediment declaration is challenged, the CBOE is required to "construe all evidence presented in the light most favorable to the voter submitting the reasonable impediment declaration" and cannot reject the impediment on the ground that it is not reasonable. See id.§ 163–182.1B(b)(5)–(6).House Bill 836 was proposed with little notice and considered as an amendment to a pending conference report, which is unusual. (Pl. Ex. 895.) The legislative record contains thirty pages of debate and indicates an intent for a broad application of the exception. (Id.) Democrats questioned the process but consented to fast-track consideration, thus enabling the law to receive immediate consideration and passage by a wide margin (103 to 4). (Id. at 30; see also id. at 29 (Representative Hall: "This conference report does contain some things that will help get us back to where we are [sic] where we did not have unnecessary restrictions on voters and where we didn't have any real measurable voter fraud.").)

In summary, as to the major components challenged herein, SL 2013–381 and SL 2015–103 had the following effect:

Voter ID:

Beginning in 2016, in-person voters who have a qualifying photo ID "bearing any reasonable resemblance to that voter" must present it to vote, either at the polling place or at the CBOE after casting a provisional ballot. N.C. Gen. Stat. §§ 163–166.13, 163–182.1A ; 2013 N.C. Sess. Law 381, § 6.2. Acceptable photo IDs are (1) a North Carolina driver's license, learner's permit, or provisional license (expired up to four years); (2) a special non-operator's identification card (expired up to four years); (3) a United States passport; (4) a United States military identification card; (5) a Veterans Identification Card issued by the United States Department of Veterans Affairs; (6) a tribal enrollment card issued by a federally recognized tribe; (7) a tribal enrollment card issued by a tribe recognized by North Carolina, so long as it is signed by an elected official of the tribe and the requirements for obtaining it are equivalent to the requirements for obtaining a special identification card from the North Carolina Department of Motor Vehicles ("DMV"); and (8) a driver's license or non-operator's identification card issued by another State or the District of Columbia so long as the voter registered to vote within ninety days of Election Day. N.C. Gen. Stat. § 163–166.13(e). Those who do not have a qualifying photo ID and who can list a reasonable impediment to getting one, can vote in person without photo ID so long as they provide alternative identification and complete a reasonable impediment declaration. Id.§§ 163–166.13(c)(2), 163–166.15. Because the reasonableness of the impediment given cannot be challenged, in practice the reasonable impediment exception is better characterized as an impediment exception. See id.§ 163–182.1B(b)(6). Only the voter's subjective belief is relevant to the reasonableness inquiry. See id. In addition, curbside voters, those with religious objections to being photographed, certain victims of natural disasters, and absentee mail voters are exempt from the photo-ID requirement. Id.§ 163–166.13(a).

Early Voting:

Early voting must now begin "[n]ot earlier than the second Thursday before an election," a reduction of seven days of permissible early voting. Id.§ 163–227.2(b). Early voting must end by 1:00 p.m. on the final Saturday before Election Day, eliminating CBOE discretion to keep early-voting sites open until 5:00 p.m. Id. Under the Stein amendment, the decrease in permissible days, however, is coupled with a requirement that the aggregate voting hours offered remain the same. Thus, the law requires that, "[f]or elections which do not include a presidential candidate on the ballot," CBOEs must "calculate the cumulative total number of scheduled voting hours at all sites during the 2010 ... elections" and "ensure that at least the same number of hours offered in 2010 is offered for [early voting] ... through a combination of hours and numbers of [early-voting] sites during [those] election[s]." Id.§ 163–227.2(g2)(2). In other words, counties must offer the same number of aggregate hours of early voting in non-presidential elections as they did in November 2010. CBOEs must make the same calculation with respect to presidential elections: e.g., the same hours in 2016 elections as in 2012. Id.§ 163–227.2(g2)(1). The CBOEs can meet these requirements either by opening more early-voting sites or keeping the existing sites open for more hours, including expanding weekend voting hours. See id.§ 163–227.2(f) ("A county board may conduct [early] voting during evenings or on weekends, as long as the hours are part of a plan submitted and approved according to subsection (g) of this section."). The law also requires that, except for CBOE offices, any early-voting site within a county maintain the same hours of operation as every other site in that county. Id.§ 163–227.2(g).

If a county determines that it cannot meet the aggregate-hours requirement or that additional hours are unnecessary, it may obtain a waiver only "by unanimous vote of the board, with all members present and voting." Id.§ 163–227.2(g3). The waiver request must also be approved by a unanimous vote of the SBOE. Id. Absent a waiver, counties must either open more early-voting sites or keep existing sites open longer to satisfy SL 2013–381's aggregate-hours requirement.

SDR:

Session Law 2013–381 repealed SDR. To be eligible to vote in any primary or general election, a voter must comply with preexisting law that requires the registration application to be postmarked or delivered in person at least twenty-five days before Election Day. Id.§ 163–82.6(c).³² Under existing federal law, those who move to the State after the registration cut-off date nevertheless remain able to vote for president and vice president. 52 U.S.C. § 10502(e).

32 Even after the passage of SL 2013–381, North Carolina continues to permit citizens who become naturalized after the registration deadline but before Election Day to both register and vote on Election Day. SeeN.C. Gen. Stat. § 163–82.6(d) –(e). Plaintiffs' fact witness, Rita Palmer, who works to register Hispanics, testified incorrectly that the elimination of SDR would cause such new citizens to lose their right to register and vote. (Doc. 329 at 151, 159–60.)

OOP:

Session Law 2013–381 prohibits the counting of OOP provisional ballots, thereby reinstating the North Carolina Supreme Court's interpretation of State law in James. N.C. Gen. Stat. § 163–55(a) now provides,

Every person born in the United States, and every person who has been naturalized, and who shall have resided in the State of North Carolina and in the precinct in which the person offers to vote for 30 days next preceding an election, shall, if otherwise qualified as prescribed in this Chapter, be qualified to vote in the precinct in which the person resides.

Thus, as a general matter, if a voter appears at the wrong precinct on Election Day, he or she will have to go to the proper precinct before the close of the polls in order to cast a valid vote.³³Pre-registration:

33 As discussed below, North Carolina law provides for out-of-precinct voting in one context even after SL 2013–381: If a registered voter moves to another precinct within the same county more than thirty days before an election but fails to report the move to the CBOE before the close of registration, the registrant is permitted to cast an out of precinct provisional ballot in her old precinct. N.C. Gen. Stat. § 163–82.15(e). As with regular OOP, the unreported mover's provisional ballot is only counted for those races for which the voter was eligible to cast a ballot in her new precinct. Id.

Session Law 2013–381 ends the practice of pre-registration. Voter registration applications now ask only one question regarding the applicant's age: "Will you be 18 years of age on or before election day?" N.C. Gen. Stat. § 163–82.4(d)(2). Thus, those who are seventeen but will be eighteen before Election Day may still register to vote in that election and in any primary before that election under SL 2013–381.

Under SL 2013–381, the repeal of pre-registration took effect in September 2013; the revisions to early voting and the elimination of SDR and OOP voting became effective in January 2014; and the voter-ID requirement took effect in January 2016. 2013 N.C. Sess. Law 381, §§ 6.2, 12.1(j), 60.2.

C. Procedural History

On the same day that Governor McCrory signed HB 589 into law, two groups sued to enjoin it. The NAACP filed its complaint in case 1:13cv658, later amended, alleging that the voter-ID requirement, elimination of SDR, reduction of early-voting days, prohibition on counting OOP provisional ballots, elimination of pre-registration, and the expansion of poll observers and ballot challenges discriminates against African Americans and Hispanics in violation of the Fourteenth and Fifteenth Amendments of the United States Constitution, as well as § 2 of the VRA. The League Plaintiffs filed their complaint in case 1:13cv660, alleging that the elimination of SDR, reduction of early-voting days, prohibition on counting OOP provisional ballots, and elimination of CBOE discretion to extend poll hours one hour on Election Day discriminates against African Americans and imposes an unjustified burden on all North Carolinians, in violation of § 2 of the VRA and the Fourteenth Amendment. On September 30, 2013, the United States filed its complaint in case 1:13cv861, alleging that the law's early voting, SDR, OOP voting, and voter-ID provisions discriminate against African Americans in violation of § 2 of the VRA. These cases were consolidated for discovery and were later consolidated for trial at the parties' request. (Doc. 252.)

On January 27, 2014, the court permitted a group of "young voters" over the age of eighteen and others to intervene as Plaintiffs in case 1:13CV660 pursuant to Rule 24(b) of the Federal Rules of Civil Procedure. Intervenors allege that the elimination of pre-registration, reduction in early voting, repeal of SDR, prohibition on counting OOP ballots, elimination of CBOE discretion to keep the polls open an extra hour on Election Day, and implementation of a voter-ID requirement violate the Fourteenth and Twenty-Sixth Amendments.

All Plaintiffs alleged that the variously challenged provisions of SL 2013–381 have a discriminatory intent and effect, although the United States has since abandoned its discriminatory effect claim to the voter-ID law after passage of the reasonable impediment exception. As relief, they seek to permanently enjoin the challenged provisions. The United States seeks the appointment of federal observers under § 3(a) of the VRA and to subject North Carolina to a pre-clearance requirement under § 3(c).

Several elections have occurred during the pendency of these cases. The first occurred in the fall of 2013, when North Carolina held municipal elections. No plaintiff sought to enjoin enforcement of the law during this election.

In December 2013, after a hearing with all parties and at their request, the Magistrate Judge issued a scheduling order, setting May 5, 2014, as the deadline for the filing of a motion for preliminary injunction and dispositive motions. (Doc. 30.) The parties later jointly moved to reset this deadline to May 19, 2014. (Doc. 91.)³⁴

34 NAACP and League Plaintiffs sought a July 2014 trial date, while the United States sought an extended schedule with the possibility of a preliminary injunction in July 2014. (Doc. 39 at 20–24, 29, 35–40, 42–44.) In any event, no party sought to schedule a hearing on a motion to preliminarily enjoin SL 2013–381 before the May 2014 primary election. Cf.League, 769 F.3d at 249–50 (Motz, J., dissenting) (expressing concern about the timing of the preliminary injunction proceeding and the election).

On May 6, 2014, North Carolina held a midterm primary election. No plaintiff sought to enjoin enforcement of SL 2013–381 during this election. Compared to the previous comparable primary midterm election, 2010, turnout increased overall: among registered white voters, it increased from 15.8% to 17.4%; among registered African American voters, it increased from 11.4% to 13.4%; and among registered Hispanic voters, it increased from 2.9% to 3.3%. (Def. Ex. 309 at 66.) Thus, the greatest increase in turnout in the 2014 midterm primary was observed among African American voters, despite the implementation of SL 2013–381.

On May 19, 2014, Defendants moved for judgment on the pleadings. (Doc. 94.) That same day, Plaintiffs filed their motion for preliminary injunction and the United States sought the appointment of federal observers. (Docs. 96, 98.) Collectively, Plaintiffs sought to enjoin the elimination of SDR, OOP voting, pre-registration, CBOE discretion to keep the polls open an extra hour, the reduction of early voting, the expansion of poll observers and ballot challengers, and the "soft rollout" of the voter-ID law (in which voters would be advised that the photo-ID requirement would apply starting in 2016).

Beginning on July 7, 2014, this court held a four-day evidentiary hearing on all motions. On August 8, 2014, after considering the testimony of multiple fact and expert witnesses and a record with over 11,000 pages of exhibits and materials, this court issued a 125–page opinion denying the motions for preliminary injunction but refusing to dismiss any claims. 997 F.Supp.2d 322 (M.D.N.C.2014).

On August 18, 2014, Plaintiffs—except for the United States—and Intervenors filed notice of appeal. (Docs. 172, 174, 175.) Plaintiffs were granted an expedited appeal, a limited briefing schedule was ordered, and on September 25, 2014, the court heard oral argument. Less than a week later, on October 1, 2014, a divided panel issued an opinion affirming in part, reversing in part, and remanding with instructions. League, 769 F.3d at 229. The majority found that Plaintiffs had demonstrated a likelihood of success on the claim that the repeal of SDR and OOP voting violated § 2 of the VRA. The court of appeals remanded the case with instructions to reinstitute the previous SDR and OOP voting provisions "in full force pending the conclusion of a full hearing on the merits." Id. at 248–49. Judge Diana Gribbon Motz dissented.

On the same day the Fourth Circuit issued its opinion, it also issued its mandate. Defendants immediately requested that the mandate be recalled and stayed pending the filing and disposition of a petition for writ of certiorari with the Supreme Court of the United States. The next day, on October 2, the panel denied the motion, over Judge Motz's dissent.

Later that day, Defendants applied to the Chief Justice for a stay and recall of the Fourth Circuit's mandate pending the filing and disposition of a certiorari petition. The Chief Justice referred the application to the full Court. Yet, while the proceedings were pending in the Supreme Court, the League Plaintiffs filed an emergency petition for a writ of mandamus with the Fourth Circuit, seeking to have the court of appeals order this court to enter the preliminary injunction. On October 3, this court entered the preliminary injunction mandated by the Fourth Circuit and set a hearing for October 7, to address the State's plan for ensuring implementation of the preliminary injunction. (Doc. 189.)

The next day, the Supreme Court, over the dissent of two Justices, recalled and stayed the Fourth Circuit's mandate and this court's preliminary injunction, pending the filing and disposition of Defendants' petition for a writ of certiorari. ––– U.S. ––––, 135 S.Ct. 6, 190 L.Ed.2d 243 (2014). Two days later, on October 10, the Fourth Circuit implemented the stay and recall of mandate. After briefing in the Supreme Court, Defendants' petition was denied on April 6, 2015. On May 5, 2015, the mandate of the Fourth Circuit issued, and this court's preliminary injunction went back into effect, reinstituting SDR and OOP until a final resolution on the merits.

Meanwhile, while the mandate and preliminary injunction were stayed by the Supreme Court, North Carolina held its 2014 general election, the third election under SL 2013–381. Compared to the last comparable midterm general election, 2010, voter participation increased: among registered white voters, it increased from 45.7% to 46.8%; among registered African American voters, it increased from 40.4% to 42.2%; and among registered Hispanic voters, it increased from 19.9% to 20.5%. (Def. Ex. 309 at 66.) Not only did African American turnout increase more than other groups in 2014 with SL 2013–381 in place, but that general election saw the smallest white-African American turnout disparity in any midterm election from 2002 to 2014. (Id. at 62; Pl. Ex. 229 at 7.)

Trial was set for July 13, 2015. But as a result of the General Assembly's June 18, 2015 passage of House Bill 836, the court immediately held a status conference to address how this change in law might affect the pending cases. Plaintiffs pushed for additional time to assess the new law but opposed delaying trial on the remaining claims; Defendants argued that the amendment rendered the challenge to the voter-ID portion of the law moot. The court proposed continuing the trial to September 2015 but, at Plaintiffs' urging, ultimately carved out the challenge to the voter-ID law from the July 13 trial setting and agreed to proceed to trial on the balance of the consolidated claims. (Doc. 282.) Thereafter, Defendants moved to dismiss the voter-ID challenge as moot (Doc. 299); the court denied the motion, setting trial on those claims for January 25, 2016.

Beginning July 13, 2015, this court held a trial on the merits of all claims except those challenging the voter-ID provisions. Over the course of three weeks, the court took the testimony of over 100 witnesses, both live and through deposition. Following trial, the parties submitted almost 300 pages of proposed findings of fact and conclusions of law. (Docs. 346, 347.)

On November 24, 2015, five months after SL 2015–103 and eight weeks before trial, NAACP Plaintiffs moved to preliminarily enjoin the implementation of the photo-ID provision of SL 2013–381, as amended by the reasonable impediment provision. (Doc. 371.) The United States did not join this motion. Briefing was completed on December 21, 2015. (Doc. 375.) On January 15, 2016, the court issued a fifty-four page memorandum opinion and order denying NAACP Plaintiffs' motion for preliminary injunction. (Doc. 383.)

On January 25, 2016, this court held a trial on the merits of Plaintiffs' claims challenging the voter-ID law.³⁵ The court took the testimony of five expert and nineteen fact witnesses over the course of six days. The parties have since submitted 138 additional pages of proposed findings of fact and conclusions of law. The court also granted Plaintiffs' unopposed request to supplement the record post-trial with materials relating to SBOE training efforts occurring in early February 2016.

35 The Intervenor Plaintiffs elected not to participate at trial.

Including the evidence from the preliminary injunction hearing, which the parties have stipulated to be considered part of the trial record pursuant to Federal Rule of Civil Procedure 65(a)(2), the record therefore consists of over 25,000 pages of exhibits, reports, and deposition transcripts.³⁶ Despite Plaintiffs' insistence on trying the claims separately, which invited this court's piecemeal adjudication, the court has remained faithful to the Fourth Circuit's direction in League, 769 F.3d at 242, to "consider the sum of those parts and their cumulative effect on minority access to the ballot box." Accordingly, this court has considered the issues from all evidentiary proceedings in reaching its decision.

36 This does not include the trial transcript or Plaintiffs' Exhibit 646, which is a database with 39,912 pages.

D. Evidence of Voter Experience Under Current Law

Plaintiffs argue that the addition of voter ID and the removal of the challenged provisions of SL 2013–381 creates inequality of opportunity, unlawfully burdens all voters, and results in a disparate, negative burden on African American, Hispanic, and "young" voters. The evidence as to each of the challenged provisions is addressed below.

1. Voter ID

Due to the soft rollout, a photo ID was not required until the March 2016 primary. Thus, unlike the other provisions at issue in this case, no data were presented from an election where the ID requirement was in place.

a. Voter Education about the Voter-ID Requirement Prior to the Reasonable Impediment Exception

Defendants have engaged in substantial efforts to educate voters about the State's photo-ID requirement. Three elections were held during the educational effort: municipal elections in November 2013, midterm primary elections in May 2014, and midterm general elections in November 2014.

Session Law 2013–381 contains a mandate to inform and educate voters about the new law. See 2013 N.C. Sess. Law 381, § 5.3. The General Assembly appropriated approximately $2 million to implement this requirement.³⁷ (Def. Ex. 535 at 2.) To accomplish these efforts, the SBOE hired election specialists "to create a mechanism to inform and provide education to the public on the requirements for [SL 2013–381], and to assist voters who do not have a photo ID for the purpose of voting in obtaining a photo ID." (Pl. Ex. 815 at 16–17.)

37 SBOE Executive Director Kim Strach testified before trial that $900,000 remained by 2016, most of which will be spent on media this year. N.C. State Conference of the NAACP v. McCrory, 156 F.Supp.3d 683, 691–93 n. 12, No. 1:13CV658, 2016 WL 204481, at *6 n. 12 (M.D.N.C. Jan. 15, 2016).

One of the SBOE's efforts was to educate voters at the polls. For the 2014 general election, the SBOE developed a color poster that depicted the photo IDs that would be accepted in 2016 and advised that voters would need a photo ID to vote beginning in 2016. (Id. at 28.) To accompany the poster, the SBOE developed a "two- sided color card that [would be handed out] to people who want[ed] information about photo ID and how to obtain it." (Id.) In order to avoid confusion that photo ID was needed prior to 2016, the SBOE also developed a large sign to be displayed at the entrance of voting sites stating that voters did not need photo ID to vote in the current election. (Id. at 29.) A similar sign facing voters exiting the voting site stated that voters would need a photo ID to vote in 2016 and encouraged them to ask poll workers for more information. (Id.) In addition, at least by the 2014 primary, poll workers were directed to tell voters they would need a photo ID to vote in 2016, show the color poster illustrating qualifying IDs, and ask voters whether they had access to one of the approved forms of ID. (Id. at 71–72.) Voters who said they had a qualifying ID were asked to sign the poll book; if they said they did not, they were asked to sign the poll book with a line that said, "I do not have a photo ID" and were told that they would need one starting in 2016. (Id.) Poll workers gave voters without qualifying ID the two-sided color push card noted above with instructions on how to get a free ID. (Id. at 72–73; Doc. 414 at 167; Def. Ex. 478.) The State kept track of those who claimed they did not have access to an acceptable photo ID. (Pl. Ex. 815 at 72–73.)

In addition to efforts to educate voters at polling sites, the SBOE created a special website dedicated to the photo-ID requirement and contacted specific voters who potentially lacked qualifying ID. This mailing effort targeted two groups. First, the SBOE sent a mailing to 10,675 registered voters who claimed they did not have acceptable photo ID while voting in the 2014 elections. (Def. Ex. 535 at 7.) Second, it sent mailings to those who could not be matched to a government database of persons with acceptable IDs. (Id. at 8–11.)

In the lead up to the enactment of North Carolina's voter-ID law, the SBOE engaged in a series of database matching efforts designed to gauge how many North Carolina registered voters lack qualifying ID. Those registrants who could not be matched to a list of persons with North Carolina DMV-issued ID appeared on a "no-match list." (See, e.g., Pl. Ex. 891 at 3.) The SBOE's no-match list contained 1,005,581 registrants as of a February 9, 2011 report, 612,955 registrants as of a January 7, 2013 report, 481,109 registrants as of a March 5, 2013 report, and 318,643 registrants as of an April 17, 2013 report. (Id.(tbl. 1).) The SBOE refined the matching criteria in each report. Although none of the SBOE's reports was "intended to be a comprehensive study on the number of voters who may not have any photo ID, regardless of source," the SBOE indicated that its April 2013 report was, at that time, the "most accurate estimate on the number of voters for whom [it] c[ould not] determine to have a photo ID issued by the [DMV]." (Pl. Ex. 534 at 1.) The SBOE's most recent no-match analysis took place in February 2015 and was based on a November 4, 2014 snapshot of the data. (Pl. Ex. 535 at 8.) It identified 254,391 registered voters who could not be matched to a qualifying ID in the DMV's database. (Id.) Further, as part of this litigation, Plaintiffs' expert Charles Stewart, Ph.D., Kenan Sahin Distinguished Professor of Political Science at the Massachusetts Institute of Technology,³⁸ engaged in a matching analysis based on a July 16, 2014 snapshot of the data. (Pl. Ex. 242 at 11, 38 (tbl. 7).) Dr. Stewart concluded that 397,971 registrants could not be matched to having a qualifying ID in the DMV and certain federal databases. (Id.)

38 Dr. Stewart was proffered without objection as an expert in American politics, election administration, research methods in political science, and the effect of election reforms in the American electorate. (Doc. 332 at 52.)

The SBOE followed up on both its February 2015 report and Dr. Stewart's report. Specifically, it contacted 218,097³⁹ registered voters on the basis of the SBOE's February 2015 no-match list and 209,253⁴⁰ registered voters on the basis of Dr. Stewart's no-match list. (Def. Ex. 535 at 8–10.) The mailing stated that photo ID would be needed to vote in 2016, listed resources for obtaining free photo ID, and provided a postage pre-paid response card where recipients were asked to "confirm[ ] whether they had acceptable photo ID, and indicat[e] whether, if they did not, they would like assistance in obtaining one." (Id. at 9.) The SBOE went through the same mailing process for voters who claimed they did not have a qualifying photo ID while voting in 2014. (Id. at 7–8.) At trial, Dr. Stewart testified that these mailings were a very good idea and an appropriate way to educate voters about the new law. (Doc. 408 at 49–50.)

39 This number is less than the number of registrants identified by the SBOE's February 2015 no-match analysis (254,391) because the SBOE removed "(1) registrants who had a registration status of 'removed' at the time of the mailing and (2) registrants whose address could not be validated by the mail house vendor responsible for sending out the mailer." (Def. Ex. 535 at 8–9.)

40 This number is less than the 397,971 appearing on Dr. Stewart's no-match list because the SBOE "eliminated voters who had been removed from the voter rolls through regular list maintenance ... [and] eliminated voters ... who also appeared on SBOE's no-match list and had therefore already been sent a mailing." (Def. Ex. 535 at 9–10.)

In sum, for nearly two years North Carolina frequently notified voters that, unless they met an exception, they would need photo ID to vote in 2016.

b. Voter Education After Enactment of the Reasonable Impediment Exception

With the advent of SL 2015–103's reasonable impediment exception on June 22, 2015, the prior information provided to voters was rendered incomplete. Session Law 2015–103 requires the SBOE to educate voters on the availability of the reasonable impediment exception, 2015 N.C. Sess. Law 103, § 8.(g), and the SBOE has engaged in substantial efforts to do so.

Creation and Distribution of Updated Materials

SBOE Executive Director Kim Strach, whom the court found credible, testified that "[i]mmediately after the enactment of S.L. 2015–103 in June 2015, SBOE staff developed new materials which would inform the public of modifications to the photo identification requirements and the availability of the reasonable impediment declaration option." (Def. Ex. 535 at 16.)

These new materials "were delivered to every county board of elections for posting and distribution at early voting and Election Day polling locations during the 2015 municipal elections"; "have been distributed to groups and associations by the SBOE Outreach Team"; "have been made available to candidates filing for the 2016 election contests"; and can be "download[ed] from the SBOE's dedicated 'Voter ID' website." (Id. at 16–17.) As of December 11, 2015, the "SBOE ha[d] distributed over 105,000 copies of these materials, including Spanish-language materials." (Id. at 17.) In December 2015, the SBOE received an additional "300,000 flyers and 13,000 full-size posters" and as of that time planned to distribute these materials to CBOEs for

posting in public buildings throughout the State, such as county courthouses and offices, municipal government offices, town or city halls, health departments, public assistance agencies, vocational

rehabilitation and mental health centers, hospitals, schools, police stations, libraries, chambers of commerce, public transit and bus stations, senior centers, community centers, shelters and temporary/emergency housing, and other facilities open to the public.

(Id.) The SBOE's plans included dissemination of these materials through outside partners⁴¹ posting them at targeted locations, "includ[ing] educational institutions, food banks and pantries, retail and business establishments, churches, and other locations open to the public." (Id. at 17–18.) Pursuant to agreements reached with the University of North Carolina system, the North Carolina Community College system, and the North Carolina Independent Colleges and Universities, the State seeks to further disseminate print materials to the campuses of "every institution of higher learning in the State." (Id. at 18.)

41 As a result of the SBOE's partnership with the United Way, the helpline system operated by the United Way includes photo-ID related messaging. (Def. Ex. 535 at 19–20.) Last year this helpline assisted 125,000 callers with needs such as "gaining access to affordable child care, counseling and support groups, health care ... and help locating local food pantries and homeless shelters." (Id.at 19.) While callers are on hold, the helpline plays a recorded message containing information about the current photo-ID requirements for voting. (Id.at 20.) Agent counselors also provide answers to basic information about the photo-ID requirements based on training from the SBOE. (Id.) United Way's partnerships means that statewide distribution of election informational materials have extended to approximately twenty to sixty different affiliated agencies in each county. (Id.)

Further, on or about November 2, 2015, the State mailed a letter to those organizations who received a prior version of educational materials not including the reasonable impediment provision directing "that recipients should provide updated current information to any individuals to whom they disseminated the original materials or information." (Id. at 17.) The letter also offered the assistance of SBOE staff and included a form to order new materials. (Id.)

Statewide Media Campaign

The State has also implemented a substantial media outreach program for the current version of the photo-ID law.⁴² The State's initial ad was entitled "Be Seen. Be Heard," (Pl. Ex. 956), and began airing on television and radio stations across the State in December 2015, (Doc. 414 at 168). The ad informed voters that "[i]f you don't have an ID, or if you are unable to obtain one, voting options are available. For more information on exceptions, or for help getting a free ID, visit VoterID.nc.gov or call 866–522–4723." (Pl. Ex. 956.) The State's most recent television and radio ad is entitled "Be Recognized." (Def. Ex. 473.) The sixty second version of the advertisement provides:

42 SeeMcCrory, 156 F.Supp.3d at 694 n. 14, 2016 WL 204481, at 8 n. 14 (providing an overview of substantial media coverage of the reasonable impediment exception preceding the March 2016 primary).

This election, voters will be asked to show a photo ID at the polls. For most voters, you can simply bring your North Carolina Drivers License or ID card, passport, military or Veterans Affairs ID or certain tribal IDs. And, if there's something preventing you from getting one, no worries—you'll still be able to vote. Just come to the polls and we'll help you cast your ballot.

(Id. at 2–3.) The thirty second version of the ad is substantially the same. (Def. Ex. 472 at 2.) Both versions of the "Be Recognized" ad began airing on television and radio in early February 2015. (Doc. 414 at 170–72.)The State also intends to implement "an expansive outdoor advertising campaign to promote general awareness of the photo-ID requirements and exceptions." (Def. Ex. 535 at 14.) Director Strach testified that its "message will be displayed throughout North Carolina in rural, suburban, and urban areas on 40 vinyl billboards through November 2016, and 100 printed billboards through roughly August 2016." (Id.; Doc. 414 at 175.) Forty digital electronic billboards across the State also displayed the message from January through March 2016. (Def. Ex. 535 at 14.) Overall, the State estimates that 16.5 million passersby viewed its billboard messages on fifty-two billboards over a five-week period leading up to the 2014 general election. (Id.)

Information Provided on SBOE and CBOE Websites

The State has also used the SBOE's primary website, CBOE websites, and the SBOE's stand-alone website dedicated to the photo-ID requirement to educate voters about the reasonable impediment exception. (Id. at 13.) The SBOE's dedicated photo-ID website appears as the first result of a search on Google®> for "North Carolina voter ID." At the top of that site is the statement, "Most Voters Will Need to Show Acceptable Photo ID at the Polls." See N.C. State Bd. of Elections, www.voterid.nc.gov (last visited April 4, 2016). To the right of that statement is an image of acceptable forms of photo ID. Id. Below the statement is the sixty second video version of the "Be Recognized" ad, and below the video, in bold, pink letters is the statement, "Reasonable Impediment: Can't Get a Photo ID? Click Here." Id. Clicking on the accompanying link produces the following prominently-displayed statement:

Declaration of Reasonable Impediment

Voters who are unable to obtain an acceptable photo ID due to a reasonable impediment may still vote a provisional ballot at the polls. (Examples of a reasonable impediment include but are not limited to the lack of proper documents, family obligations, transportation problems, work schedule, illness or disability, among other reasonable impediments faced by the voter.)

Voters must also:

1. Sign a declaration describing their impediment; and

2. Provide their date of birth and last four digits of their Social Security number, or present their current voter registration card or a copy of an acceptable document bearing their name and address. (Acceptable documents include a current utility bill, bank statement, government check, paycheck, or other government-issued document.)

The provisional ballot will be counted when the information on the declaration is verified and all other eligibility requirements are met.

Id. The website has a "button" at the top labeled "Español" that allows users to receive voter-ID related information in Spanish. Id.

Judicial Voter Guide

The SBOE also included voter ID-related information as part of the State's Judicial Voter Guide, which is required by statute to be mailed to "every household in North Carolina not more than twenty-five days prior to the start of early voting in each election in which there is a statewide judicial contest." (Def. Ex. 535 at 14.) The front of the guide features a prominent statement informing voters that important information about the voter-ID requirement is contained inside. (Def. Ex. 537 at 1.) The statement also directs those who "can't obtain an acceptable photo ID" to the page of the guide where the reasonable impediment exception is described. (Id.) Pages four through six of the guide contain information on the voter-ID requirement. (Id.) Page four lists the acceptable forms of ID, page five describes the reasonable impediment exception, and page six lists other exceptions to the ID requirement. (Id. at 4–6.)

Targeted Mailing of Those Previously Contacted

Most pertinently, the SBOE has taken specific steps to reeducate those individuals that it previously contacted regarding the photo-ID requirement. As noted above, individuals who signed the "Acknowledgment of no Photo ID" form while voting and individuals appearing on no-match lists were mailed information about the need for photo ID in 2016 and how to acquire it. (Def. Ex. 535 at 7–10.) These mailings predated SL 2015–103. (Id. at 11.) After the fall elections in November 2015, the SBOE sent every individual who received a prior mailing (315,755 voters)—except those who had reported they already possess acceptable photo ID and those for whom prior mailings were returned to the SBOE as undeliverable—an additional mailing describing the reasonable impediment exception and other exceptions to the photo-ID requirement. (Id.) In December 2015, the SBOE sent a similar mailing to the 823 voters who indicated they lacked qualifying ID while voting during the 2015 municipal elections. (Id. at 8; Doc. 414 at 163–64; Def. Ex. 484.)

Election Official Training

The SBOE has provided CBOEs with substantial training on implementing the voter-ID requirement and the reasonable impediment exception. According to Director Strach, CBOEs "are responsible for providing in-person training to the local election workers and officials who will staff polling places," while the SBOE's role is to "provide[ ] oversight and resources to the counties' training efforts, including developing training materials and programs for use by [CBOEs]." (Def. Ex. 535 at 5; accord Doc. 414 at 139.) CBOE training of election workers has "historically [been] conducted in the months immediately preceding an election," and, Director Strach says, there is "no precedent for county boards of elections to train elections workers on new elections procedures before the training they will receive for the 2016 elections." (Def. Ex. 535 at 5–6.) According to Strach, "[t]raining election officials immediately in advance of an election is preferable to conducting the training at any earlier time ... [as it] allows the training to be fresh in the minds of election workers." (Id. at 6.) In addition, given that election workers "typically work only a few days each year, ... they receive training only on the procedures which will be in effect during the election for which they are being trained." (Id.)

As noted above, the SBOE began to develop and disseminate information on the reasonable impediment provision soon after it was enacted. In August 2015, the SBOE began to provide training to CBOE officials on the reasonable impediment exception at the statewide conference for CBOE members and staff. SeeN.C. State Conference of the NAACP v. McCrory, 156 F.Supp.3d 683, 697–98 n. 17, No. 1:13CV658, 2016 WL 204481, at *10 n. 17 (M.D.N.C. Jan. 15, 2016) (describing the mandatory nature of the meeting). On the first day of training, attendees received a presentation entitled "Duties and Responsibilities: Directors and Board Members." (Def. Exs. 480, 483; Doc. 414 at 151–52.) The presentation provided a non-exhaustive list of acceptable impediments, described the alternative documentation requirement, and provided guidance on counting provisional ballots cast under the reasonable impediment exception. (Def. Ex. 483.) On the second day of training, attendees received a presentation entitled "preparing for voter ID in 2016." (Def. Exs. 480, 482; Doc. 414 at 149–50.) The presentation contained information on acceptable photo IDs and exceptions to the photo-ID requirement, including the reasonable impediment exception. (See Def. Ex. 482.) Another presentation on the second day also featured information on the reasonable impediment exception. (Def. Exs. 465, 480; Doc. 414 at 147–48.)

In January 2016, the SBOE conducted regional training sessions for the CBOE elections personnel who would conduct the poll worker and election official training for the March 2016 primary. (Def. Ex. 535 at 5.) The SBOE encouraged CBOEs to invite poll workers and election officials to attend regional training. (Doc. 414 at 145.) Training sessions were held in Greenville, Buies Creek, Charlotte, Graham, Asheville, and Raleigh. (Def. Ex. 532.) The SBOE also made its presentation available via webinar so that additional election officials could receive training. (Doc. 414 at 146.) In total, 1,400 election officials and poll workers participated. (Id.)

A primary purpose of regional training was to provide training on how to use the SBOE's "Station Guide." (Id. at 145.) The Station Guide is a 123–page document that is placed "on every table or station at each polling site." (Id. at 183–84; Def. Ex. 531.) It is designed to provide election workers with "step-by-step instructions" for processing voters both with and without acceptable photo ID. (Doc. 414 at 183.) The January 2016 regional training presentation described the purpose and organization of the Station Guide; acceptable photo ID, along with pictures and expiration requirements; check-in procedures and the process for referring those without acceptable ID to the Help Station; the standard for applying the reasonable resemblance requirement; the reasonable impediment exception and the process for implementing it; and the parts of the reasonable impediment paperwork that must be completed by the voter and the parts that must be completed by the election worker. (Def. Ex. 532.)

Although the regional training presentation's primary reference material was the Station Guide, it also referenced and provided links to on-demand training videos that the SBOE made available to CBOEs in December 2015. (Id.; Def. Ex. 476.) The training videos consist of eleven modules for use by CBOEs in training their election workers. (Doc. 414 at 140.) The tenth module covers the reasonable impediment exception and provides the election worker with a hands-on demonstration of how to process a reasonable impediment voter. (Id. at 140–41; Def. Ex. 476 at 17–20.)

Although the Station Guide and training modules are designed to be detailed, the more comprehensive guide to election administration is the Election Official Handbook ("Handbook"). (Doc. 414 at 183–84; see Def. Ex. 475.) The Handbook provides guidance on "every aspect of the voting experience [and] ... step-by-step instructions and scripted language to deal with any potential scenario that an election official may encounter."⁴³ (Doc. 414 at 141; see Def. Ex. 475.) A copy of the Handbook will be available at every early-voting and Election Day polling place in 2016. (Doc. 414 at 142.)

43 "In the event that the Handbook fails to address a particular eventuality, poll workers will be trained to contact their county board of elections staff and/or SBOE personnel for assistance." (Def. Ex. 535 at 4.)

On February 1–2, 2016, the SBOE conducted a statewide educational conference for CBOE elections personnel. (Id. at 146.) Director Strach gave a presentation dedicated exclusively to voter-ID requirements and exceptions. (Def. Ex. 551.) The presentation described various training tools, including the video modules discussed above, the Station Guide, and webinars, (id. at 6); the roles of various election workers in applying the ID requirement, including greeters, check-in workers, curbside attendants, Help Station workers, judges of election, the county office, and the CBOE, (id. at 8–15); acceptable forms of photo ID and expiration requirements, (id. at 17–22); the options for voting when the voter lacks qualifying ID, (id. at 24); the process for evaluating reasonable resemblance, including a reasonable resemblance flowchart, (id. at 28–42); the reasonable impediment declaration process and when voters qualify to use it, (id. at 45–47); the challenge process for reasonable impediment declarations, including burdens of proof and standard of review, (id. at 52–58); and the alternative documentation requirement of the reasonable impediment exception, (id. at 49). The SBOE's Veronica Degraffenreid presented on voting site uniformity and provided substantial training on voter-ID requirements and exceptions. (Def. Ex. 553.) The uniformity presentation described election worker training resources for CBOEs, (id. at 3–6 (Part I)); the role of the various voting stations in processing voters, both with and without qualifying ID, (id. at 12–16); required signage, including voter ID related signage, (id. at 17–19); materials to be handed out to voters, including the voter-ID push cards described above, (id. at 20); the Station Guide's organization and purpose, (id. at 27–30); acceptable ID and expiration requirements, (id. at 31–39; id. at 1–2 (Part II)); step-by-step check-in procedures for voters with and without acceptable ID, (id. at 5–16); the reasonable resemblance requirement and the applicable standards to apply, (id. at 14–15); and the provisional voting options available to voters without acceptable ID and how to implement them at the Help Station, including the reasonable impediment exception, (id. at 6–15 (Part III); id. at 1–19 (Part IV)). Both presentations were detailed and extensive. (See Def. Exs. 551, 553.) A final presentation, entitled "What Will you Do?," quizzed attendees on their knowledge. (Def. Ex. 554.) The presentation displayed various types of photo ID for hypothetical voters and asked attendees whether the ID was acceptable for voting. (Id. at 5–18.) The presentation also quizzed attendees on how to process voters, (id. at 21–29), how to apply the reasonable resemblance requirement, (id. at 51–58), and how to apply the reasonable impediment exception, (id. at 64, 70).

In sum, the SBOE has engaged in substantial efforts to educate voters and election officials about the requirements of and exceptions to the voter-ID requirement.

c. Voters' Experience in Acquiring Qualifying ID

To acquire a free voter ID, voters must present at a DMV location providing North Carolina DMV services. Testimony at trial indicated that this process, at least for some, has not been as easy as one might expect.

To acquire a free voter ID, an applicant must do the following: (1) be a registered voter or complete a voter registration application at the time of applying for a voter ID; (2) sign a declaration stating that the registrant does not have an acceptable ID to vote; (3) provide proof of North Carolina residency, or, in the alternative, sign an affidavit of residency (there is no cost for this affidavit when applying for free voter ID); (4) provide a valid SSN; and (5) prove age and identity by providing two supporting documents. (Def. Ex. 533 at 2 (tbl. 4).)

There has been some inconsistency within the DMV about which supporting documents are sufficient to prove age and identity. (See Doc. 410 at 180–81.) Historically, the "officially acceptable" list of documents has been published in "Table (1)" of the DMV's required-documents form, "DL-231", which appears on its website. (Seeid. at 180–81, 187.) Documents in Table (1) are more traditional forms of supporting identification and include (1) a driver's license or State-issued ID card from North Carolina, another State, Puerto Rico, a U.S. territory, or a Canadian province (expired less than two years); (2) a certified birth certificate⁴⁴ issued by a government agency in the United States, Puerto Rico, a U.S. territory, or Canada or U.S. Report of Consular Birth Abroad; (3) an original social security card; (4) tax forms that reflect the applicants full name and full SSN; (5) a Motor Vehicle Driver's Record; (6) a North Carolina school transcript or registration signed by a school official, or a diploma or GED from a North Carolina school, community college, or North Carolina university; (7) a valid and unexpired U.S. military ID; (8) a valid, unexpired passport from any nation; (9) a certified document from a Register of Deeds or government agency in the United States, Puerto Rico, a U.S. territory, or Canada; (10) a limited driving privilege issued by a North Carolina Court (expired not more than one year); valid, unexpired documents issued by the Department of Homeland Security or the United States Citizenship and Immigration Services; or (11) a court document from a U.S. jurisdiction, Puerto Rico, a U.S. territory, or Canada. (Def. Ex. 533 at 1 (tbl. 1).)

44 As noted, SL 2013–381 prohibits the State from charging any fee "to a registered voter who signs a declaration stating the registered voter is registered to vote in [North Carolina] and does not have a certified copy of that registered voter's birth certificate or marriage license necessary to obtain [acceptable] photo identification." N.C. Gen. Stat. § 130A–93.1.

As a practice, however, DMV examiners did not always limit themselves to documents in Table (1). (Doc. 410 at 181.) Over time this took the form of an alternate document list, which benefitted applicants. (Id.) This list was not publicly available, and it does not appear to have been uniformly followed by examiners. (See id.) But it was used by at least some examiners from January to August 2014. (Id.) In any event, in January 2016, the DMV officially incorporated documents on the alternate list into the list of acceptable supporting documentation and made that list available on its website. (Id. at 186–87.) Voters are still encouraged to bring a document from Table (1), but the DMV will consider the following unexpired forms of alternate identification in an application for a free voter-ID: (1) certificate of adoption; (2) college or student ID; (3) concealed handgun permit; (4) Department of Revenue tax document; (5) employee or Government ID; (6) extended health care facility record; (7) hunting or fishing license; (8) license to carry firearms; (9) life insurance policy; (10) Medicaid/Medicare card; (11) medical, clinic, or hospital record; (12) military dependent's ID card; (13) military draft record; (14) passport card; (15) payment statement or check stub; (16) prison ID or inmate record; (17) retirement benefits record; (18) traffic citation or court record; (19) U.S. Coast Guard merchant ID card; (20) U.S. vital statistics official notification of birth registration. (Def. Ex. 533 at 2 (tbl. 4).) Because this list is not exhaustive, the DMV will review any documents that an applicant has in his possession. (Id.)

Once a voter has all the necessary materials to acquire a free voter ID, he must travel to a DMV office providing driver's license services. (Pl. Ex. 1044 at 142.) The evidence indicated that as of the beginning of January 2016, approximately 2,172 applicants had sought no-fee voter-ID cards across the State, and 2,139 had been issued. (Doc. 410 at 177–79; Def. Ex. 494.) The DMV currently has 114 brick and mortar sites that provide driver's license services. (Doc. 410 at 164.) Some have limited hours. For example, in Allegany County, the location is only open on "Wednesday and Thursday 9:30 a.m. to 12:00 p.m., 1:00 p.m. to 3:30 p.m." (Pl. Ex. 1044 at 166–167.)

Sixteen of North Carolina's 100 counties do not have a brick and mortar site. (Pl. Ex. 241 at 13.) Eleven of these counties are serviced by five DMV mobile units, which currently appear at twenty-four mobile sites. (Id.; Doc. 410 at 198–99.) No mobile site offers services more than three days per month. (Doc. 410 at 204; Pl. Ex. 241 at 13 n.3.) Nevertheless, the DMV estimates that 98% of the its "market population" (those age 15 and older) lives within a thirty-minute drive of a DMV license service station, whether a brick and mortar or mobile site. (Doc. 410 at 168–69.) A December 2013 customer survey indicated that wait time, not DMV accessibility, was a top concern for respondents. (Pl. Ex. 1044 at 82–83; Doc. 410 at 156–57.) While the current DMV sites may be adequate for those with access to a vehicle, the court has substantial questions about the accessibility of free voter ID for those who lack transportation, especially in rural communities that lack public transportation.

At trial, Plaintiffs presented the video depositions of several witnesses who experienced difficulty in acquiring certain qualifying licenses from the DMV. Their depositions were taken prior to SL 2015–103's enactment of the reasonable impediment exception.

Alonzo Phillips is a sixty-one year old African American male who lives with his mother in Halifax County, North Carolina. (Pl. Ex. 1048 at 5.) Ten years ago, Mr. Phillips attempted to acquire a non-operator ID card (not a license) from the DMV, available for a nominal fee. (Id. at 14.) He presented his social security card and birth certificate, but the DMV refused to issue him an ID because his birth certificate listed "Alonz," while his social security card displayed his correct name, Alonzo. (Id. at 15.) To the extent Mr. Phillips seeks to use his birth certificate as a supporting document in the future, federal law requires him to correct the document.⁴⁵ Under federal law, "a driver's license cannot be issued unless there is an exact match of your name, SSN, and birthdate with the Social Security Administration." (Doc. 410 at 183.)⁴⁶ Moreover, because Mr. Phillips was born in New York, correcting his birth certificate would be logistically difficult for him. (Pl. Ex. 1048 at 17–18.) However, his aunt, who provides him with support, has offered to help him get a court order to update his birth certificate. (Id. at 46–47.) Most importantly, there is no indication that Mr. Phillips will have difficulty voting under the reasonable impediment exception. In fact, he testified that he can walk to his voting location, which is close by, and that he has a cellphone and would feel comfortable calling the SBOE for help. (Id. at 46, 52.)

45 A birth certificate, of course, is only one form of acceptable supporting documentation, of which there are many. (SeeDef. Ex. 533 at 1 (tbl. 1), 2 (tbl. 4).)

46 It was undisputed at trial that federal law requires such a match. No party provided a citation to the applicable federal law in their conclusions of law and findings of fact, but the requirement appears to derive from the REAL ID Act of 2005, Pub. L. No. 109–13, 119 Stat. 302, 312.

Rosanell Eaton is a ninety-three year old African American and a lead plaintiff in this case. She is, in every respect the court can imagine, a remarkable person. In January 2015, Ms. Eaton presented to the DMV to renew her driver's license. (Pl. Ex. 1045 at 18–19, 25.) It is not clear from her testimony, but it appears that she was concerned whether her current license would comply with the new law. Because the name on her birth certificate (Rosanell Johnson) did not match the name on her social security card, federal law prohibited the DMV from issuing her a driver's license. (See id. at 19–20.)⁴⁷ Ms. Eaton testified that the DMV told her she needed to get her SSN changed. (Id. at 20.) Presumably, she was actually told to get the name on her social security card changed so it matched the name she sought to use at the DMV, but here, too, the record is not clear. (See id. at 19–20.) In any event, Ms. Eaton says the DMV refused to take further action until she made changes at the social security office. (Id.). It took her ten trips (and two tanks of gas) back-and-forth between the DMV and the social security office before she got her license on January 26, 2015. (Id. at 20–21, 24, 51, 55.) Ms. Eaton is confident now that she will be able to vote using her new license. (Id. at 29.)

47 When she married, she had changed her name to Rosanell Johnson Eaton, (Pl. Ex. 1045 at 20), but her license was in the name of Rosa Johnson Eaton, (Pl. Ex. 300).

Ms. Eaton's testimony does not make clear why her ordeal was so involved, but it is troubling that any individual could be subjected to such a bureaucratic hassle. Here, too, the problems delaying her license renewal stem from a federal law requirement. It is unclear why she did not encounter this problem previously, as her last renewal was in April 2010. (Pl. Ex. 300.) But most importantly, the voter-ID law did not require Ms. Eaton to endure this hassle. To the extent Ms. Eaton wanted to continue to drive, which appears to have been the case, she had to renew her license. (Pl. Ex. 1045 at 49–50.) Driver's licenses required renewal long before SL 2013–381. But, to the extent she simply wished to comply with the ID requirement, her expiring license would have been compliant for voting. As a voter over seventy, she can use an expired license indefinitely so long as it expired after her seventieth birthday, which hers did.⁴⁸SeeN.C. Gen. Stat. § 163–166.13(f) ; (Pl. Ex. 300). Also, under the law's "reasonable resemblance" requirement—which Plaintiffs do not challenge—a photo ID is acceptable as long as the name appearing on it is "the same or substantially equivalent to the name contained in the registration record." 08 N.C.A.C. 17.0101(c). Ms. Eaton's previous license meets this test. ("Rosa Johnson Eaton" on her former license, (Pl. Ex. 300), compared to her voter registration card of "Rosanell Eaton," (Pl. Ex. 302)).⁴⁹

48 This exception arguably existed even under SL 2013–381, but SL 2015–103 clarified any ambiguity.

49 It appears that her current registration is in the name of "Rosa Nell Eaton," seeNC Public Voter Information ("Rosa Nell Eaton"), https://enr.ncsbe.gov/voter_search_public/voter_details.aspx?voter_reg_num=000000015723&county=35 (last visited April 5, 2016), which also qualifies. "Rosa" qualifies as a "variation or nickname" of her formal first name. 08 N.C.A.C. 17.0101(c)(4)(B) (illustrating that "Sue" is an acceptable variation or nickname of Susanne). "Nell" as her middle name qualifies as a typographical error. Id.at 17.0101(c)(4)(F). Finally, "Johnson" qualifies as the use of a former/maiden name. Id.at 17.0101(c)(4)(D).

Silvia Kent is a caretaker for her three disabled sisters, Katherine, Ester, and Faydeen. (Pl. Ex. 1049 at 12–13.) Ms. Kent's sisters are registered to vote and vote regularly. (Id. at 15–16.) While taking her sisters to vote in the November 2014 election, Ms. Kent was informed that voter ID would be required beginning in 2016. (Id. at 16–17.) Ms. Kent then took her sisters to the DMV to acquire ID. (Id. at 19.) Katherine was issued an ID, but Esther and Faydeen were not. (Id.) The birth date on Esther's supporting ID was incorrect. (Id. at 22–23.) The DMV examiner told Ms. Kent that she would need to go to the Register of Deeds to remedy the problem. (Id. at 23.) But when Ms. Kent did, she discovered another problem: the spelling of Esther's name on her birth certificate and the date of birth listed were incorrect. (Id. at 26–27.) Ms. Kent claims she was told that the Register of Deeds could do nothing about the error. (Id. at 27.) A similar problem blocked Faydeen from acquiring an acceptable ID. (Id. at 29–30.) Ms. Kent attributes her sisters' problems to the fact that, despite being born in the 1940s, their birth certificates were not filed until 1962. (Id. at 35–36.) Notwithstanding having sought legal assistance and spending "lots of hours" attempting to get ID for her sisters, as of June 5, 2015, neither Esther nor Faydeen had acquired acceptable ID. (Id. at 34, 38–39.) Nothing in Ms. Kent's testimony indicates that Ester or Faydeen would have difficulty voting under the reasonable impediment exception.

Maria Del Carmen Sanchez is a fifty-eight year-old United States citizen born in Cuba. (Pl. Ex. 1051 at 9–11.) She took the name Thorpe when she married her husband. (Id. at 25.) She has lived in North Carolina since 1990 and is registered to vote under the name Maria Sanchez Thorpe. (Id. at 11–12.) Her unexpired passport lists her name as Maria Del Carmen Sanchez. (Pl. Ex. 836.) In 2007, six years before SL 2013–381, Ms. Sanchez went to the DMV to renew her expiring license, which bore the name (consistent with her voter registration) Maria Sanchez Thorpe. (Pl. Ex. 1051 at 23; Pl. Ex. 839.) She attempted to use her passport as a supporting document to establish her age and identity. (Pl. Ex. 1051 at 23.) But, because the name did not match the name listed with the DMV, it would not issue her a license.⁵⁰ (Id. at 24–25.) According to her, the DMV told her that the only way she could cure her problem and use her passport to establish identity was to drop Thorpe from her name by "get[ting] a divorce or ... go[ing] to [the] Social Security Administration." (Id. at 34.) After some research, Ms. Sanchez learned she could file an affidavit for name change at the DMV. (Id. at 35–36; Pl. Ex. 840.) She and her husband returned to the DMV, where she instructed DMV workers as to the location of the affidavit, completed it, had it notarized, and received a new license. (Pl. Ex. 1051 at 36–37.)

50 At the time, Ms. Sanchez considered her legal name to be Maria Sanchez Thorpe. (SeePl. Ex. 1051 at 25.) Her driver's license from 1994 had used that name, and she testified that after getting married she updated "everything"—except her Passport—to use the name Thorpe. (Seeid.) She never updated her passport because "it never occurred to [her] to change [her] passport" and that it "didn't seem important." (Id.)

While Ms. Sanchez's testimony demonstrates the ineptitude of government bureaucracy, her difficulty arose from her failure to update her documents after getting married and changing her name. (Id. at 25–26.) Had she updated her passport as she did her other legal documents, she could have used it to renew her license without issue. The DMV should be faulted for being unaware of the name change affidavit option, but it cannot be faulted for finding Ms. Sanchez's passport bearing a different name to be insufficient to establish her identity. In any event, Ms. Sanchez's testimony is not especially probative of the current state of affairs at the DMV, as her incident occurred almost six years before SL 2013–381 and over eight years before trial.

Balanced against these testimonials was evidence of significant improvements at the DMV since the passage of SL 2013–381. In October 2013, retired U.S. Army Brigadier General Kelly J. Thomas took over as the Commissioner of Motor Vehicles for the State of North Carolina. (Doc. 410 at 153–54.) Commissioner Thomas was given a broad mandate to be the "lead change agent" for the DMV. (Id. at 154–55.) His first task was to "analyze the problems at DMV." (Id. at 155.) To do so, he commissioned a "Voice of the Customer" survey in December 2013 that identified eighteen things that DMV customers wanted fixed. (Id. at 156.) Respondents were brutally frank:

The survey said that our DMV employees and our process was not helpful. They said that we were very ugly to customers. They thought that we didn't like our jobs—our employees didn't like their jobs. They thought the process was cumbersome. They wanted online access to more DMV access, practices and functions. They wanted credit card and debit card access.

(Id. at 157.) At trial, Commissioner Thomas testified credibly that DMV has addressed sixteen of the eighteen issues identified. (Id. at 156–57.) The two remaining are fingerprinting of customers and electronic identification, which are in study. (Id. at 157.)

To address customer service, Commissioner Thomas partnered with Wake Technical Community College to create "customer 101" training for DMV employees. (Id. at 162.) In addition, recognizing that DMV examiners did not have any form of continuing education, he implemented a continuing education program for all 550 examiners. (Id. at 163.) The DMV is also in the process of implementing an online training program for examiners, which is intended to educate examiners on changes as they happen. (Id. at 164.)

The DMV has also implemented several strategies to address customer wait times. It has implemented online license renewal, used by 219,000 North Carolinians since June 2015 and saving over 60,000 wait hours, (id. at 165–66); significantly updated examiner stations to provide each examiner with a "customer-facing computer screen," vision and sign tester, credit card/debit card machine, and camera for taking ID pictures, (id. at 166–67); and rolled out an extended-hours project (keeping offices open until 6 p.m. at twenty-one sites and every Saturday until noon at eleven others) to "offer hours that citizens didn't have to take off work," (id. at 168; Pl. Ex. 664 ¶¶ 80–83). When the extended-hours program was offered at nineteen locations (now twenty-one), eighty-six percent of DMV's market population lived within a thirty-minute drive of an extended-hours office. (Doc. 410 at 168.) As a result of these and other changes at the DMV, the average customer wait time across the State since April of 2015 has been reduced to nineteen minutes and forty-two seconds. (Id. at 170.)

In an effort to bring DMV locations closer to customers, the DMV has made substantial efforts to expand its mobile unit fleet. When Commissioner Thomas took over, there were five Winnebago mobile units, yet only one was operational. (Id.) Recognizing that the aging and maintenance-prone mobile units were not a productive option, the DMV has developed a "footlocker" mobile unit. (Id. at 170–71, 173.) These approximately 100–pound footlockers are basically a single examiner station in a box. (Id. at 173–74.) They can be hauled in the back of a truck and can plug into a standard electrical outlet. (Id.) The DMV has a patent pending on the units, and Commissioner Thomas says that other States are interested in them. (Id. at 174.) The State plans to have fourteen footlocker units operating across North Carolina. (Id. at 175.) Commissioner Thomas projects that the footlockers will allow the DMV to go from the twenty-four mobile sites it has today to "45 by the end of April ... [and] 70 by the summer of 2016." (Id. at 211.) He sees this development as a key part of his goal to put a DMV brick and mortar or mobile site within a twenty-minute drive of 98% of the DMV's market population. (Id. at 175–76.) That said, at the time of trial the DMV only had one footlocker mobile unit in operation, along with four Winnebago units. (Id. at 198–99.)

All told, the DMV was an obvious choice to issue voter IDs, given its experience in issuing driver's licenses, but it is undisputed that the agency had significant flaws in 2013. There is little persuasive evidence the legislature was aware of them, and the law did contemplate a two-and-one-half-year rollout. The evidence showed that the DMV has made substantial improvements under Commissioner Thomas during this time period. Nevertheless, the DMV has room for continued improvement, and Commissioner Thomas had to concede as much. For example, Plaintiffs' counsel identified substantive inaccuracies in an April 10, 2015 training presentation on the voter-ID requirement that mistakenly advised, without qualification, that another State's license was an acceptable photo ID. (See id. at 207–09.)⁵¹ Moreover, the evidence demonstrated that, for some voters, personal circumstances, including mismatched or error-riddled documents, made acquiring even a free voter ID more complicated than one might expect. No doubt, the experience of individuals like Mr. Phillips, Ms. Eaton, and Ms. Thorpe highlighted these problems and were likely instrumental in the legislature's adoption of a reasonable impediment exception.

51 This training module was offered to examiners beginning in May 2015. (Doc. 410 at 208.) One slide of the presentation correctly informed examiners that customers are not eligible to receive free voter ID unless they have "none of the other forms of ID acceptable for voting." (Def. Ex. 534 at 4.) However, another slide educated examiners that a driver's license issued by another state is an acceptable form of photo ID. (Id.at 6.) This is not always true, as this form of identification is only acceptable if the voter registered "within 90 days of the election." N.C. Gen. Stat. § 163–166.13(e)(8).

d. Evidence of North Carolina Voters Without ID

Plaintiffs claim that hundreds of thousands of registered North Carolina voters lack qualifying ID and that African Americans are disproportionately likely to be among them. In light of the adoption of the reasonable impediment exception, the focus of this contention is redirected to include the adequacy of the exception as a legal matter and the burden of its use.

Plaintiffs' expert, Dr. Stewart, testified as to results of his attempts to match North Carolina registered voters' names to names in various databases of acceptable IDs.⁵² Dr. Stewart describes database matching as "the technique of allowing the computer to take unique information about an entity in one database and find the information related to that entity in another database using some form of variables to link." (Doc. 408 at 8.) Database matching is commonly used in the social sciences. (Id.) In some databases, the entities have a unique identifier that can be matched. (Id. at 27, 29.) North Carolina's voter registration database does not contain a unique identifier for registrants. (Id. at 26–27.) In an attempt to compensate for this shortcoming, Dr. Stewart designed a series of iterative "sweeps" intended to match voters on the basis of non-unique identifiers. (Id. at 27.)

52 Dr. Stewart did not have access to a database "identifying holders of tribal IDs or out-of-state driver's licenses." (Doc. 419 at 22 n.6.) Therefore, registered voters who only possess these forms of acceptable ID appear as no-matches in Dr. Stewart's analyses.

Dr. Stewart's first no-match list was based on a July 16, 2014 snapshot of the voter registration files. (Pl. Ex. 242 at 11.) He found that at least 397,971 (6.1%) registered voters could not be matched to a qualifying ID. (Id. at 50 (tbl. 11).) By race, he found that 147,111 (10.1%) African Americans could not be matched to a qualifying ID, compared to 212,656 (4.6%) whites. (Id. at 38 (tbl. 7).)

Dr. Stewart updated his no-match analysis in December 2015, after SL 2015–103, to account for its expansion of acceptable IDs, but he continued to rely on his July 16, 2014 snapshot of data. (Pl. Ex. 891 at 14 n.26.) Thus, despite relying on data that is over a year and a half old during a period of rolling out notice of the photo-ID requirement, he proffers it as the "best estimate" of current conditions. (Doc. 408 at 52.) Dr. Stewart incorporated at least some of Defendants' critiques from his prior no-match list, which he characterizes as his "refined matching criteria." (Id. at 19 (tbl. 11).)⁵³ Based on this updated criteria, he found that 224,863 (3.5%) registered North Carolina voters could not be matched to having an acceptable ID. (Id.) Thus, SL 2015–103's expansion of acceptable IDs (driver's licenses and non-operator's IDs that have been expired for less than four years) and Dr. Stewart's refinement of his matching criteria reduced the size of the no-match list by 173,108 voters. (Id. at 14 n. 26, 19 (tbl. 11); Pl. Ex. 242 at 50 (tbl. 11).) By race, Dr. Stewart found that 83,470 (5.7%) African Americans could not be matched to a qualifying ID, compared to 116,344 (2.5%) whites.⁵⁴ (Pl. Ex. 891 at 19 (tbl. 11).)

53 Defendants had identified several flaws in Dr. Stewart's methodology. For example, they argued that Dr. Stewart's failure to include the "Driver_Hist" table from the DMV database ("SADLS") led him to include at least 38,801 voters on his no-match list even though the "Driver–Hist" table indicated they had an unexpired, DMV-issued ID. (Doc. 391 at 14; Pl. Ex. 891 at 14 n.26.) By using the "Driver-Hist" table, Dr. Stewart was able to make additional matches. (Pl. Ex. 891 at 14 n. 26.) A substantial number of additional matches were able to be made on account of the "refined matching criteria." The updated criteria reduced the size of the no-match list by 47,837 (272,700—224,863). (Pl. Ex. 891 at 19 (tbl. 11).) By race, the updated criteria reduced the number of African Americans on the no-match list by 14,988 (98,458—83,470). As a percentage of registered voters, this reduced the percentage of African Americans who could not be matched from 6.8% to 5.7%. (Id.) As for whites, the refined criteria permitted 28,876 (145,220—116,344) whites to be matched. (Id.) That reduced the percentage of whites who could not be matched from 3.2% to 2.5%. (Id.)

54 Dr. Stewart did not include Hispanics as a separate group in his no-match analysis. (Pl. Ex. 891 at 19 (tbl. 11).)

In sum, Dr. Stewart was able to match 94.3% of African American registrants, 97.5% of white registrants, and 96.5% of all registrants to a qualifying ID. (Id.)

Dr. Stewart's matching analysis in this case differs materially from his analysis in the DOJ's attempt to block preclearance of South Carolina's voter photo-ID law, which also includes a reasonable impediment exception. (Doc. 408 at 60–61.) First, because South Carolina's voter registration database contained a unique identifier—the voter's full SSN ("SSN9")—his results were more precise and were not based on the multiple "sweeps" of the lists that were necessary here. (Id. at 28–29.) Second, he did not include "inactive" voters in his South Carolina no-match analysis because, in his view, they were less "likely to vote in the future" and more likely to "soon ... be moved to 'archived' status," (id. at 21–22), but he did here.⁵⁵ This significantly increased the size of the no-match list in this case.⁵⁶

55 Both States define an "inactive" voter substantially the same. (Doc. 408 at 66.) In both States, an inactive voter can present at the polls and vote if certain conditions are satisfied. (Id.at 131–32.) Dr. Stewart's primary reason for including inactive voters here, despite excluding them in South Carolina, is that they are, regardless of status, registered voters and are able to vote. (Id.at 156–57.)

56 There is an additional distinction between Dr. Stewart's no-match analysis in South Carolina and here. In South Carolina, Dr. Stewart excluded individuals that were flagged as having received a license in another State on the logic that they were less likely to be a resident of South Carolina. (Doc. 408 at 46–47, 90–91.) Here, however, Dr. Stewart permitted such persons to appear on the no-match list. (Id.at 90–91.)

Despite breaking his results down by active and inactive voters in his first no-match list, Dr. Stewart omitted it from his December 2015 report even though he had apparently done the analysis. (Pl. Ex. 891; Doc. 408 at 138.) However, after cross-examination at trial, he produced an additional exhibit with the no-match results. (Pl. Ex. 1063.) Dr. Stewart found that 151,005 (2.6%) active voters could not be matched to a qualifying ID. (Id.) By race, 60,312 (4.8%) African American active voters could not be matched to a qualifying ID, compared to 73,143 (1.8%) whites. (Id.) Thus, Dr. Stewart was able to match 95.2% of African American active voters, 98.2% of white active voters, and 97.4% of all active voters.

The comparison between Dr. Stewart's North Carolina and South Carolina analyses is telling. As noted above, even though Dr. Stewart had a unique identifier in South Carolina, he matched a higher percentage of active and inactive voters in North Carolina than he did active voters in South Carolina (96.5% in NC vs. 93.3% in SC). (Pl. Ex. 891 at 19 (tbl. 11); Def. Ex. 504 at 36 (tbl. 4).)⁵⁷ This is true when broken down by race (94.3% of African Americans in NC vs. 90.5% in SC; 97.5% of whites in NC vs. 94.5% in SC). (Id.) The magnitude of the difference only becomes more pronounced when the percentage of active voters matched in North Carolina is compared to the percentage of active voters matched in South Carolina, which is the more apples-to-apples comparison (97.4% in NC vs. 93.3% in SC.). (Pl. Ex. 1063; Def. Ex. 504 at 36 (tbl. 4).) This, too, is true when broken down by race (95.2% of African Americans in NC vs. 90.5% in SC; 98.2% of whites in NC vs. 94.5% in SC). (Id.) Notably, the percentage of matched voters in North Carolina exceeds the data relied upon by the three judge panel that upheld the South Carolina law.⁵⁸To be sure, Dr. Stewart's no-match list purports only to note the lack of a match; it does not equate to lack of a qualifying ID. Even if this court were to assume that everyone on Dr. Stewart's December 2015 no-match list lacks a qualifying ID, however, the data suggest that the number who would wish to vote, and thus use the reasonable impediment exception, will be very low. As noted above, the SBOE removed 52,765 individuals from Dr. Stewart's list of 224,863 voters as part of its list maintenance and address validation process and sent the 172,098 remaining individuals a mailing asking whether or not they have qualifying ID. (Def. Ex. 511 at 16–17.) Defendants' expert, Janet Thornton, Ph.D., a labor economist and applied statistician,⁵⁹ performed an analysis of the voting history of those 172,098 individuals. (Id. at 18 (tbl. 6).) She found that 69.8% had not voted in 2012 or 2014 and that 39.5% had never voted. (Id.) Broken down by election, 92.6% did not vote in the 2012 primary, 72.1% did not vote in the 2012 presidential election, 96.3% did not vote in the 2014 primary, and 87.8% did not vote in the 2014 midterm. (Id.(tbl. 7).)

57 At trial, Defendants used exhibit 504 to make the comparison to South Carolina, without objection. Dr. Stewart's no-match analysis in that exhibit was limited to South Carolinians who could be matched to having a "valid driver's license/ID" card. (SeeDef. Ex. 504 at 36 (tbl. 4).) It did not include other acceptable forms of ID. (Id.) Dr. Stewart later supplemented this analysis with a no-match list that took into account "military IDs and passports." (SeeDef. Ex. 311 at 33 (tbl. 11).) Even though a copy of Defendants' exhibit 311 was given to the court, it does not appear to have been moved into evidence. So, the court does not rely upon it. It bears noting for the sake of comprehensiveness, however, that Dr. Stewart's updated no-match list from South Carolina (reflected in Defendants' exhibit 311) nevertheless exceeds his December 2015 no-match list for North Carolina. After taking other acceptable forms of ID into consideration, Dr. Stewart was able to match 94.8% of active South Carolina voters (compared to 96.5% of active and inactive voters in NC), including 91.7% of African Americans (compared to 94.3% in NC) and 96.1% of whites (compared to 97.5% in NC). (SeePl. Ex. 891 at 19 (tbl. 11)); Def. Ex. 311 at 33 (tbl. 11).) Of course, as noted herein, the magnitude of the difference only becomes more pronounced when the percentage of active voters matched in North Carolina is compared to the percentage of active voters matched in South Carolina, which is the more apples-to-apples comparison. (SeePl. Ex. 1063; Def. Ex. 311 at 33 (tbl. 11).)

58 Despite the data provided by Dr. Stewart and the racial disparities that existed in ID possession, the three-judge panel in South Carolina v. United States, 898 F.Supp.2d 30, 40 (D.D.C.2012), precleared South Carolina's voter-ID law under § 5 of the VRA in light of its reasonable impediment exception. Perhaps recognizing that the reasonable impediment exception made the exact number of individuals without qualifying ID less critical, the court relied on ballpark figures. Id.It found that "about 95% of South Carolina registered voters," and "[a]bout 96% of whites and about 92–94% of African Americans," had qualifying ID. Id.Even under the analysis of Dr. Stewart, the percentage of matched voters in North Carolina exceeds those matched and relied upon by the three judge panel (96.5% to "about 95%"; 97.5% to 96%; 94.3% to "[a]bout 92–94%"). Id.; (Pl. Ex. 891 at 19 (tbl. 11)).

59 Dr. Thornton was proffered "as an expert in the field of economic and applied statistical analysis." (Doc. 338 at 92.) Plaintiffs did not object to Dr. Thornton's data analysis and statistical analysis. (Id.at 93.)

As of December 30, 2015, 45,692 of the SBOE's mailings had been returned to the SBOE; 38,815 of those were undeliverable, (id. at 17), and 4,992 stated they already had a qualifying ID, (id. at 18 (tbl. 6)). Although these respondents do not reveal how many of the "undeliverable" or "non-responses" lack qualifying ID, the data show that these respondents behave much more like normal voters than the no-match list as a whole. For example, only 30.1% of those who said they have qualifying ID did not vote in the 2012 presidential election. (Id.(tbl. 7).) This is consistent with the turnout of that election, where 67.2% of registered African Americans, 60.4% of registered whites, and 60.9% of all registered voters voted. (Pl. Ex. 242 at 161 (App'x U).)

Defendants' expert, M.V. Hood, III, Ph.D., Professor of Political Science at the University of Georgia,⁶⁰ provided a glimpse into the likely prevalence of reasonable impediment voting by providing data from South Carolina, the only other State to offer that option. In the 2014 general election, of 1,195,741 votes cast, 131 reasonable impediment affidavits were completed.⁶¹ That constitutes 1.1 reasonable impediment affiants for every 10,000 voters. (Def. Ex. 500 at 2.) Of course, the 2014 election was not a presidential election, and North Carolina has substantially more voters than South Carolina. (Pl. Ex. 242 at 161 (App'x U).) However, there is simply no evidence in this case to suggest that more than a fraction of a percent of voters will rely upon the reasonable impediment exception.

60 Dr. Hood was proffered as an expert "in the field of American politics ... [and] the areas of electoral politics, racial politics, southern politics, and election administration" without objection. (Doc. 339 at 56–57.)

61 Dr. Hood acquired data from every county in South Carolina except Spartanburg County. (Doc. 410 at 234–35.) He requested the data but was told it did not exist. (Id.) If any reasonable impediment affidavits were completed in that county, they are not reflected in Dr. Hood's analysis.

Further, the characteristics of individuals on Dr. Stewart's no-match list raise serious questions about its reliability. For example, those on Dr. Stewart's no-match list were far more likely not to have included the last four digits of their SSN ("SSN4") on their registration: 59.4% of those on the no-match list were missing their SSN4, compared to 14.6% of registered voters. (Def. Ex. 511 at 12 (tbl. 1).) It may be that those who do not include their SSN4 are less likely to have an ID. But another at least equally plausible explanation is that North Carolina's voter registration form did not ask for a registrant's SSN4 prior to 2004, and even today the information is not required. (Doc. 408 at 116.) Half of Dr. Stewart's no-match list registered before 2004. (Doc. 416 at 40.) As Dr. Thornton testified, "the fewer types of information there are to compare, the less likely it is to find potential matches." (Def. Ex. 511 at 11.) At least nine of Dr. Stewart's sweeps used SSN4 as a data field. (Pl. Ex. 242 at 29 (tbl. 3).)

Dr. Thornton credibly identified several problems affecting the reliability of Dr. Stewart's methodology. This court will not set forth every criticism herein, but here are the more significant ones. Most relevant, Dr. Thornton questioned whether Dr. Stewart had done sufficient manual review of his results to ensure accuracy. (Doc. 416 at 64, 86.) Dr. Thornton has done database matching of DMV records for private sector clients in the past and asserted that she and her staff spend hundreds of hours on manual review. (Id. at 64.) Dr. Stewart did not perform a manual review of his December 2015 no-match report, but he did perform an "informal review" of his initial no-match list. (Doc. 408 at 103, 105.) Of the 397,971 individuals on the no-match list, Dr. Stewart extracted fifty for manual review. (Id. at 104.) As part of this review, he looked for common typographical errors and mistakes. (Id. at 103–04.) He identified 10–15% false negatives—where the computer did not match an individual but the voter nevertheless has an acceptable ID in the databases. (Id. at 27, 104.) Dr. Stewart considered this to be an acceptable degree of error. (Id. at 104.) By contrast, in looking for false positives—the computer makes a match even though a voter did not have an acceptable ID in the databases—Dr. Stewart and his research assistant manually reviewed 100 individuals for each sweep. (Id. at 137, 153; Pl. Ex. 254 at 15–17.) As a result, Dr. Stewart manually reviewed 3,600 individuals for false positives, but only fifty individuals for false negatives. (Id. at 153–54.)

Dr. Thornton also criticized the fact that Dr. Stewart did not have actual access to the federal databases he used for matching. (Doc. 416 at 97.) Rather, he was forced to give instructions to the federal agencies and let them run the sweeps. (Doc. 408 at 21.) The federal databases included information on the possession of passports and veteran IDs. (Id.) The problem, according to Dr. Thornton, is that database matching is an iterative process in which you identify criteria for sweeps by working with the database. (Doc. 416 at 96–97.) Without access to the database, she claims, it is difficult to determine what number and type of sweeps should be conducted. (Id.)

In sum, Dr. Stewart's no-match list is itself only an estimate of how many voters lack qualifying ID in North Carolina.⁶²(See Doc. 408 at 44.) There is reason to believe that that it substantially overestimates the number of registrants who lack qualifying ID. (Id. at 117 (Dr. Stewart, stating that "it's possible that there is an overestimate" in the no-match list).) But even under Dr. Stewart's estimate, the percentage of North Carolinians who could not be matched to a qualifying ID is less than in South Carolina. In addition, the voting history of those on the no-match list and the evidence of reasonable impediment voting in South Carolina suggest that only a fraction of the small fraction of individuals who lack qualifying ID will cast a ballot under the reasonable impediment exception. This is not because these voters will be deterred by the ID requirement; it is because they did not vote at a significant rate before the requirement existed, even in high turnout elections. As noted below, this has important implications for the feasibility of administering the reasonable impediment provision, as the data suggest that poll workers are not likely to encounter an overwhelming number of such voters.

62 Plaintiffs claim Dr. Stewart's first and second no-match analyses underestimate the true number of voters who lack usable ID because they count voters with suspended licenses as matches. (Doc. 419 at 26.) North Carolina's voter-ID law does not contemplate whether suspended licenses are acceptable for voting. SeeN.C. Gen. Stat. § 163–166.13. It is not possible to tell whether a DMV ID is suspended just by looking at it. However, North Carolina law prohibits possessionor displayof any "driver's license, learner's permit, or special identification card" that the individual knowsto be canceled, revoked, or suspended. Id.§ 20–30(1). In Dr. Stewart's initial no-match analysis, he accounted for suspended licenses. (Pl. Ex. 256 at 2 (revised tbl. 11).) When Dr. Stewart counted suspended licenses as qualifying ID, he was not able to match 397,971 individuals. (Id.) When he did not count suspended licenses as qualifying ID, he was not able to match 653,995 individuals. (Id.) Dr. Stewart has not provided an analysis of suspended licenses in his most recent no-match report, which he calls his "best estimate." (SeePl. Ex. 891; Pl. Ex. 408 at 52.) Accordingly, while Dr. Stewart's prior analysis suggests the number of individuals with a suspended license may be sizable, this court has no current estimate of that number. This court cannot simply assume that Dr. Stewart's prior analysis is correct because, in addition to updating his matching to account for expired licenses under SL 2015–103, Dr. Stewart substantially "refined" his matching criteria between his first and second report. (SeeDoc. 408 at 41–43; Pl. Ex. 891 at 19 (tbl. 11).) Dr. Stewart did not include this refined matching criteria in his initial no-match analysis, but it has made a substantial difference in his most recent analysis. (Pl. Ex. 891 at 19 (tbl. 11).) For example, without the refined criteria (which came in part from Defendants), 272,700 individuals would have been on Dr. Stewart's no-match list. (Id.) With the refined analysis, which Dr. Stewart "agree[s]" was necessary, (Doc. 408 at 42), 224,863 individuals are on the most recent no-match list (47,837 fewer), (Pl. Ex. 891 at 19 (tbl. 11)). Moreover, South Carolina also prohibits the possession or display of "any canceled, revoked, suspended, or fraudulently altered driver's license or personal identification card." S.C. Ann. Code § 56–1–510. Yet there is no indication that the court considered suspended licenses as non-qualifying ID in its analysis. SeeSouth Carolina, 898 F.Supp.2d at 40. An apples-to-apples comparison to South Carolina would not exclude suspended licenses from consideration. In any case, those with suspended licenses are eligible for a no-fee voter ID, and the fact that they previously had a DMV-issued ID strongly suggests they have the materials and resources necessary to acquire one. (SeeDoc. 419 at 49 n.18.) Additionally, it is not even necessary for these individuals to actually acquire a no-fee voter ID. If an individual follows the law and surrenders his license (so he no longer "possesses" it), then he can vote under the reasonable impediment exception so long as he has an impediment to acquiring the voter ID (lack of qualifying documents, etc.). An individual who lost his suspended license would also be eligible to vote under the reasonable impediment exception. SeeN.C. Gen. Stat. § 163–166.15(e)(1) f.

e. Availability of the Reasonable Impediment Exception

Voters who do not have a qualifying ID retain the opportunity to vote through the reasonable impediment exception. Those so voting must complete the following process.

When voters present to vote, the first election worker they will encounter is the greeter. The greeter's job is to "[p]rovide preliminary guidance to voters on voting procedures, [p]rovide information on acceptable photo ID, [a]sk voters to remove their ID from their wallets or purses prior to presenting to the check-in station ... [a]nd provide confidence and reassurance that all voters will be given the opportunity to cast a ballot."⁶³ (Def. Ex. 551 at 9.)

63 If the county wishes, the greeter can also "complete help referral [forms] and route voter[s] to [the] Help Station." (Def. Ex. 551 at 9.)

Voters will next move to the check-in station. The first question election workers will ask the voter at the check-in station is whether the voter has acceptable photo ID. (Doc. 412–2 at 7.) If the voter does not, the election worker completes a "Help Referral Form" and refers the voter to the Help Station. (Id. at 8.) The Help Referral form contains the voter's name, registration number, address, and reason for referral. (Id.) This permits the person at the Help Desk to understand why the voter is being referred to them. (Id.) The Help Station existed long before the reasonable impediment exception was created, and it serves to assist any voter who has an issue that may prevent him from casting a regular ballot. All provisional voters, including OOP voters, are referred to the Help Station. (See id.)

Once at the Help Station, the voter lacking photo ID must be informed of all alternative voting options available. (Doc. 412–3 at 12.) Voters who have a qualifying photo ID but forgot it or those who do not have ID but wish to acquire it prior to the canvass can cast a provisional ballot that will be counted so long as the voters present a qualifying ID at their CBOE by noon on the day before the county canvass. (Id.; Doc. 412–4 at 2.) Voters who choose this option do not complete a reasonable impediment declaration. Voters who subjectively believe a reasonable impediment prevented them from acquiring ID are entitled to vote under the reasonable impediment exception. (Doc. 424–4 at 2.) Both of these options are available throughout early voting and on Election Day. (Doc. 412–3 at 14.) Finally, voters can request an absentee ballot at the early-voting site up until the deadline for doing so (a week before Election Day). (Id. at 13–14.)

Voters who elect to vote under the reasonable impediment exception must complete a two-step process at the Help Station. First, they must complete a provisional voting application ("PVA"). (Def. Ex. 546 at 3.) This form is not unique to reasonable impediment voters and must be completed by all voters casting a provisional ballot. (Id.(Ex. 1).) The top of the PVA is labeled "Voter Registration/Update Form" and in substance asks for the information the voter would have provided when they registered. (Id.) At polling places with electronic poll books (i.e., all early-voting sites), this part of the form will be pre-populated automatically by the electronic equipment and will not need to be completed by the voter. (Id. at 3.) The middle part of the PVA is labeled "Voter's Affirmation of Eligibility to Vote." (Id.(Ex. 1).) This section contains in substance the same attestation to vote that every voter casting a ballot, including those casting regular ballots, must complete. (Id. at 3; see Pl. Ex. 1056.) Reasonable impediment voters, like all other voters, will need to sign the attestation that they are otherwise authorized to vote. (Def. Ex. 546 at 3.) The bottom portion of the PVA asks voters to indicate the reason they are voting provisionally. (Id.(Ex. 1).) For those without photo ID, this part of the PVA will pre-populate "to indicate the provisional voting reason is 'No Acceptable ID.' " (Id. at 3.) In addition to checking an acknowledgment that they were provided alternative voting options, (Doc. 412–4 at 9), the voter will "only [need] to sign the application," (Def. Ex. 546 at 3). The election official will need to sign the form at the bottom of this box. (Id.(Ex. 1).) Voters can receive assistance from election officials at the Help Station in completing the PVA. (Id. at 3.)

Second, reasonable impediment voters must complete a reasonable impediment declaration ("RID"). The RID comes in two forms: the "pre-printed" version and the "SEIMS-generated" version. (Id. at 3.) As the names would suggest, the SEIMS-generated form is printed at the Help Desk after certain information from the SBOE's registration database ("SEIMS") is pre-populated into the form, whereas the "pre-printed" version is printed before the voter presents.

With regard to the pre-printed version, the top of the form is to be completed by the election official and includes information such as "[l]ocation voted." (Id.(Ex. 2).) Moving down the form, the next box is to be completed by the voter and asks for the voter's name, email address, phone number, date of birth, and SSN4. (Id.) The next box is labeled "Voter's Declaration of Reasonable Impediment." (Id.) Voters must declare that they "suffer from a reasonable impediment that prevents [them] from obtaining acceptable photo identification." (Id.) The form then asks the voter to list the impediment(s) he suffers. (Id.) The form contains template boxes for the following impediments: "Lack of transportation"; "Lack of birth certificate or other documents needed to obtain photo ID"; "Work schedule"; "Lost or stolen photo ID"; "Disability or illness"; "Family responsibilities"; "Photo ID applied for but not received"; and "State or federal law prohibits ... listing [the] impediment." (Id.) If any one of these applies, the voter need only check the appropriate box, although a voter may check all that apply. (Id.("My reasonable impediment is due to the following reason(s).").) No further explanation is required. (Id.) If none of these reasons applies or the voter wishes to be more specific, there is also a box for "other reasonable impediment" followed by a line where the voter can explain the impediment. (Id.)

Below the statement of impediment, the voter is asked to check one of three options indicating which alternative identification document or information he is providing. (Id.) The first box is to be checked if the voter has provided his SSN4 and date of birth. (Id.) If the voter already provided this information in the top of the RID, he need only check the box. (Id.) The second box is to be checked if the voter is presenting a HAVA document showing his name and address. (Id.) The HAVA documents that qualify are listed, and the voter need only check which applies. (Id.) They include "a current utility bill; bank statement; government check; paycheck; or other government document." (Id.) The "other government document" option provides a line where the voter can write in the applicable document. (Id.) The third box is to be checked if the voter has provided his voter registration card. (Id.) Finally, if the voter fails to "provide any alternative identification document or information," there is a box for the election official to check. (Id.) The final portion of the pre-printed RID requires the voter to attest that it has not been completed "fraudulently or falsely." (Id.)

The SEIMS-generated version of the RID is substantively the same as the pre-printed form, but differs in the following ways. (Id.(Ex. 3).) First, instead of only asking for the voter's name, email address, phone number, date of birth, and SSN4, the form contains the "Voter Registration/Update Form" box that appeared on the PVA. (Id.) This form contains more voter information than is required on the pre-printed form, but the information is pre-populated by the computer so the voter does not need to complete it. (Id.(Exs. 2, 3).) Second, the portion addressing the statement of reasonable impediment (where the voter checks which impediment applies) and proof of identity (where an alternative identification document or information is provided) contains smaller typeface. (Id.(Ex. 3).) As with the pre-printed version, the voter must attest that the form has not been "fraudulently or falsely" completed. (Id.) In sum, the SEIMS form offers the benefit of pre-population, while the pre-printed form offers the benefit of larger print. (Id.(Exs. 2, 3).) To compensate for the absence of pre-population, the pre-printed form requires less voter information than the SEIMS-generated form. Voters can receive assistance from election officials at the Help Station in completing either version of the RID. (Id. at 3.)

The final document the voter will receive from the poll worker is the provisional voter instructions. This form, which is provided to all provisional voters, gives the voter the information necessary to determine whether his vote was counted. (Doc. 412–4 at 13.)

Plaintiffs concede, as they must, that the reasonable impediment exception ameliorates the burden of the photo-ID requirement for some voters. Plaintiffs nevertheless claim that the reasonable impediment process places a discriminatory burden upon African Americans and Hispanics. They claim that members of these groups are (1) more likely to lack qualifying photo ID (and thus need the reasonable impediment exception) and (2) more likely to struggle in completing the RID.

African Americans are more likely to lack qualifying ID and thus elect to use the reasonable impediment exception. As noted above, Dr. Stewart's no-match results do not establish how many North Carolinians lack qualifying photo ID. However, regardless of the actual number, it is more likely than not that racial disparities exist in the population that lacks qualifying photo ID. In each of the SBOE's four no-match analyses, African Americans were less likely to be matched to a qualifying ID. (Pl. Ex. 891 at 4, 6.) In addition, African Americans have disproportionately been on the no-match list in each of Dr. Stewart's no-match analyses. (Id. at 13 (tbl. 7), 19 (tbl. 11).) In fact, the evidence shows that racial disparities grow as the no-match list becomes smaller. (Id.) Similar disparities have been found in Georgia and South Carolina. (Id. at 8–9.) Further corroborating the results in North Carolina, Dr. Stewart has presented studies showing racial disparities in ID possession nationwide. (Id. at 9.) Dr. Stewart claims that he has yet to find a combination of acceptable IDs that will eliminate the disparities in photo-ID possession. (Doc. 408 at 159–60.) Accordingly, this court finds that, whatever the true number of individuals without qualifying IDs, African Americans are more likely to be among this group than whites.

The second part of Plaintiffs' argument is less clear and turns on whether the reasonable impediment exception sufficiently ameliorates any alleged burden arising from disparities in photo ID possession. Plaintiffs make several arguments for why this is not the case.

Plaintiffs first express concern over the fact that reasonable impediment declarants will be provided provisional ballots. As Plaintiffs indicate, although HAVA requires provisional ballots to be given to voters in certain circumstances, it only requires those ballots to be counted "in accordance with State law." 52 U.S.C. § 21082(a)(4). But the problem with Plaintiffs' argument is two-fold. First, it is in conflict with Plaintiffs' position at trial in July 2015, where they advocated for OOP provisional ballots on the grounds that they ameliorate burdens. (Doc. 346 at 90–94); seeSouth Carolina v. United States, 898 F.Supp.2d 30, 42 (D.D.C.2012) ("[T]he Supreme Court characterized provisional ballots as curing problems and alleviating burdens, not as creating problems and imposing burdens."). Second, with regard to reasonable impediment declarants, North Carolina law provides for counting their ballots. Despite the provisional label, North Carolina law provides that ballots cast under the reasonable impediment exception must be counted so long as (1) an acceptable alternate form of identification can be verified (SSN4 and date of birth, etc.) and (2) the stated reason is not factually false, merely denigrating to the ID requirement, or obviously nonsensical. N.C. Gen. Stat. § 163–182.1B(a).

Plaintiffs next claim the State's educational efforts have not been sufficient to make voters, especially minority voters, aware of its availability. This court addressed the sufficiency of the State's educational efforts in its denial of Plaintiffs' motion to preliminarily enjoin implementation of the ID requirement in the March 2016 primary. (Doc. 383.) Defendants have updated those efforts, as noted above, and the State's educational efforts have continued and increased since the preliminary injunction decision was issued.

Plaintiffs' primary witness on the sufficiency of the State's educational efforts was Barry Burden, Ph.D., Professor of Political Science at the University of Wisconsin-Madison.⁶⁴ Dr. Burden opined that the public is not sufficiently informed about the reasonable impediment exception. (Pl. Ex. 889 at 4.) Dr. Burden has done no study of North Carolinians' knowledge of the exception. (Doc. 407 at 82.) Instead, he relies upon surveys from other States, including Wisconsin and Pennsylvania, none of which has a reasonable impediment exception or experienced North Carolina's education effort. (Id.) These studies are plainly unhelpful. Dr. Burden also cites the nationwide Survey on the Performance of American Elections, which in 2014 found that 8% of whites and 14% of African Americans and Hispanics cited not having "the right kind of ID" as a factor in why they did not vote. (Pl. Ex. 889 at 6.) But only one State in that survey—South Carolina—had a reasonable impediment exception. (Doc. 407 at 83–84.) If such data exist, Dr. Burden has not broken them down for South Carolina. (Id.; Pl. Ex. 889 at 6.) This court thus lacks any data on voter knowledge of the reasonable impediment exception in South Carolina. (Doc. 407 at 83–84.) Further, Dr. Burden did not perform any study comparing North Carolina's education and training efforts to the education and training efforts in the States studied. (Id. at 81.) In short, there is no reliable evidence on which to credit Dr. Burden's opinion that North Carolinians are unaware of the reasonable impediment exception. What is apparent, as described more fully above, is that North Carolina's voter outreach and education efforts pertaining to the voter-ID requirement and its reasonable impediment exception have been substantial.

64 Dr. Burden was proffered without objection as an expert in the analysis of election laws and administration and their effect on voter behavior. (Doc. 331 at 69.)

Plaintiffs next argue that the reasonable impediment process requires a high degree of literacy and is intimidating. Dr. Burden concluded that, even as amended, "the photo ID provision remains burdensome on voters in North Carolina, more so for African Americans and Latinos than for whites." (Doc. 407 at 42–43.) To reach this conclusion, Dr. Burden relied on what he calls the calculus of voting.⁶⁵ (Id. at 44.) This framework centers on the idea that there are costs and benefits to voting. (Id.) The State controls some costs, such as the ease of using certain voting mechanisms. (Id. at 45.) In Dr. Burden's view, however, the State does not control any of the benefits of voting. (Id.) Because Dr. Burden does not consider the benefits of voting, he consequently fails to meaningfully engage the voter's propensity to vote. Although it speaks of costs and benefits, Dr. Burden's analysis is in effect reduced to the following two questions: (1) did the State increase the cost of voting and, if so, (2) was the increased cost justified? (See id. at 53.) For example, Dr. Burden cites a State's imposition of a registration requirement—found in virtually all States—as a burden on voters, but sustains it because of its "compelling State interest." (Id.("Mandating voter registration at one point became sort of obvious reform for many states. That was a new restriction, but it had a strong motivation behind it.").)

65 Dr. Burden also relied on a second framework centered on the role that habit plays in voting behavior. (Doc. 407 at 48.) In Dr. Burden's view, North Carolina voters were habituated to not providing ID. (Id.)

Dr. Burden's conclusion that the reasonable impediment exception does not sufficiently ameliorate the alleged burden from SL 2013–381 is in part based on his belief that the paperwork and process for the exception will deter low literacy individuals, who are disproportionately African American and Latino. (Pl. Ex. 889 at 3, 6.) This opinion is without support. Dr. Burden is not an expert in literacy. Accordingly, he has not offered a review of the applicable forms or an opinion as to what level of literacy would be required to complete the reasonable impediment process. In sum, Dr. Burden's testimony was of limited practical assistance to the court, as it was heavy on theory and light on facts.

Plaintiffs also offered the testimony of several fact witnesses who work with low literacy individuals.

Ashley Lasher is the Executive Director of the Literacy Counsel of Buncombe County, North Carolina. (Pl. Ex. 1050 at 8.) The Literacy Counsel's mission is to "increase comprehensive literacy and English language skills through specialized instruction by trained tutors and access to literacy resources." (Id. at 11.) The group offers two adult literacy programs to adults in western North Carolina: an English-as-a-second-language course serving about 250 students per year, 85–90% of whom are Spanish-speaking; and an adult education class for low literacy individuals whose first language is English, which serves about fifty students per year split equally between African Americans and whites. (Id. at 44–46.) Ms. Lasher is not a literacy expert, nor does she work directly with students. (Id. at 11, 14.) Instead, her role is to manage the organization, provide oversight, and participate in fundraising efforts. (Id. at 11.) Plaintiffs sought to have her opine on the ability of the group's students to navigate a draft of the SEIMS-generated version of the RID. (Id. at 62–65.) Plaintiffs did not tender or qualify Ms. Lasher as an expert, nor did they lay adequate foundation for Ms. Lasher to provide lay opinion testimony based upon her personal knowledge and perception. Fed. R. Evid. 701. In fact, her testimony made clear that her opinions were not based on her personal knowledge and perceptions, but on the personal knowledge and perceptions of her organization's program directors, who were not available for cross examination.⁶⁶ Consequently, Ms. Lasher's opinions are not admissible, but, even if they were, this court would not find them helpful.

66 Ms. Lasher did not review the form with any of the organization's students, nor does she regularly review forms for level of difficulty. (Pl. Ex. 1050 at 65, 69.) Instead, she reviewed it with her organization's program directors, who conduct intake assessments, and sought to give her opinion based on that. (Id.at 69, 76–77.) Based on her feedback from the directors, she claimed that the reasonable impediment form requires a higher level of vocabulary, the small font would make it more difficult for lower literacy readers, and low literacy individuals might be fearful in completing the form in light of the fact that it's a felony to "fraudulently or falsely" complete it. (Id.at 62–65.) Ms. Lasher was not aware that reasonable impediment declarants can receive assistance in completing the necessary forms from either a poll worker or a person of their choice. (Id.at 38.) In any event, she testified that many of the organization's low literacy students have significant coping skills that frequently consist of acquiring the assistance of trusted family members or friends in completing documents. (Id.) She attributed the fact that many of the organization's students are registered to vote to this coping ability. (Id.)

Michelle Kennedy is the Executive Director of the Interactive Resource Center ("IRC") in Greensboro, North Carolina. (Doc. 409 at 82.) The IRC provides a series of services to the homeless or likely-to-be-homeless population. (Id. at 82–83.) One such service involves assisting the homeless in document recovery and acquiring ID. (Id. at 83–84.) Ms. Kenney testified that ID is critical to the ability for homeless individuals to transition back to self-sufficiency. (Id. at 85.) Unfortunately, due to the nature of their living circumstances, homeless individuals are more likely to lose their ID once it is acquired. (Id.) Ms. Kennedy estimates that 90% of the individuals the IRC serves are African American. (Id. at 84.) With regard to the RID, Ms. Kennedy was concerned about the form's request for residential address and its warning that fraudulently or falsely completing it is a felony. (Id. at 100–01.) Even though the IRC's "guests" use the IRC's address in applying for ID and in registering to vote, (id. at 109–11), she was concerned that a "vett[ing]" of that address "would show ... that [it's] a commercial address and, therefore, not a residence," (id. at 100). Ms. Kennedy was also concerned that many of the IRC's guests would not know their SSN4. (Id. at 99.)

Maria Unger Palmer is a plaintiff in this case and has extensive experience in get out the vote ("GOTV") efforts, including efforts targeting Hispanics. (Doc. 410 at 7–8.) In her future outreach efforts, Ms. Palmer does not plan to use the reasonable impediment exception because she believes it is intimidating and "requires a high level of literacy." (Id. at 12.) She is not a literacy expert but "was a schoolteacher and a school principal and trained [individuals] in testing." (Id. at 25.) Although Ms. Palmer has volunteered as a translator at the polls in the past, she says State-provided interpreters are not available. (Id. at 14.) Without an interpreter, she believes, many low literacy Hispanic voters will not have the literacy skills to complete the reasonable impediment process. (Id. at 14–15.)

Examination of the reasonable impediment voting process and the process of other voting mechanisms reveals that the concerns of these fact witnesses are not well-founded. Every North Carolina county uses electronic poll books during early voting. (Id. at 90; Doc. 414 at 123.) Many, but not all, also use electronic poll books on Election Day. (Doc. 410 at 90.) Where electronic poll books are used, the PVA (step 1) and the SEIMS-generated version of the RID (step 2) will pre-populate with the voter's registration information, including residential address. (See Doc. 546 at 3.) This means that if an IRC guest were to use the IRC's address when registering to vote, that address will pre-populate into the residential address field on both forms. (See id.) Those who present at one of the voting locations without electronic poll books can complete the pre-printed version discussed above. (See id.(Ex. 2).) This version features larger print and requires less information from the voter. (See id.) Further, in completing either form, voters can receive assistance from any person of their choosing,⁶⁷ except "their employer or their union representative." (Doc. 414 at 138–39.) If the voter does not have anyone to assist him, he can seek assistance from poll workers in completing both forms. (Id. at 189.) Poll workers are instructed to assist voters in completing the provisional process without inquiring into whether the voter is illiterate. (Id.) For example, if the voter tells the poll worker that she does not understand the form, "the precinct official is supposed to do everything they can to try to provide as much explanation to [the voter] as possible until they do understand it." (Id. at 211.) This type of assistance predates North Carolina's voter-ID law. For example, in implementing SDR, poll workers were trained to inspect the voter registration form to ensure it was properly completed. (Id.) This is in part why Plaintiffs claim it is a valuable fail-safe. Director Strach testified that poll workers will provide the same review function for reasonable impediment voters. (Id. at 212 ("The person at the help station is to ensure that [the reasonable impediment declaration] is complete before ... they provide [the voter] the ballot in order to vote the provisional ballot.").) If an administrative defect in the declaration, such as a failure to list an impediment or provide other necessary information, nevertheless remains, the CBOE has the ability to reach out to the voter to acquire the missing information. (Id. at 212–13.) Accordingly, the reasonable impediment voting process is designed to permit and provide significant assistance.

67 Plaintiffs contend that N.C. Gen. Stat. § 163–166.8, which is entitled "Assistance to voters," prohibits the type of assistance for reasonable impediment voters that Director Strach says will be provided. (SeeDoc. 414 at 185–89.) This is incorrect. That statute applies to persons qualified to vote who need "assistance with entering and exiting the voting booth and in preparing ballots" and provides that a voter "who, on account of illiteracy, is unable to mark a ballot without assistance" "is entitled to assistance from a person of the voter's choice, other than the voter's employer or agent of that employer or an officer or agent of the voter's union." N.C. Gen. Stat. § 163–166.8(a)(2). Most importantly, Director Strach testified that if a voter asks an election official for help filling out a form, officials are trained to assume it is on account of literacy without requiring proof of illiteracy. (Doc. 414 at 189.)

Further, Plaintiffs have failed to demonstrate that the reasonable impediment voting process is more difficult than other voting mechanisms that Plaintiffs either advocate for or have not challenged.

First, the PVA stage of the reasonable impediment process (step 1) must be completed by all provisional voters, including OOP voters. All such voters must report to the Help Station and complete the PVA, which requires the voter to provide the same voter registration-related information as required by the RID and attest that it is not provided "fraudulently or falsely." (Def. Ex. 546 (Ex. 1).) As with most voter-related forms, the PVA contains phrases, such as "attest," "provisionally," "affirmation," and "fraudulently." (Id.) Residential address is also a required field. (Id.) Nevertheless, as noted below, a disproportionate share of African Americans and Hispanics cast OOP ballots, and thus necessarily completed the provisional voting application, when OOP was in place.

Second, all voters are required to complete a voter registration form. (See Pl. Ex. 212A.) Those wishing to use SDR were required to complete a voter registration form at the polling place. Residential address is a required field on the voter registration application. (Id.) In bold, red print next to the signature line, the registrant is warned that "fraudulently or falsely completing" the registration application is a felony. (Id.) Nevertheless, as noted above, African American registration rates exceed those of whites in North Carolina. (Pl. Ex. 684.)

Third, every voter is required to complete an ATV (authorization to vote) form. (Doc. 410 at 91–93.) This was true both before and after the voter-ID law. (Id.) Accordingly, as Plaintiffs' counsel pointed out, if over 4.3 million North Carolinians voted in the 2008 Presidential election, then over 4.3 million North Carolinians completed an ATV form. (Id. at 93.) To complete the ATV form, the voter must attest that the address he provided is correct and that he has not voted in the election. (Pl. Ex. 1056.) Here, too, the voter is warned that "fraudulently or falsely completing" the ATV is a felony. (Id.) Thus, to the extent that the IRC's guests are concerned about attesting that the IRC's address is their residential address, this concern predates the voter-ID law and will remain regardless of the method of voting employed. (Id.) In addition, as with many voting related forms, there are phrases such as "fraudulently," "hereby certify," and "violation of NC law." (Id.)

The fact that so many minority voters have successfully navigated these forms over the years strongly suggests that their experience with the RID will not be different. The SEIMS-generated version may have slightly smaller print, (Def. Ex. 546 (Ex. 3)), but the same is true of the forms discussed above. The pre-printed version appears to have larger print than any of these forms. (Id.(Ex. 1).) The RID only requires two pieces of information from the voter that are not redundant of other forms. First, the voter must provide alternative identification. (Id.(Ex. 3).) Even assuming Ms. Kennedy is correct that many of the IRC's guests do not know their SSN4, (Doc. 409 at 99), the SSN4 is just one form of acceptable alternative identification, (Def. Ex. 546 (Ex. 3)). Voters also can provide an acceptable HAVA document or their voter registration card. (Id.) All registered voters receive a voter registration card, and all SDR voters had to present a HAVA document in order to vote. The second piece of non-redundant information is the section where the voter states his impediment to acquiring acceptable ID. (Id.) But even this part of the form is designed for ease of use. Rather than require the voter to write in his own impediment, the form contains a non-exhaustive list of eight qualifying impediments, and the voter need only check one or more that apply. As noted above, the voter can receive whatever assistance is necessary to make this determination.

Plaintiffs next argue that the reasonable impediment challenge process is likely to be implemented in a discriminatory and intimidating fashion. As noted above, a provisional ballot cast under the reasonable impediment exception must be counted and can only be rejected on the basis of the impediment provided if the listed impediment is "factually false, merely denigrate[s]" the ID requirement, or is "obviously nonsensical." N.C. Gen. Stat. § 163–182.1B(a)(1). A voter's reasonable impediment declaration can become subject to scrutiny through either a voter's evidentiary challenge or the CBOE's review of provisional ballots. (See Def. Ex. 547.)

CBOEs are required to "make redacted copies of all Reasonable Impediment Declaration forms available to the public upon request," (id. at 1), and any voter registered in the same county as the reasonable impediment voter may make an "evidentiary challenge" to the reasonable impediment declaration, N.C. Gen. Stat. § 163–182.1B(b)(1). There are significant procedural limitations on the reasonable impediment challenger. Challenges may only be made on the SBOE's Evidentiary Challenge form, (Def. Ex. 547 at 7), and must be "submitted no later than 5:00 P.M. on the third business day following the election" , N.C. Gen. Stat. § 163–182.1B(b)(2). In addition, the scope of the challenge is "strictly limited" to the facts the challenger alleges in the written challenge form. (Def. Ex. 547 at 7.) To deter unwarranted or improper challenges, the challenge form clearly provides that "fraudulently or falsely" completing it is a felony. (Id.)

Once a challenge has been made to the factual veracity of the reasonable impediment, the CBOE office is instructed to closely inspect the evidentiary challenge form to "ensure the form has been completed fully, including a signature and contact information for the challenger." (Id. at 2.) To be complete, the challenge form must be notarized. (Id. at 7.) If the CBOE determines that the challenge form is complete and timely presented, the CBOE must send written notice of the challenge to the voter and the challenger by mail. (Id. at 3.) At a minimum, the notice must contain the following:

Name and address for the voter and the challenger; A statement indicating that an evidentiary challenge has been entered ... disputing the factual truthfulness of the reasonable impediment claimed by the voter; A statement that the county board of elections will hold a hearing at [date, time, and location] during which it will decide whether the challenger has shown by clear and convincing evidence that the claimed impediment merely denigrates the photo identification requirement, is obviously nonsensical, or is factually false; A statement that the voter may appear in person or through an authorized representative to present evidence supporting the factual veracity of the impediment; [and] Copies of the Reasonable Impediment Declaration form (redacted) and the completed Evidentiary Challenge Form.

(Id.) CBOEs are directed to provide the "maximum notice possible to the voter," and in addition to written notice "should make every effort to contact the voter via phone, email, and any other available means of contact." (Id.) The CBOE office is also directed to notify CBOE board members, the county attorney, and the SBOE of the challenge. (Id.) All Evidentiary Challenge forms must be forwarded to the SBOE. (Id.) The SBOE plans to use its legal team to provide oversight of challenges and ensure CBOEs are following proper procedure. (Doc. 414 at 215–16.)

On the day of the canvass, the CBOE "is required to conduct a hearing on the challenge in an open meeting and render a determination on the provisional ballot." (Def. Ex. 547 at 3.) The county attorney's role at the evidentiary hearing is to provide CBOE board members with guidance on the standard of review. (Id. at 4.) The standard of review to be applied by the CBOE is "[w]hether, having considered all facts in the light most favorable to the voter, the challenger has shown by clear and convincing evidence that the stated impediment (1) merely denigrates the photo identification requirement, (2) is obviously nonsensical, or (3) is factually false." (Id.(emphasis added).) CBOEs have been trained that "[l]ight most favorable to the voter" means "[i]f you can view a fact in a way that helps the voter, you must view it that way." (Def. Ex. 551 at 55.) They have also been trained that "[c]lear and convincing evidence" "is greater than 'more likely than not' " and means "[e]vidence which should fully convince you." (Id.) "The challenger bears the burden of proof and persuasion at the hearing." (Def. Ex. 547 at 4.)

At the challenge hearing, the CBOE is to first provide the challenger the opportunity to speak and present evidence. (Id. at 5.) The challenger's presentation is to be limited to "substantiating facts already alleged in the Evidentiary Challenge form." (Id.) CBOEs are reminded "that the statute sets an intentionally high bar for a challenger." (Id.) If the voter is present, the board must next provide the voter an opportunity to speak and present evidence. (Id.) A voter's absence cannot be held against the voter. (See id.("[A]ll evidence must be construed in the light most favorable to the voter, even if the voter is not present at the hearing.").) In addition, the CBOE is directed to "[k]eep in mind that a voter who has claimed a reasonable impediment may face material constraints different from those experienced by members of the [CBOE]." (Id.) The CBOE is not permitted to "second-guess the voter's priorities or scheduling constraints." (Id. at 2, 5.)

After giving the parties an opportunity to speak and present evidence, the CBOE must deliberate in open session. (Id. at 5.) Each CBOE is comprised of three members, no more than two of which can be from the same party as the governor. N.C. Gen. Stat. § 163–30. A CBOE cannot "find a challenge valid if it provides only evidence regarding the reasonableness of the impediment." N.C. Gen. Stat. § 163–182.1B(b)(6). For example, CBOEs have been trained that if a voter checks the box for "photo ID applied for but not received," a challenge could not be sustained on the ground that the voter " 'waited until the last minute' to apply for a photo ID." (Def. Ex. 547 at 2.) If the SBOE has reason to believe that a CBOE has rejected a RID on the basis of the reasonableness of the impediment provided, the SBOE intends to use its supervisory powers to correct this problem. (Doc. 414 at 216.) In sum, the CBOE must reject the challenge unless two of its three members find that, even viewing the evidence in the light most favorable to the voter, the challenger has carried his burden of showing by clear and convincing evidence that the stated impediment is either factually false, merely denigrates the photo-ID requirement, or is obviously nonsensical. (Def. Ex. 547 at 5.)

The second way a RID may become subject to scrutiny is through the CBOE's review of provisional ballots. (Id. at 6.) The CBOE's ability to reject a RID on the basis of the impediment provided is very limited. First, the CBOE "may not question the factual veracity of a claimed impediment" without completing the formal hearing process described above. (Id.) Second, if the voter checked one of the "template impediments," such as "lack of transportation" or "work schedule," the CBOE cannot reject the impediment on the basis that it merely denigrates the ID requirement or is obviously nonsensical. (Id.) This is because the SBOE considers the template impediments to be non-denigrating and not nonsensical as a matter of law. (Id.) Accordingly, the only way the SBOE claims a CBOE can reject a reasonable impediment declaration without a formal hearing is if the voter has checked the "other reasonable impediment" box and provided a written description that the CBOE has grounds to believe merely denigrates the ID requirement or is obviously nonsensical. (Id.) All facts must still be viewed in the light most favorable to the voter,⁶⁸ and the CBOE cannot "reject a provisional ballot if there is any possible question of fact." (Id.) The SBOE provided the following example during CBOE training:

68 Director Strach's memorandum also instructs CBOEs that the voter's due process rights must be preserved by providing the voter with "meaningful notice and an opportunity to be heard before an unbiased board of elections." (Def. Ex. 547 at 6.)

[A] voter who writes "baseball player" on the Reasonable Impediment Declaration form could be attempting to more specifically indicate the voter's profession, which has impeded the voter from

obtaining acceptable photo ID. Such a claim would be equivalent to the "work schedule" impediment already deemed valid as a matter of law. Accordingly, a county board could not disregard the factual question at issue and consider "baseball player" as merely denigrating the photo identification requirement or as nonsensical, and as a result throw out the ballot.

(Id.) Finally, CBOEs have been instructed that, in considering non-template impediments during the canvass, the CBOE must

bear in mind that (1) the voter has declared the impediment under penalty of a Class I felony, (2) the voter could have easily chosen to mark one of the template impediments, and (3) election officials were able to review alternative identification documents or validate the voter's social security number and date of birth.

(Id.) Even if a voter's listed impediment is ultimately found to be factually false, merely denigrating, or obviously nonsensical, the SBOE will refer the voter for prosecution only where its investigative team finds that "there was intent to commit a violation." (Doc. 414 at 137.)

The statute governing challenges to RIDs does not provide an appeal process for reviewing a CBOE's rejection of a declaration. SeeN.C. Gen. Stat. § 163–182.1B ; (Doc. 410 at 134). However, the SBOE has supervisory authority over all elections in the State and has "statutory authority to take any petition or complaint of any alleged misconduct of a county board of elections or their failure to carry out their duties in administering the law." (Doc. 414 at 138); N.C. Gen. Stat. § 163–22(a), (c). Pursuant to this authority, Director Strach asserts that a voter who believes her RID has been erroneously rejected could file a petition or complaint that the SBOE could review. (Id. at 216–17); N.C. Gen. Stat. § 163–22(c) ("[T]he [SBOE] shall have the right to hear and act on complaints arising by petition or otherwise, on the failure or neglect of a [CBOE] to comply with any part of the election laws imposing duties upon such a board. The [SBOE] shall have power to remove from office any member of a [CBOE] for incompetency, neglect or failure to perform duties, fraud, or for any other satisfactory cause."). If the CBOE's failure to follow proper procedure affected the outcome of an election in the county, then the complaint could be made in the form of an election protest. (Doc. 414 at 217.) Election protests must be resolved before votes are canvassed and the results certified. (Id.)

At trial, Director Strach answered hypothetical questions, including some extensive questioning from the court, on whether certain challenges would contest factual falsity or the reasonableness of an impediment. She was asked "[i]f someone says, I have [a] lack of transportation, and the challenger says they have access to a car one day a week, can that be demonstrated to be factually false, or is that a question of reasonableness?" (Doc. 414 at 207–08.) She replied that she "think[s] that's a question of reasonableness." (Id. at 208.) She was then asked the same question, but the challenger presented evidence that the voter has access to a bicycle. (Id.) Again she replied that she believes that would only go to reasonableness. (Id.) In her view, "[i]f [the challenger] is only able to provide that [the voter] has access to transportation ... that would go to the reasonableness of it, and that would not be able to be deemed not factual." (Id. at 207.)

With regards to "disability or illness," Director Strach was asked "if someone checks the box of disability or illness, and somebody comes forward and says they have evidence that they are not ill or not disabled," could the challenge be sustained as factually false? (Id. at 208.) For the challenge to be sustained, Director Strach testified that the challenger would have to "prove the absence of disability or illness." (Id. at 209.) If the challenger could not prove that the voter had never been disabled or never been ill, then the question would be whether it was reasonable for the voter not to have acquired a qualifying ID in light of whatever disability or illness existed. (See id. at 208–09.) Of course, impediments cannot be rejected on the ground that they are not reasonable.

Director Strach was next asked, "[i]f a voter checks the box [for] lack of birth certificate or other documents needed to obtain photo ID and the person is challenged, and the challenger comes forward with proof that the person factually has two of the documents [that are sufficient to establish identity and age at the DMV]," then can the challenge be sustained? (Id. at 209.) She replied that the challenge could only be sustained if the challenger could show by clear and convincing evidence that the voter "actually possessed" all of the necessary documents to acquire a qualifying ID. (Id. at 209–10 (emphasis added).) Of course, the voter's evidence would be considered as well. Thus, if, for example, a challenger were merely to offer evidence from a database that the voter was issued a certain type of supporting document, it would not be sufficient to sustain the challenge under Director Strach's testimony because it would not prove actual possession. (See id.)

Director Strach was next asked "[i]f somebody comes in and says they are homeless and they've had their ID stolen, can they check the box [for] 'lost or stolen ID'?" (Id. at 210.) She replied that this would be a proper use of the RID. (Id.)

Director Strach was also asked "[i]f a voter were to name any family responsibility, and as long as factually there was such a responsibility, can that then be questioned further, or does that then become a question of reasonableness?" (Id.) She replied that so long as any family responsibility existed, any challenge would go to reasonableness. (Id.)

Finally, Director Strach was asked "[w]hat happens if [a voter] check[s] more than one box and it turns out that one of the boxes is factually false but another box is not?" (Id. at 213.) She replied that "in the light most favorable to the voter ... it can still be counted if at least one of them is correct and truthful."⁶⁹ (Id.)

69 The SBOE represents that it would not refer a case for prosecution unless the voter "intentionally provided false information on a declaration." (Doc. 410 at 137.)

In light of the reasonable impediment challenge process described above and the testimony of Director Strach, the court finds that Plaintiffs have failed to show that the reasonable impediment challenge process is likely to be applied in an intimidating or discriminatory manner. The law gives every advantage to the voter and places every burden upon the challenger. A challenge cannot be made without the challenger first putting his own neck on the line and swearing before a notary that "all the facts ... alleged in connection with th[e] challenge are true and accurate to the best of [the challenger's] knowledge." (Def. Ex. 547 at 7.) Fraudulently or falsely completing the challenge form is punishable as a felony. (Id.) Based on this record and absent actual fraud, reasonable impediment challenges appear to be highly unlikely. Although the United States monitored South Carolina's implementation of its voter photo ID and reasonable impediment law, neither it nor any other Plaintiff has directed this court to a single challenge there, much less a challenge where the factual falsity provision was used to arbitrarily disenfranchise a voter. This is significant, because South Carolina has been applying effectively the same reasonable impediment exception since 2013.2. Change in the Early-Voting Schedule

Over the past two decades, early voting has grown in popularity nationally, while participation in Election Day voting has waned. Absentee mail-in voting, however, remains more popular nationally than early in-person voting. (Pl. Ex. 40 at 5–6.) Despite the national growth in popularity, sixteen States do not offer any form of early in-person voting; two of these States—Oregon and Washington—conduct elections almost entirely through the mail.⁷⁰ (Def. Ex. 270 at 21 (compiled by Plaintiffs' expert, Sean P. Trende)⁷¹ .)

70 During trial, Plaintiffs objected to Defendants' comparisons with other States, yet Plaintiffs' experts themselves repeatedly did just that in arguing the intent and effect of SL 2013–381. (E.g., Pl. Ex. 40 at 8.) Under § 2, the examination is assuredly a very local, practical appraisal. League, 769 F.3d at 243. At least one legal commentator endorses such comparisons, however, as properly within the totality of the circumstances analysis:

Given the distinctive characteristics of each state's election ecosystem, evidence of other states' laws and practices may be of limited probative value. The Sixth Circuit's opinion in Ohio State Conference of the NAACP v. Husted[, 768 F.3d 524 (6th Cir.2014), vacated on other grounds, No. 14–3877, 2014 WL 10384647 (6th Cir. Oct. 1, 2014) ] dismissed entirely evidence of other states' practices, stressing the "intensely localized assessment" that the statute requires. The court was right about the need for such an assessment but wrong, in my view, to dismiss evidence of other states' practices entirely. While the main focus should be on how the challenged practice interacts with social ... conditions withinthe state, other states' experience may well shed light on that inquiry. The fact that an ID requirement is unusually strict may be taken into consideration. So too, the fact that a state offers extraordinarily generous opportunities for early voting—in comparison with other states—might be taken into consideration as part of the totality of circumstances, should the state try to reduce that period. Evidence of other states' practices may be of limited probative value, given the particularized local inquiry that § 2 requires, but should not be disregarded entirely.

Daniel P. Tokaji, Applying Section 2 to the New Vote Denial, 50 Harv. C.R.–C.L. L.Rev. 439, 484 (2015) (footnote omitted). At least one other court has said as much. See, e.g., Frank v. Walker, 768 F.3d 744, 747 (7th Cir.2014) ("The record also does not reveal what has happened to voter turnout in the other states (more than a dozen) that require photo IDs for voting. If as plaintiffs contend a photo-ID requirement especially reduces turnout by minority groups, students, and elderly voters, it should be possible to demonstrate that effect. Actual results are more significant than litigants' predictions. But no such evidence has been offered."), cert. denied, ––– U.S. ––––, 135 S.Ct. 1551, 191 L.Ed.2d 638 (2015).

71 In this case, Mr. Trende was proffered as an expert in "U.S. campaigns and elections, the voting laws at issue in this case, United States demographics, and voting behavior." (Doc. 339 at 190.) Plaintiffs objected and sought to exclude Mr. Trende under Daubert. (Doc. 271; Doc. 339 at 207.) Plaintiffs, however, did not object to Mr. Trende's testimony so long as it was limited "as to what the laws are in each of the 50 States." (Doc. 339 at 209 ("If Mr. Trende simply wants to testify as to what the laws are in each of the 50 states, I don't think we would have an objection. To the extent that he wants to characterize particular laws as being within the mainstream or outside of the mainstream, he lacks any particular expertise in that subject.")). In any case, this court concludes that Mr. Trende is qualified to present and organize the laws of the fifty States.

Among the States offering early voting, tremendous variation exists, ranging from three to forty-six days. (Id. at 23.) And even within a State, there can be variation from county-to-county and election-to-election. (Id. at 24.) In 2014, when North Carolina offered ten days of early voting, the national median of all States and the District of Columbia was eleven days of early voting. (Id. at 23.) Twenty-one States offered fewer than ten days of early voting; twenty-six States offered more than ten days of early voting. (Id.) An analysis of the length of the early-voting period offered by each State (rather than the number of precise days offered), yields similar results. (Id. at 20.)

The types of days offered for early voting also varies by State. North Carolina, both before and after SL 2013–381, is in the minority of States that offer any weekend voting. (Id. at 35.) North Carolina is in a super-minority of States that permit voting on a Sunday. (Id.)

Election law scholars, including Plaintiffs' own expert witnesses, refer to early voting as a form of "convenience voting." (See, e.g., Pl. Ex. 42 at 59; Def. Ex. 2 (Ex. 11) at 639; Def. Ex. 348 at 95.) A fundamental component of Plaintiffs' claim is that early voting increases participation. It would seem obvious that the introduction of convenience voting would have the effect of increasing political participation. But there is a somewhat surprising scholarly consensus, created in no small part by Plaintiffs' own expert witnesses, that not only is this not demonstrated, but that empirically early voting actually tends to depress participation. (See, e.g., Def. Ex. 346 at 92–93; Def. Ex. 348 at 95.)

According to Plaintiffs' experts, political participation is defined in terms of voter turnout and registration rates. (See, e.g., Doc. 331 at 113.) As Plaintiffs' expert, Paul Gronke, Ph.D., Professor of Political Science at Redd College,⁷² wrote in a peer-reviewed publication before this litigation arose:

72 Dr. Gronke was proffered without objection as an expert in early voting, election administration, political science and research methods, voter behavior and the effect of election reforms on voters. (Doc. 332 at 206.)

[W]e remain skeptical of those who advocate in favor of early voting reforms primarily on the basis of increased turnout. Both these results, and prior work in political science, simply do not support these claims. There may be good reasons to adopt early voting—more accurate ballot counting, reduced administrative costs and headaches, and increased voter satisfaction—but boosting turnout is not one of them.

(Def. Ex. 2 (Ex. 11) at 644; see also id.(Ex. 12) at 26 ("The research thus far has already disproved one commonly made assertion, that early voting increases turnout. It does not.") (emphasis added).) Early voting "mak[es] it more convenient to be sure, but pal[es] in significance to such effects as feelings of citizen empowerment, interest in and concern about the election, and political mobilization by parties, candidates, and other political organizations." (Id.(Ex. 11) at 644.)

Another of Plaintiffs' experts, Dr. Burden, has written:

The added convenience of early voting decreases the direct costs of voting, but this effect is more than offset by a reduction in mobilization efforts, resulting in lower net turnout. ... Our unambiguous empirical claims are based on multiple data sources and methods: despite being a popular election reform, early voting depresses net voter turnout.

(Def. Ex. 348 at 95–96, 108 (emphasis added).)⁷³ Social scientists have suggested that this counter-intuitive result occurs because early voting detracts from the energy of Election Day and the mobilization efforts of campaigns and GOTV efforts of political activists. (Id. at 97–99; see also Def. Ex. 346 at 96 ("The law of unintended consequences seems to have rendered early in-person voting counterproductive to the goal for which it is often adopted: increased voter turnout.").)

73 At trial, Dr. Burden attempted to distance himself from his previous article by saying that it only analyzed jurisdictions first implementing early voting. However, the article's explanations for the depressed turnout, expressed before Dr. Burden was retained in this case, are not so limited. For these reasons, the court finds Dr. Burden's pre-litigation analysis more reliable. SeeFed. Rule Evid. 702 advisory committee notes (noting that courts consider "[w]hether experts are 'proposing to testify about matters growing naturally and directly out of research they have conducted independent of the litigation, or whether they have developed their opinions expressly for purposes of testifying' " (quoting Daubert v. Merrell Dow Pharm., Inc., 43 F.3d 1311, 1317 (9th Cir.1995) )).

Given these findings by Plaintiffs' own experts, it is of little surprise that there is no evidence in this case that North Carolina's introduction of early voting or use of seventeen days of early voting caused increased political participation either overall or for any racial subgroup.

That said, following national trends, North Carolinians have begun using early in-person voting with increasing frequency. For example, in 2000, 89.3% of North Carolinian voters voted on Election Day, while only 8.1% voted early. (Pl. Ex. 42 (Ex. 36).) In 2008, however, early voting saw the greatest increase in use ever and constituted the most popular method of voting, being used by 48.7% of North Carolinian voters. (Id.) In 2014, only 37.4% of voters used early voting, compared to 60.0% voting on Election Day. (Pl. Ex. 242 at 159.) Thus, while early voting has become increasingly popular, its popularity in relation to Election Day voting varies by year and election cycle.

The rates of early voting by racial subgroups also varies. Among voters for North Carolina general elections held from 2000 to 2012, white and African American use was nearly the same, except for three elections.⁷⁴ (Pl. Ex. 40 at 15.) In the 2004 general election, white voters disproportionately used early voting over African American voters by a difference of 5.39%. (Id.) In the 2008 general election, early voting's largest increase in use coincided with President Obama's candidacy, when African American voter use exceeded white voter use by 15.89%. (Id.) The disparity was similar but slightly less in degree in the 2012 general election during President Obama's re-election campaign. (Id.) However, a similar disparity in African American use was observed in 2014, even after SL 2013–381 became effective. (Pl. Ex. 242 at 159.) Overall, African American use of early voting has exceeded African American use of Election Day voting only in 2008 and 2012. (Id.) The same is true of white voters.⁷⁵ (Id.)

74 Broken down by year, 19.83% of African Americans and 25.22% of white voters voted early in 2004; 60.36% of African Americans and 44.47% of whites voted early in 2008; and 64.01% of African Americans and 49.39% of whites voted early in 2012. (Pl. Ex. 40 at 15). In 2000, 2002, and 2010 the white and African American use of early voting differed by less than one and a half percentage point. (Id.(8.99% African American/7.88% white in 2000, 5.21% white/4.93% African American in 2002, and 28.53% African American/28.36% white in 2010).) In 2006, 11.95% of whites used early voting, compared to 9.19% of African Americans. (Id.)

75 Those are the proportional numbers. In terms of actual numbers of early votes cast, there have always been more white than African American early votes cast. (Pl. Ex. 40 at 22.) Even in 2008, when African American voters used early voting more than in any other election, there were more than twice as many white early voters as African American early voters. (Id.) In the 2012 primary election, there were more than five times as many white early voters as African American early voters. (Id.) The court provides these figures here and elsewhere only for the sake of comprehensiveness but has not relied on them to reach any legal conclusion, for discriminatory intent or otherwise.

Results also vary when broken down by the first seven days of early voting removed by SL 2013–381.⁷⁶ From 2006 to 2012, 2,128,693 votes were cast during the first seven days of early voting—1,429,667 by whites, and 616,483 by African Americans. (Pl. Ex. 40 at 30.) Thus, 67.16% of these votes were cast by whites, while 28.96% were cast by African Americans. (Id.) Whites used the first seven days of early voting at a higher rate than African Americans in 2006 and 2010, while African Americans used the first seven days at a higher rate than whites in 2008 and 2012. (Id.)

76 When analyzed as a proportion of first week voters, the African American/white numbers are as follows: 2006 general (8.58%/90.03%); 2008 primary (31.50%/65.27%); 2008 general (31.88%/64.24%); 2010 primary (21.92%/75.05%); 2010 general (17.58%/80.25%); 2012 primary (16.12%/79.62%); 2012 general (32.93%/62.47%). (Pl. Ex. 40 at 30.)

Accordingly, the evidence shows that, although African Americans disproportionately used the first seven days in the aggregate, racial disparity turns on whether the election is a midterm or general election. (Id.) Moreover, the last ten days of the seventeen day early-voting period were the ones most heavily used, even by African American voters.⁷⁷ (See Def. Ex. 362 at 1–3; Doc. 338 at 134–41.)

77 Additionally, in the 2014 general election, by the second day of early voting (day nine on the pre-SL 2013–381 schedule), the cumulative number of African American voters had already surpassed the cumulative total from the first nine days of the seventeen day schedule in 2010. (Def. Ex. 268 at 42–43.)

In terms of age, the turnout of registered "young" voters (those aged 18 to 24, by the Plaintiffs' own definition) increased from 17.5% in 2010 to 18.0% in 2014.⁷⁸ (Def. Ex. 309 at 78.) The number of registered young voters also increased from 9.7% in 2010 to 10.4% in 2014. (Id. at 77.) The number of young early voters during that period appears to have decreased by 9.3%, according to Plaintiffs' expert Peter Levine, Ph.D., Associate Dean for Research in the Jonathan Tisch College of Citizenship and Public Service at Tufts University.⁷⁹ (Pl. Ex. 236 at 22–23.) Young early voters are disproportionately likely to put off voting until the last day of early voting. (Id. at 22.)

78 Plaintiffs' expert, Peter Levine, Ph.D., stated in his sur-rebuttal report that young voter turnout decreased from 16.7% in 2010 to 16.4% in 2014, disagreeing with Dr. Thornton's figures. (Pl. Ex. 248 at 1.) Dr. Thornton testified at trial that she reviewed this criticism, confirmed her original calculations, and compared her results with those reported on the SBOE website, which were "nearly identical." (Doc. 338 at 118.) The difference appears to arise because Dr. Levine believes that turnout as a percentage of voting age population ("VAP") is a better figure to use than Dr. Thornton's use of turnout as a percentage of registered voters. (SeePl. Ex. 236 at 15 (tbl. 2b).) Dr. Levine's method differs from the use of citizen voting age population ("CVAP") in that he counts non-citizens as potential voters, even though they are not eligible to vote. (SeeDoc. 338 at 109.) Because Dr. Levine's figures are not limited to eligible voters (i.e., citizens), the court finds that Dr. Thornton's figures are a better representation of turnout than Dr. Levine's for the purposes of this case, even though Dr. Thorton's figures give a narrower window on the changes. CVAP would be preferable, where available.

79 Dr. Levine was proffered as an expert in civic engagement and the effects of voting laws on youth voting without objection. (Doc. 334 at 102.)

To look at the impact of the change in the early-voting schedule, it is helpful to compare specifically the 2014 midterm general election—the first general election under SL 2013–381—to the prior comparable midterm general election in 2010. If having fewer days of early voting harms political participation, one might expect there to be evidence of decreased turnout between the elections. Plaintiffs' expert, Dr. Gronke, predicted as much to this court in the run-up to the 2014 general election:

I conclude from the analyses in this report that the changes to early in-person

voting that I have reviewed—eliminating the first seven days of one-stop early voting—will have a differential and negative impact on the ability of African Americans to cast a ballot in North Carolina. I know of no empirical argument by which one could conclude that African-American voters—or any voters for that matter—will successfully adjust to 40% fewer early voting days, regardless of the possibility of longer hours on those days.

(Pl. Ex. 40 at 39.) Dr. Stewart made similar predictions of adverse impact. (Pl. Ex. 42 at 56–59, 89 ("Provisions in HB 589 intended to ameliorate the reduction in early voting days are unlikely to succeed.") These analyses were cited by Plaintiffs to support their claimed need for a preliminary injunction in this case.

Contrary to these prognostications, however, turnout in the 2014 midterm general election (compared to the 2010 midterm general election), actually increased for both African Americans and whites after SL 2013–381. (Def. Ex. 309 at 59–62.) More pertinent, the 2010 disparity in turnout rates between white and African American voters decreased in 2014, after SL 2013–381. (Id.) African American use of early in-person voting increased by 7.2%, which exceeded the 1.9% increase observed among whites and the 1.6% increase among Hispanics. (Id. at 68–69.) These turnout numbers are contrary to Plaintiffs' experts' predictions and contradict the claim that SL 2013–381 has a negative, disparate impact on African Americans or Hispanics. (See also id. at 62; Def. Ex. 268 at 35.) As an apparent response to this data, Plaintiffs articulate their claim that, while they have increased their registration and turnout, it has become harder for them to do so. But this is unpersuasive.

Drs. Gronke and Stewart reached their inaccurate predictions, in part, by extrapolating from Florida's experience when it reduced early voting from fourteen to eight days.⁸⁰Brown v. Detzner, 895 F.Supp.2d 1236, 1239 (M.D.Fla.2012) ; (Pl. Ex. 40 at 25–29; Pl. Ex. 42 at 83). But, just as Florida's experience proved to lack any predictive power for North Carolina's 2014 midterm election, so, too, the court finds it unpersuasive for predicting the 2016 general election.

80 Yet another example of Plaintiffs' reliance on other States' practices. Indeed, Plaintiffs frequently cited the experience of other States when it was helpful to do so.

In Florida's 2012 general election, after the reduction of six days of early voting, there was significant congestion and a decrease in the number of early voters when compared to 2008. (Pl. Ex. 42 at 83–87.) In attempting to extrapolate Florida's experience to North Carolina, however, Plaintiffs' experts failed to consider several material differences between the two States' programs, including the type, quantity, and quality of the voting machinery; the capacity or number of the early-voting facilities; the complexity of the ballot; and the number of available poll workers. (Doc. 333 at 73–76.) Importantly, Florida had a fairly complex ballot in 2012, with multiple referenda in multiple languages, which likely affected congestion. (Id. at 75–76; Doc. 335 at 52–54; Pl. Ex. 49 at 6 (Plaintiffs' expert Theodore T. Allen, Ph.D., Professor of Integrated Systems Engineering at Ohio State University, opining that the length of the ballot directly relates to congestion).) Florida also offers fewer early-voting sites than North Carolina. (Def. Ex. 270 at 36–37.) And while "most" counties maintained similar hours as before, not all did or were required to do so, as in North Carolina. (Pl. Ex. 40 at 28.) Compared to 2008, Floridians' use of early voting declined by 10.7% in 2012, after eliminating six of its fourteen days of early voting. (Id. at 26.) By contrast, after North Carolina reduced seven days of early voting but introduced other compensating reforms, the number of North Carolinians using early voting increased by 21.1% from 2010 to 2014. (Pl. Ex. 242 at 159.) True, Florida involved presidential elections while North Carolina involved mid-terms, but the mid-term had highly contested races. And SL 2013–381's same hours requirement will ensure that in 2016 counties maintain the same number of hours as in 2012, the previous presidential election. So, North Carolina's experience was nothing like Florida's. Plaintiffs have not shown that any voters were deterred in 2014, nor have they offered persuasive evidence that the 2016 general election will be any different.

Plaintiffs supported their congestion argument with a "wait-time" analysis of early voters by Dr. Stewart. The analysis was based on his internet survey of persons who claimed to have voted in the 2008 and 2012 general elections (before SL 2013–381). In his April 2014 report, Dr. Stewart concluded that North Carolina's early-voting lines were already congested in 2008 and 2012, based on his survey data showing that 27.2% of North Carolina's early voters waited more than thirty minutes, compared to only 15.8% of early voters nationwide.⁸¹ (Pl. Ex. 42 at 75.) He opined at trial that, given this difference, there was greater early voting congestion in 2014 than 2010 because early voting use increased "roughly 20%" while available hours decreased 3%.⁸² (Doc. 332 at 84.)

81 Dr. Stewart organized his internet survey via the Survey of the Performance of American Elections and selected the questions respondents would be asked. (Doc. 332 at 138–39.) Respondents were recruited through website pop-up ads and similar internet advertisements and were promised points redeemable for prizes for completing the survey. (Id.at 139–41.)

82 A decrease in hours in 2014 is not persuasive evidence that early-voting lines will worsen. Session Law 2013–381 requires that a CBOE can reduce hours only upon unanimous agreement, strongly supporting a fact-based inference that such extra hours were unnecessary in that county.

Dr. Stewart's wait-time opinion suffers from a number of important flaws, rendering it unpersuasive. First, his conclusion assumes that North Carolina's early-voting system in 2010 was operating at full capacity, such that any additional burden would automatically result in greater wait times. Such an assumption is not supported by the record. And to the extent that increased wait times correlate with persons becoming too frustrated to vote, the actual early-voting figures from 2014 demonstrate an increase in the number of people successfully casting an early ballot.

Second, Dr. Stewart's surveys were based on very few observations. In 2008, only ninety-five respondents claimed to have been North Carolina early voters; in 2012, only ninety-one, and in 2014, 425.⁸³ (Pl. Ex. 42 (Ex. 42); Pl. Ex. 242 at 85 (tbl. 16).) Looking at just 2014, Dr. Stewart extrapolated the sampled survey responses of the 425 purported North Carolina early voters onto a population of 1,097,942 early voters. (See Pl. Ex. 242 at 159.) Conclusions drawn from this data are subject to a high margin of error.⁸⁴ (See Def. Ex. 246 at 22–23; Def. Ex. 309 at 79–87.)Third, the survey responses themselves have plain indicia of unreliability. For the 2014 survey, 73.8% of respondents claim to have voted either on Election Day or through early voting. (Pl. Ex. 242 at 85.) However, this figure far exceeds North Carolina's overall turnout rate for the 2014 general election, which was only 38.8% of the voting age population. (Pl. Ex. 242 at 161 (App'x U).) Thus, either the survey respondents were untruthful about whether they voted—perhaps in hopes of being compensated (in which case their voting experiences are highly suspect),⁸⁵ or the survey has an inherent bias toward selecting actual voters motivated to comment, showing that the sampling is far from random. The former would be consistent with the testimony of Plaintiffs' experts throughout the trial, who noted that certain survey respondents of the decennial Census Current Population Survey ("CPS") consistently over-report whether they have voted. (See e.g., Doc. 339 at 111.) Either way, the survey is suspect.

83 In 2014, North Carolina was "oversampled": after surveying 200 North Carolina respondents, an additional 1,000 were also sampled. (Pl. Ex. 242 at 84–86.) The oversampling seems to have been at Dr. Stewart's suggestion.

84 The survey questions were also open to variable interpretations among respondents. (Def. Ex. 309 at 80.) In addition, the survey did not reveal any statistically significant difference in wait times encountered by African Americans and whites. (Id.at 84–87.)

85 It is notable that Dr. Stewart's survey could have asked for identifying information, which would have permitted him to confirm, through information publicly available on the North Carolina SBOE website, whether respondents had in fact voted. But Dr. Stewart made no such effort to do so. (Doc. 332 at 143–44.)

Fourth, even assuming the respondents truthfully reported whether they voted, the survey design assumes a level of human memory that is unrealistic.⁸⁶ When the respondents went to vote, they had no idea they would later be asked to calculate the number of minutes they spent waiting and voting. Given that some respondents voted early but were not surveyed until a week after Election Day, they were asked to recall their wait times from up to nineteen days earlier.

86 Respondents were asked: "Approximately, how long did you have to wait in line to vote?" (Pl. Ex. 42 at 74.) The available responses were: "(1) Not at all, (2) Less than 10 minutes, (3) 10–30 minutes, (4) 31 minutes—1 hour, (5) More than 1 hour [with follow-up prompts], and (6) I don't know." (Id.)

In addition, Defendants' evidence indicated that long wait times were not common in the 2014 general election.⁸⁷ For example, the SBOE surveyed all CBOEs (who are presumably more attuned to focusing on wait times than Dr. Stewart's internet respondents) after the November 2014 general election as to early voting and Election Day wait times. (Def. Ex. 210.) Of the 368 early-voting sites, the vast majority (64 counties) reported wait times of 0–30 minutes, and 23 counties reported experiencing a wait time of 30–60 minutes. (Id. at 4.) Only thirty-six early-voting sites reported wait times of more than an hour, and those were either on the first two days or last three days of early voting, with thirty occurring on the last day. (Id. at 3.) For the middle five days of early voting, no site experienced wait times greater than an hour. (Id. at 3–4.) Similar figures are reported for Election Day waits. (Id. at 5.)

87 Some of the voting lines from 2014 were caused by electronic touchscreen equipment, (Def. Ex. 210 at 6), which Dr. Stewart believes to cause lines, (Doc. 332 at 148–49), and which is set to be phased out by SL 2013–381, § 30.8.

The SBOE's survey, however, suffers from its own methodological shortcomings. First, there is no evidence that CBOEs were notified that they would be asked about wait times until after the election. (Pl. Ex. 817 at 73–74.) Second, it does not appear that CBOEs had any mechanism to measure wait times. (See id.) Nevertheless, if a significant voting problem occurs, CBOEs are likely to learn of it. Accordingly, while the SBOE's survey has its own reliability problems, it is some evidence that, contrary to Dr. Stewart's assertions, major wait-time problems did not occur in the 2014 general election.

For all of these reasons, the court declines to credit Dr. Stewart's wait-time analysis.⁸⁸

88 Another of Plaintiffs' experts, Dr. Allen, relied on Dr. Stewart's wait-time analysis to try to predict wait times on Election Day in 2016. (Pl. Ex. 49 at 14.) Dr. Allen's opinion is vulnerable, in part, because it relied on Dr. Stewart's unreliable analysis. Moreover, Dr. Allen's analysis merely gave various possible effects based on the numbers of early voters transitioning to Election Day voting. However, it is unknown how many net voters—if any—will transition. Dr. Allen also failed to factor into his analysis SL 2013–381's same-hours requirement, a critical component of the new early-voting schedule. Therefore, Dr. Allen's analysis is of little assistance.

An additional reason Plaintiffs' experts' predictions did not come to pass is that they refused to engage in meaningful analysis of SL 2013–381's same-hours requirement. Before the 2014 general election, Dr. Stewart opined that North Carolinians most frequently early vote in the middle of the day; he believed that any new hours added to satisfy the same-hours requirement would have to be added at less-used times, such as in the evenings. From this, he opined that the same-hours requirement would have little ameliorative effect on the reduction of early-voting days. (Pl. Ex. 42 at 75–79.)

This opinion made little sense then and has been further discredited by the results of the 2014 general election. Dr. Stewart examined figures from 2012, which showed that most early voters went to the polls between 11:00 a.m. and 4:00 p.m. (Id. at 77.) From this use data, he concluded that voters prefer to vote in the middle of the day and will be neither able nor willing to vote at other times of the day. (Id. at 76–79.) However, before SL 2013–381, relatively few evening or weekend hours were offered. (See Pl. Ex. 242 at 80 (fig. 12).) Therefore, while patterns of early-voting use could have been a function of voter preference, it appears more likely they were a function of early-voting availability.

A simple example reveals the false assumptions in Dr. Stewart's logic. One of Plaintiffs' challenges to the reduction of early-voting days is that African American voters prefer to vote early on that first Sunday when their church provides transportation to polling sites. However, in 2010, no African American voted on the first Sunday of early voting. (Def. Ex. 268 at 40.) One might conclude, therefore, that African Americans do not prefer to vote on Sundays. But that would be wrong because, in fact, no county elected to offer early voting on the first Sunday during early voting in the 2010 midterm election. (See Doc 126–4 at 45–90.) Therefore, use can be a function of mere availability, not necessarily preference.

Similarly, based on the 2014 data, it is clear that North Carolinians respond to new early-voting opportunities. In 2014, counties complied with the same-hours requirement by expanding evening and weekend hours. (See, e.g., Pl. Ex. 242 at 80, 167.) Looking at the data, it is apparent that the change in use from 2010 to 2014 followed the change in availability. The number of available weekday evening hours (from 5:00 p.m. to 8:00 p.m.) increased by 75.6%, and the number of votes cast during evening hours increased by 87.6%. (Pl. Ex. 242 at 167–68.) Likewise, the number of available weekend hours increased by 55.4%, while the number of weekend votes increased by 42.2%. (Id.) Evening hours are more convenient for many voters than midday hours because citizens can vote after leaving work. (See Doc. 335 at 80.)

In addition, SL 2013–381 resulted in more early-voting sites than were available not just in the previous 2010 midterm, but in the 2012 presidential election as well. (Doc. 340 at 205; Def. Ex. 13 (showing a 24.32% increase in early-voting sites from 2010 to 2014).) Even more hours and sites will be available in 2016. (Doc. 340 at 206.)

Even if preferences can be inferred from use, an inference Dr. Stewart consistently tries to draw, then the data suggest that voters "prefer" the early-voting schedule of 2014 over that of 2010 because, in actuality, they heavily used the new hours. Actual 2014 turnout suggests strongly that the new early-voting schedule did not deter voters and that the prior schedule was not necessarily the preferred one.

Plaintiffs have urged that it will be difficult for voters—African Americans in particular—to adjust to the new early-voting schedule. Dr. Gronke supplemented his 2014 report after the 2014 general election to conclude that African American early voters from 2012 were more likely not to vote in 2014 than white voters, thus asking the court to infer that such voters were likely deterred by the new early-voting schedule. He did this through a "voter transition" analysis, explaining:

[R]ather than look at aggregate turnout totals, we can examine the behavior of individual early voters before and after the reductions to early voting were implemented. This transition analysis has the advantage of comparing the same pool of voters across different elections and different legal contexts, and focuses on voter behavior at the individual level, rather than on aggregate vote totals. This is perhaps the best way to try to isolate the impact of the legal changes on an individual's tendency to cast a one-stop ballot.

(Pl. Ex. 234 at 11.)

Dr. Gronke identified white and African American voters who had voted early in 2012 to examine how they voted in the 2014 midterm election. He provided an illustration of his analysis, (id. at 12 (fig. 4)), and pointed to multiple disparities. First, 39.41% of African American 2012 early voters did not vote at all in the 2014 midterm election, which Dr. Gronke denominates a "drop-off rate," compared to only 31.86% of white early voters. (Id.) White 2012 early voters were also more likely to vote early again in 2014 or to vote on Election Day in 2014 compared to African Americans. (Id.) From these disparities, he concludes, "There are a number of possible, non-mutually exclusive, reasons for these disparities. But they provide some evidence that, contrary to the claim that voters can easily adapt to a shorter period of time for early voting, African American voters may have been less able to adapt than were White voters." (Id. at 13.)

A more comprehensive analysis, however, reveals that Dr. Gronke's "disparities" are actually part of a pattern unrelated to, and in fact pre-dating, SL 2013–381.

Overall, Dr. Gronke's analysis disguises the fact that white turnout levels frequently exceed African American turnout levels in midterms, but that African American turnout levels have exceeded white turnout levels in the 2008 and 2012 presidential elections. (Pl. Ex. 242 at 161 (App'x U).)

More importantly, Dr. Thornton applied Dr. Gronke's method of transition analysis of 2012–2014 to 2008–2010, the previous comparable transition (presidential-to-midterm), which was a period not impacted by SL 2013–381. She found similar disparities between whites and African Americans as Dr. Gronke had found for the impacted transition period—except that the disparities were even greater in the 2008–2010 transition. Among African American early voters in 2008, 41.18% did not vote in 2010, compared to only 33.14% of whites. (Def. Ex. 309 at 73.) Thus, the white-African American drop-off disparity from 2008– 2010 actually decreased in the 2012–2014 transition analysis. (Compare id., with Pl. Ex. 234 at 12 (fig. 4).) The racial disparity in whether an early voter was likely to vote early again also decreased from the 2008–2010 period to the 2012–2014 period. (Id.) Thus, if a voter transition analysis "is perhaps the best way to try to isolate the impact" of SL 2013–381, as Dr. Gronke urges, then his conclusion is wrong, and the reduction in early-voting days tends to benefit, rather than harm, African American voters.

Further undermining Plaintiffs' contention that African Americans are less able to adjust to the remaining days of early voting is Dr. Thornton's drop-off transition analysis regarding users of the eliminated seven days of early voting. She identified those who voted early during the first seven days in 2010 and examined whether they voted in 2014. (Def. Ex. 362 at 1.) She found that those who voted in the first seven days of early voting in 2010 were more likely to have voted in 2014 than those who voted in the last ten days. (Id.; Doc. 338 at 134–41.) This conclusion is valid for both African American and white voters. (Def. Ex. 362 at 1.) Dr. Thornton conducted similar analyses for the transition periods of 2008–2010 and 2012–2014. (Id. at 2–3.) She found that those who voted during the first seven days of early voting in 2012 were more likely to vote in 2014 than were the same 2008 early voters transitioning to 2010. (Id.) Importantly, this conclusion is valid for both African American and white voters. (Id.) Accordingly, although Plaintiffs have established that African Americans disproportionately used the first seven days of early voting during general elections, Dr. Thornton's analysis tells us something about these early voters regardless of their race: they are not the marginal voter, but instead are more motivated and adaptable than other early voters.

Plaintiffs also offered the testimony of several fact witness in support of their claims.⁸⁹ Illustrative here is the video testimony played at trial of two affected voters who explained their problems with early voting and lines. Tawanda Pitt, an African American nurse and a resident of Wilson, North Carolina, testified that she tried to vote on Election Day in 2014. (Pl. Ex. 798 at 8.) The first time she arrived at her precinct, there was a long line. (Id. at 10.) She understood that the precinct would be receiving a new computer that would expedite the process, so she left and returned two hours later. (Id.) According to Ms. Pitt, the precinct had only two computers, which was down from the four or five computers it had in the past. (Id. at 10, 19.) When she returned, however, the line was longer. (Id. at 10.) Both times she waited about "30, 35 minutes." (Id. at 11.) She gave up the second time when the poll worker could not tell her when the new computer would arrive, (id. at 23), so she could fix dinner and help her son with his homework, (id. at 25). She ended up not voting in 2014.

89 Plaintiffs submitted the deposition transcripts of sixty-five fact witnesses. At trial, however, Plaintiffs highlighted only a handful of them. The court presumes those were Plaintiffs' stronger witnesses, but it has nevertheless reviewed the transcripts of all fact witnesses.

It is unfortunate that Ms. Pitt did not vote, but her difficulties, and the line she experienced, were at least in part due to technological difficulties (fewer computers than in prior years) not attributable to SL 2013–381. Put simply, SL 2013–381 did not change the number of computers available to precincts on Election Day. In addition, Ms. Pitt had not tried to vote early and did not know how long the lines were during early voting. (Id. at 25.)Sherry Durant is African American and, due to her cerebral palsy, lives in a group home, which severely restricts her mobility. (Pl. Ex. 721 at 6, 9.) She wanted to vote in 2014, but was incapable of getting herself to a polling place. (Id. at 12.) Several other residents at the group home also wished to vote. (Id. at 17.) A group home worker, Ms. Graves, proposed to take them all to vote in person. (Id.) It was clear that Ms. Durant was not aware of how many days of early voting were offered before or after SL 2013–381. (Id. at 19.) Ms. Durant testified that Ms. Graves was not able to organize a voting trip for any of the residents during the early-voting period or on Election Day. (Id. at 17–18.) Ms. Graves did not testify, but Ms. Durant claimed that every day Ms. Graves planned to take the residents to vote "one of [the] residents had to go to the doctor or [Ms. Graves] had to go to the doctor herself or [Ms. Graves] needed to be present elsewhere or it was just chaotic." (Id. at 28.) Ms. Durant did not know Ms. Graves' schedule in the previous seven days, when early voting would have been available without SL 2013–381, and there is nothing in Ms. Durant's testimony that suggests that Ms. Graves's availability would have been any different then. (Id.)

Finally, Plaintiffs contend that the new early-voting schedule disproportionately burdens African Americans by removing a "souls-to-the-polls"⁹⁰ Sunday. However, the evidence demonstrated that many churches who provide Sunday transportation also provide transportation to the polls throughout the early-voting schedule. (E.g., Pl. Ex. 793 at 27–28.) Accordingly, in addition to retaining one Sunday, these churches are positioned to take advantage of the additional night and weekend hours created by the same-hours requirement.

90 These are church-related efforts to engage congregants and provide resources to get them to the polls to vote, including during Sunday early voting.

In sum, the court has evaluated all of the evidence surrounding the impact resulting from the change in the early-voting schedule. In light of the same-hours requirement, the evidence does not demonstrate that the new early-voting schedule results in a reduced opportunity to vote or imposes a burden on voters. Nor does the evidence show that the new schedule disparately and negatively impacts the political participation of African Americans, Hispanics,⁹¹ or young voters.⁹²

91 Plaintiffs have not provided any evidence that Hispanic voters disproportionately used early voting. (See, e.g., Doc. 346 discussing Hispanics only with regard to SDR, OOP, and pre-registration). In fact, the evidence Plaintiffs' experts provided on racial disparities in the eliminated seven days of early voting do not include Hispanics. (See, e.g., Pl. Ex. 40 at 30; Pl. Ex. 42 (Ex. 41).) Accordingly, Plaintiffs have failed to show that Hispanic voters have been disparately impacted by SL 2013–381's change in the early-voting schedule based on disproportionate use.

92 The evidence as to young voters was that they did not focus on elections until closer to Election Day, but, like other groups, there was no evidence that young voters will not benefit from the additional night and weekend hours created by SL 2013–381.

Contrary to all of Plaintiffs' dire predictions, turnout actually increased for all voters under SL 2013–381. In many ways, the new early-voting schedule is an improvement for all North Carolina voters. Comparing 2010 to 2014, the new schedule resulted in 24.32% more early-voting sites, 72.14% more evening hours (with 45 counties newly offering evening hours), 4 counties newly offering Sunday hours, and 26.62% more Sunday hours overall. (Def. Ex. 13.) There was also no persuasive evidence that the new schedule increased lines at early-voting centers or that such lines deterred minority voters. (See Doc. 332 at 160–61.)⁹³ Conversely, there is no credible evidence that the old schedule itself increased political participation generally or among any subgroup of voters; the previous academic consensus was to the contrary. What can be said is that all parties acknowledge that strong minority use of early voting in 2008 and 2012 was a result, in some measure, of a Democratic campaign strategy in North Carolina, particularly President Obama's campaign, which specifically encouraged the use of early voting over Election Day voting. (Def. Ex. 270 at 58–62; Doc. 331 at 90–91 (Dr. Burden: stating that the Obama campaign "emphasized [early voting] fairly heavily"); Doc. 332 at 158–59 (Dr. Stewart: stating that it is "certainly true" that the "Obama campaign had an impact on the modes of voting by Obama supporters").)

93 Plaintiffs' evidence of voters waiting was largely anecdotal and not representative of any systemic issue. (E.g., Doc. 330 at 110–12 (Isabel Najera: waited "around two hours," but her delay was attributable to a poll worker's investigation after discovery that she was not listed on the voter roll); Pl. Ex. 792 at 13–14 (Quisha Mallette: UNC law student who "had to sit in there for a little while" while waiting to vote provisionally because she did not switch her registration to her county of residence); Pl. Ex. 798 at 19, 25 (Tawanda Pitt: did not try to early vote and left polls on Election Day after waiting thirty-five minutes on two occasions).)

There was also no persuasive evidence that voters were habituated to the old schedule or had any difficulty adjusting to the new schedule. In fact, voters who testified at trial did not even seem to be aware of how many days were offered under the old or new law without being prompted by Plaintiffs' counsel. (E.g., Doc. 331 at 167–69, 173 (Nadia Cohen: did no research into voting deadlines, conceding that "voting is not my top priority"); Pl. Ex. 721 at 19 (Sherry Durant: "it was basically going from what did you say, 15 to ten" days).)

For these reasons, Plaintiffs have failed to show that it is harder for any voter, including African Americans, to vote under the ten-day early-voting schedule given the same-hours requirement. Plaintiffs' predictions for the 2016 presidential election are unpersuasive, and the 2014 results demonstrate that the ten-day voting schedule and the same-hours requirement combine to produce more high-convenience voting hours. In addition, the evidence shows that, regardless of race, those who voted during the first seven days of early voting under the seventeen-day early-voting schedule are more likely to vote under the ten-day schedule than are those who voted in the last ten days of the former seventeen-day schedule. Likewise, the evidence indicates that churches are positioned to take advantage of the new voting sites and hours in their GOTV efforts. For these reasons, while the ten-day early-voting schedule makes early voting different, Plaintiffs have failed to prove that it makes voting harder.

3. Elimination of SDR

During the six years it was permitted, SDR allowed citizens to register and then vote at an early-voting site during the early-voting period. Session Law 2013–381's elimination of SDR returned North Carolina to the pre-2007 state of affairs, and voters must comply with North Carolina's twenty-five day registration cut-off in order to be eligible to vote. Even after the repeal of SDR, however, a voter who has moved within his county may update his registration information, including a change of address, during early voting or on Election Day, and vote. SeeN.C. Gen. Stat. § 163–82.6A(e).

SDR had limitations. It was available only during the early-voting period and not on Election Day (the latter being known as Election Day Registration ("EDR")). It was also available only at designated early voting sites in the county in which the citizen resided. See 2007 N.C. Sess. Law 253, § 1. SDR was unavailable in the gaps between the twenty-five day cut-off and the start of early voting, and following the close of early voting to Election Day. Moreover, because North Carolina requires residency in the assigned precinct for thirty days before any election, N.C. Gen. Stat. § 163–55(a), SDR did not aid a voter who had moved within thirty days of an election who sought to register to vote for elections specific to his new precinct (although even now, if he had moved to his new precinct within the same county more than thirty days before the election, he can update his registration as an unreported mover and vote the full slate in his new precinct).

Plaintiffs' expert, Dr. Stewart, described the type of person who tended to use SDR:

Some people register in "blackout periods" in the weeks preceding elections. Based on research about voter registration in the political science literature, it is clear that many of the registrations that occur during blackout periods are people who are not normally attentive to public affairs, who have become attuned to politics during the presidential election season—a brief period every four years where matters of politics and elections dominate a wide variety of media channels.

(Pl. Ex. 42 at 48.)

When it offered SDR, North Carolina was in a small minority of States that did so.⁹⁴ Three States offer EDR—a very different electoral mechanism—but not SDR. (Def. Ex. 270 at 29–31.) Twelve jurisdictions offer both EDR and SDR; this figure includes the District of Columbia, as well as North Dakota, which does not require registration at all. (Id.) Currently, thirty-six States, including North Carolina, offer neither SDR nor EDR.⁹⁵ (Id.) In fact, before SL 2013–381, North Carolina was the only State in the Nation to offer only SDR during early voting.⁹⁶

94 Whether certain States should be classified as offering SDR or EDR is subject to interpretation and coding. (Def. Ex. 270 at 29–31.) However, it was undisputed at trial that the majority of States do not offer SDR or EDR. (Doc. 331 at 101 (Plaintiffs expert, Dr. Burden, conceding that a majority of States do not offer SDR)).

95 At the time of his report, Defendants' expert Sean Trende put this number at thirty-seven. (Pl. Ex. 270 at 29–32.) In that report, Vermont was coded as having neither EDR nor SDR, but appears to have enacted some variation since. Vt. Stat. Ann. tit. 17, § 2144a(4) (effective January 1, 2017); (Doc. 340 at 15 (Trende: saying that Vermont enacted SDR "a few weeks" before trial); see alsoMcCrory, 997 F.Supp.2d at 351 n. 34 (setting forth registration cut-off dates for States without EDR or SDR, many of which are longer than the twenty-five day cut-off period in North Carolina, (e.g., Alaska Stat. Ann. § 15.07.070(c) –(d) (30 days); Ariz. Rev. Stat. Ann. § 16–120 (30 days); Ark. Code Ann. § 7–5–201(a) (30 days)).

96 Ohio had "Golden Week," when normal registration overlapped with early voting for five days. (Def. Ex. 270 at 31–32.)

Plaintiffs' experts claim that SDR boosts turnout. (Doc. 342 at 130–31.) But there is no reliable statistical evidence that this is so. As recently as a January 2014 article, Plaintiffs' own expert, Dr. Burden, found that while his statistical analyses "suggest" that SDR has the potential of offering a mechanism to enhance the mobilization of certain voters, the results are not statistically significant (i.e., the 95% confidence interval includes the hypothesis of no effect). (Def. Ex. 348 at 101–02.) Similarly, a 2011 study involving North Carolina's 2008 general election found it "impossible" to isolate the effect of SDR in turnout. (Def. Ex. 346 at 93 (noting that the variable measuring both SDR and competitive gubernatorial races showed "no statistically significant influence on turnout" because of no comparative data). Thus, no reliable conclusions can be drawn.

The effects of EDR are quite different. The academic consensus is that EDR has a consistent, positive effect on turnout. As Plaintiffs' own expert, Dr. Burden, explains:

The only consistent way to increase turnout is to permit Election Day registration. Early voting reduces turnout by robbing Election Day of its stimulating effects. This depressant effect is only partially offset if SDR is present or if EDR offers a vehicle for the last-minute mobilization of marginal voters.

(Def. Ex. 348 at 108.) Other researchers have confirmed this finding, while also maintaining the important distinction between EDR and SDR. (See, e.g., Def. Ex. 346 at 78, 80, 89, 93, 96–97.)

Despite this body of scholarship, created in part by Dr. Burden, other Plaintiffs' experts were unaware of the different effects of SDR and EDR and carelessly comingled the two, scuttling the difference. For example, in his 2014 report, Dr. Stewart references how EDR boosts turnout, (Pl. Ex. 42 at 51), a point that has no relevance to the impact of adding or removing SDR in North Carolina. At trial, Dr. Stewart stated that the literature supports the notion that SDR increases turnout. (Doc. 342 at 130.) When pressed to identify this literature, however, he could only point to a chapter in a book dealing with EDR (or combining EDR with SDR). (Id. at 130–31.) Dr. Stewart testified that he was unaware of any study that examined SDR separately from EDR. (Id. at 131.) This was odd, since one of his co-experts, Dr. Burden, had performed just such a study about which Dr. Burden testified at trial.

Dr. Gronke went further than Dr. Stewart. He cited the relevant scholarship distinguishing the effects of SDR from EDR, but totally mischaracterized the articles. In his April 14, 2014 report, Dr. Gronke wrote, "For same-day or Election Day registration, there are essentially no dissents; there is essentially universal agreement among scholars that this is an election reform that has a substantially positive impact on voter turnout." (Pl. Ex. 40 at 33.) He went on to characterize Dr. Burden's article (Def. Ex. 348) as finding "a positive effect when early voting was offered in conjunction with EDR, as was the case in North Carolina prior to recent election law changes." (Pl. Ex. 40 at 33–34 (latter emphasis added).)

Dr. Gronke conceded at trial that his SDR analysis was not at all accurate. First, there is no scholarship finding an overall positive effect on turnout from SDR, since it requires the existence of early in-person voting, a mechanism that depresses turnout. What is troubling is that Dr. Gronke should have known as much, since he had served as a peer reviewer of Dr. Burden's article prepared prior to this litigation. (Doc. 333 at 50–51.) Second, contrary to Dr. Gronke's representation, North Carolina has never offered EDR. (Id. at 53.) As a result, the court finds Drs. Stewart and Gronke unreliable on the scholarly literature on SDR.

In 2014, before the preliminary injunction hearing in this case, Dr. Gronke predicted, as he had regarding the reduction of early voting, that the elimination of SDR would certainly reduce African American turnout in the 2014 election:

I conclude from the analysis in this report that, because same-day voter registration has been shown to be a strong and consistent predictor of higher turnout, the elimination of same-day registration during the election process, whether during one-stop voting or on

Election Day,⁹⁷ will lower turnout overall. In particular, I conclude that eliminating same-day registration will have a disparate impact on African-American voters because they take advantage of same-day registration at a significantly higher rate.

97 Again, Dr. Gronke is misinformed. North Carolina never had EDR.

(Pl. Ex. 40 at 39 (emphasis added).) This turned out to be a poor prediction because, as noted above, African American turnout actually increased in 2014. (Def. Ex. 309 at 66; Pl. Ex. 229 at 7.) Plaintiffs have provided no evidence in this case that the African American share of the 2014 vote would have been any higher had SDR (or OOP voting, or the first seven days of early voting) not been eliminated. Against this backdrop, it is not surprising that Dr. Gronke avoided giving any opinions about SDR in his 2015 report. (See Pl. Ex. 234.) But he never amended his inaccurate 2014 report, despite reserving the right to do so. (Pl. Ex. 40 at 39.)

In examining the use of SDR in North Carolina, it is helpful first to examine the changes in registration rates. SDR was only in place for three general elections: 2008, 2010, and 2012. After 2006, African American registration rates exceeded those of whites, and a disparity favoring the African American electorate has been growing ever since. (Pl. Ex. 684.) Plaintiffs assert that, since African American registration rates exceeded white registration rates after SDR was implemented, this was most likely because SDR was implemented. Dr. Stewart has employed this post hoc ergo propter hoc reasoning: "There is no doubt that the same-day registration vehicle has been an important part of the laudable parity in black-white registration rates achieved in North Carolina ...." (Pl. Ex. 42 at 23.) Yet Dr. Stewart conceded at trial that he had done no analysis to reach a causal conclusion. (Doc. 332 at 151–52.) Being nothing but his ipse dixit, the court need not, and does not, accept this conclusion. See Gen. Elec. Co. v. Joiner, 522 U.S. 136, 146, 118 S.Ct. 512, 139 L.Ed.2d 508 (1997) ("Nothing in either Daubert or the Federal Rules of Evidence requires a district court to admit opinion evidence that is connected to existing data only by the ipse dixit of the expert.").

A closer examination of the time period at issue reveals something different. Dr. Stewart presented the following findings as to registrations among the voting age population ("VAP"):

Black White Black pct. minus Number Percent Number Percent White Year VAP^a registered registered VAP registered registered pct. 2000 1,218,470 988,134 81.1% 4,527,155 4,082,850 90.2% -9.1% 2006 1,356,761 1,116,818 82.3% 4,860,639 4,248,469 87.4% -5.1% 2008 1,427,617 1,354,976 94.9% 5,067,215 4,596,476 90.7% 4.2% 2010 1,502,563 1,339,180 89.1% 5,240,438 4,534,617 86.5% 2.6% 2012 1,566,467 1,492,839 95.3% 5,385,029 4,728,853 87.8% 7.5% 2014 1,634,801 1,491,076 91.2% 5,605,570 4,674,320 83.4% 7.8%

(Pl. Ex. 684.) From 2000 to 2006, when SDR was not in place, the African American/white disparity shrank from 9.1% to 5.1%. (Id.) The disparity further shrank from 2006 to 2008, when SDR was first implemented and President Obama first ran for national office, and in fact resulted in a disparity advantaging the African American electorate. (Id.) However, from 2008 to 2010, the African American advantage remained, but declined, despite the availability of SDR. (Id.) The African American advantage rose again in 2012, when President Obama ran for re-election. (Id.) Then, in 2014, after SDR had been eliminated, the African American advantage in registration rates rose even further to 7.8%. (Id.) This was quite surprising given that, from 2008 to 2010, the previous presidential-to-midterm election transition, when SDR was available, the African American advantage fell. (Id.)

Dr. Stewart supported his conclusion by performing a "churn" analysis. The churn refers to the dynamic nature of the voter rolls. The voter rolls are in constant flux: new registrations add voters onto the rolls (e.g., those turning voting age or migrating into the State); list maintenance removes voters from the rolls (e.g., those who have died or migrated out of the State). Dr. Stewart examined two two-year periods to demonstrate the net effects of the churn. For the period from 2010 to 2012 (transitioning from a midterm to a presidential election), when SDR was in place, he found that 663,927 voters were removed from the rolls, but that 1,112,412 voters were added; thus, there was a net gain. (Pl. Ex. 686.) But in the period from 2012 to 2014 (from a presidential election to a midterm election), during which SDR was eliminated, 662,305 voters were removed from the rolls, and only 640,417 were added; thus, there was a net loss of 21,888 registrations. (Id.) From this analysis, Plaintiffs argue that the repeal of SDR affects voter churn and will, over the long term, negatively impact registration rates. (Doc. 346 at 67.)

As it turns out, however, Dr. Stewart's analysis is incomplete, if not misleading. Defendants' expert, Dr. Thornton, widened the period for Dr. Stewart's churn analysis to include the period from 2008 to 2010 (from a presidential election to a midterm), another period during which SDR was in place. (Def. Ex. 359.) She found that, during this period, 581,188 voters were removed from the polls and only 517,181 voters were added—a net loss of 64,007. (Id.) The churn for the 2008 to 2010 transition, from a presidential election to a midterm election, was worse(almost three times so) for registration rates than was the analogous 2012 to 2014 transition, when SDR was eliminated. (Id.) So, a more complete churn analysis seems to support, rather than rebut, Defendants' argument that the elimination of SDR does not harm registration rates.⁹⁸

98 At trial, Dr. Stewart conceded the accuracy of Dr. Thornton's churn analysis. (Doc. 342 at 126–27.) However, he refused to concede that her figures undermine his conclusions. (Id.) Insofar as Dr. Stewart's opinions appear to be impervious to new facts or data and he was content to present what he must have known was, charitably put, an incomplete analysis, the court views his opinions with a skeptical eye. See, e.g., United States v. Williams, 731 F.3d 1222, 1230 (11th Cir.2013) ; Gutheil & Simon, Narcissistic Dimensions of Expert Witness Practice, 33 J. Am. Acad. Psychiatry L. 55, 57 (2005) ("[R]efusing to concede even valid points ... seriously impairs credibility.").

Plaintiffs point to the timing of the implementation of SDR, arguing that the increase in registration rates in 2008 is strong evidence of the benefit it provided. However, the data show that, before SDR was implemented, African American registration rates were already increasing at a rate higher than those of whites. And, after SDR was eliminated, African American registration rates have continued to grow faster than those of whites. What confounds the inquiry as to 2008 (and 2012) is that the largest increase in African American SDR use coincides with the candidacy of the first African American president of the United States, which Plaintiffs concede has "surely" been a factor in the increase in African American participation. (Pl. Ex. 44 at 6.) That said, Plaintiffs downplay the role the President's candidacy had in increasing African American turnout in 2008 and 2012. Yet, they also dismiss the increase in African American turnout after SDR was repealed because of what they characterize as an unusually heightened interest in North Carolina's Senate race, the Nation's most expensive. While campaign effects surely play a role in these elections, these are inconsistent positions.⁹⁹ Further confounding the inquiry as to opportunity, burden, and cause is evidence that African Americans in similarly-situated States that did not have SDR in 2008 and 2012 also saw similar increases in African American registration rates, as campaigns exploited registration and voting opportunities available in those States. (Def. Ex. 270 at 46.)

99 The large increase in African American participation in 2008, when State spending was $22 million, compared to 2012, when State spending was almost $100 million, is strong evidence that the novelty of the candidacy of the first African American candidate for the presidency played a significant role in turnout. (Def. Ex. 270 at 10–11.)

In total aggregate numbers, it is indisputable that African American voters disproportionately used SDR when it was available. According to Dr. Stewart, African Americans comprised 35.5% of registrants during the SDR period for the 2008 presidential election and 32.0% of registrants during the SDR period for the 2012 presidential election,¹⁰⁰ which exceeded their roughly 22% proportionate share of all registered voters.¹⁰¹ (Pl. Ex. 42 at 46–47; Def. Ex. 309 at 76.) Plaintiffs argue that they have become burdened by the elimination of SDR because African Americans became habituated to using SDR during its six years of availability.

100 Dr. Stewart considered only those using SDR to become new North Carolina registered voters. (Pl. Ex. 42 at 43–47.) Further, Dr. Stewart focuses on those who registered during the early-voting period, rather than on those who actually registered using SDR. These two numbers are not necessarily the same. For example, in 2002, 2,326 African Americans registered during the early-voting period. (Id.(Ex. 31).) Because SDR did not exist, they clearly registered via the traditional method. Accordingly, while the court accepts that those who register during the early-voting period may be more likely to prefer SDR, which allows them to vote, over non-SDR, which does not, Dr. Stewart's data do not tell us how those who registered during the early-voting period registered. (Id.at 46–47.)

101 In raw numbers, far more whites used SDR during these two years. (Pl. Ex. 42 at 46.) This fact is not considered in the calculus, however, as it is the disproportionate use that is at issue.

Habituation is an individual-level characteristic, not an aggregate one. (Doc. 333 at 64–65.) Plaintiffs' experts acknowledged this and tracked individuals in their early-voting analysis, but they did not do so for individual voters using SDR, even though the data are available to do so. Such an analysis would be designed to determine whether whites or African Americans were more likely to use SDR more than once.¹⁰²

102 Moreover, in the 2004 presidential election (before SDR), African American voters were still disproportionately likely to register during the early-voting period even though it would not enable them to vote in the upcoming election. (Pl. Ex. 42 (Ex. 31).) It cannot be that these African American registrants were habituated to using SDR, since SDR did not exist.

Second, statistics about SDR use do not demonstrate what these particular voters, of any race, would have done had SDR not been an option, especially given that there are a multitude of easy ways to register in North Carolina apart from SDR. The registration period is open year-round, but to be eligible to vote in an election a registrant must register twenty-five days before the applicable Election Day. N.C. Gen. Stat. § 163–82.6(c). Applications are available online at the SBOE website, at the SBOE, CBOEs, public libraries, public high schools, and college admissions offices throughout the State. Every State resident can register to vote by mail, seeN.C. Gen. Stat. § 163–82.6(a) ("The county board of elections shall accept any form described in [N.C. Gen. Stat. §] 163–82.3 if the applicant submits the form by mail, facsimile transmission, transmission of a scanned document, or in person."), which permits the registrant to enlist the assistance of a family member and others to navigate and complete the one-page, seven-question form, see id.; (see also Doc. 331 at 41–42). Thus, those with transportation, economic, or other challenges need not physically appear to register.¹⁰³ Certain State agencies, as required by the NVRA, 52 U.S.C. § 20506 (formerly 42 U.S.C. § 1973gg–5 ), also offer voter registration services. Those agencies include departments of social services and public health, disability services agencies (vocational rehabilitation offices, departments of services for the blind and hard of hearing, and mental health departments), the North Carolina Employment Security Commission, and, for those engaged in a DMV transaction (including acquiring a no-fee voter ID), any DMV office, pursuant to N.C. Gen. Stat. § 163–82.19. (Doc. 126–1 at 4.) The League Plaintiffs acknowledged that these other avenues mean that "many people who are of lower socioeconomic status have an opportunity to register to vote elsewhere." (Doc. 167 at 135–36.)

103 Cf.Miss. State Chapter, Operation Push v. Allain, 674 F.Supp. 1245, 1250–52 (N.D.Miss.1987) (describing Mississippi law that initially prevented all registration outside of the office of the county registrar).

In addition, State law permits any individual, group, or organization—such as the GOTV efforts conducted by some Plaintiffs—to conduct a voter registration drive, without any special training, pursuant to SBOE-published guidelines and with materials the SBOE and CBOEs provide. (Doc. 126–1 at 4.)¹⁰⁴

104 And, as noted, even after SL 2013–381, a voter who has moved within the county can still update his or her registration during early voting or on Election Day (i.e., after the 25–day registration cut-off) and vote. N.C. Gen. Stat. § 163–82.6A(e).

Plaintiffs argue that these other methods of registration are not a substitute for SDR and its in-person effectiveness. (Doc. 346 at 74–76.) DMV only offers registration services to those seeking DMV services, and it is true that the poor (which African Americans represent disproportionately) are less likely to use DMV services, as they are less likely to drive or own a vehicle. Also, because of a foul-up at DMV in implementing SL 2013–381, in September 2014 some 2,726 seventeen-year-olds were denied the right to register, and the SBOE had to send them a letter with a voter registration form and the promise to file it for them if completed and returned. (Pl. Ex. 726.) As for public assistance, Plaintiffs argue that it, too, is an insufficient substitute for the removal of SDR because it is only offered for those applying for such services and that public assistance registrations declined from 41,162 in 2012 to 13,340 in 2014. (Pl. Ex. 725 at 4.) While Plaintiffs' statistical evidence correlated African Americans disproportionally with the purpose of such services, suggesting it as an ideal registration opportunity, there was no direct evidence as to why registrations at such services did not occur more frequently. As Defendants pointed out, the reduction in use of these sources may very well be due, in some significant measure, to the fact that voter registrations have been offered to millions of Americans under the Affordable Care Act, passed in 2010.

Aside from prior use data, Plaintiffs seek to use data from the 2014 election to bolster their claim that African Americans have been disparately impacted by the elimination of SDR.

In his 2015 report, based on data provided by the SBOE, Dr. Stewart notes that in the 2014 midterm 12,983 people registered to vote after the registration deadline but before Election Day. (Pl. Ex. 242 at 163.)¹⁰⁵ Dr. Stewart did not know whether the dates reported reflected when the registration applications were signed or when the SBOE processed them. (Doc. 332 at 122.) He noted that, overall, 273 people registered during the seven days of early voting eliminated by SL 2013–381. (Pl. Ex. 242 at 163.) These registrants were more likely to be white than African American. (Id. at 164.) He also noted that 11,993 people registered to vote during the ten-day early-voting period. (Id. at 163.) However, Dr. Stewart did not remove from this figure (or from any of the above figures) those who registered at locations other than an early-voting site, (Doc. 332 at 123), even though only those registering at an early-voting site can use SDR. Nor did Dr. Stewart remove voters who were registering for a future election (thus ineligible to vote in the upcoming election) because they had not resided in their precinct for more than thirty days. SeeN.C. Gen. Stat. § 163–55(a) (requiring that a voter have "resided ... in the precinct in which the person offers to vote for 30 days next preceding an election"). Therefore, Dr. Stewart's figure likely exceeds the number of potential same-day registrants from 2014, but the court has no way of knowing the extent.

105 Among these registrants, 374 registered after the deadline but before either the old or new early-voting period began. (Pl. Ex. 242 at 163.) An additional 343 registered after early voting ended but before Election Day arrived. (Id.) Such registrations would not have enabled a citizen to vote before or after SL 2013–381 and are thus not relevant to the analysis.

Instead of providing the actual number of African American and white registrants included in the 11,993 subtotal (surely a knowable figure), Dr. Stewart provided a percentage (by race) of all registrations during the two-year period preceding the 2014 election that occurred on the indicated day. (Pl. Ex. 242 at 164.) Thus, he found that the number of registrations during early voting in 2014 was 1.415% of all registrations for the preceding two-year period. (Id. at 163.) He then broke this figure down by race, calculating that African American registrations during the 2014 early-voting period constituted 1.994% of the registrations for the two-year period. (Id. at 164.) The comparable white registrations were 1.800%.¹⁰⁶ (Id.) He noted the difference in numbers by race—0.194 percentage points—and concluded that this represented a disparity showing that African American citizens were disparately impacted by the unavailability of SDR in the 2014 general election. He acknowledged, however, that the disparity in this election was smaller than the disparity for previous elections. (Doc. 332 at 128.)

106 With these figures and others, Plaintiffs attempt to present the percentages to say that African Americans were 11% more likely to register during the early-voting period. However, as the Seventh Circuit has earlier explained, such mathematical manipulations conceal the true inquiry:

We have given the percentages of persons who have these documents. Plaintiffs express the figures differently, giving the percentages of persons who lack the documents (2.4% of whites, 4.5% of blacks, and 5.9% of Latinos), then dividing one percentage by another to yield an expression such as "registered Black voters in Wisconsin were 70% more likely than white voters to lack a driver's license or state ID." That is a misuse of data. Dividing one percentage by another produces a number of little relevance to the problem. If 99.9% of whites had photo IDs, and 99.7% of blacks did, the same approach would yield the statement "blacks are three times as likely as whites to lack qualifying ID" (0.3 ÷ 0.1 = 3), but such a statement would mask the fact that the populations were effectively identical. That's why we do not divide percentages.

Frank v. Walker, 768 F.3d 744, 752 n. 3 (7th Cir.2014).

Defendants have offered a different characterization of the 2014 data. Defendants would have the court first calculate the actual number of African American and white registrations during the 2014 early-voting period. (Doc. 332 at 124–28.) This yields approximately 2,714 African American registrations and 7,507 white registrations. (Pl. Ex. 242 at 163–64.) Of the 11,993 registrations during this period, the number of African American registrations thus constituted 22.630%. (Id.) This compares to the African American share of 22.5% of all voters registered as of 2014.¹⁰⁷ (Def. Ex. 309 at 76.) Thus, the proportion of African American registrants during the 2014 early-voting period is virtually identical to the proportion of African American registered voters as of 2014.

107 The proportion of the African American North Carolina citizen voting age population ("CVAP") in 2014 is not yet known. In 2013, however, African American North Carolinians were 21.88% of the North Carolina CVAP. (Def. Ex. 309 at 77.) This figure is also nearly identical to the proportion of African American registrations (22.630%) during the 2014 early-voting period.

Given the closeness of these percentages, and considering that the data include people registering at sites other than those offering early voting (and thus formerly offering SDR), these data are at best weak evidence that the elimination of SDR caused African Americans to be affected disproportionately.¹⁰⁸ This conclusion is bolstered by the fact that the African American advantage in registration rates grew from 2012 to 2014, and African American turnout rates increased from 2010 to 2014, all while SDR was eliminated.

108 There is a further concern about the reliability of these figures. The president of Plaintiff North Carolina Chapter of the NAACP, the Reverend Dr. William Barber II, gave a speech in October 2014, just before the general election, imploring audience members at the State Chapter's annual banquet to take people to the polls who were not registered and demand they be given a provisional ballot:

Also, if people did not get registered we want you to take them to the polls anyway. Federal law requires that they have to be given a provisional ballot. ... If they didn't get registered and can't register in early voting, we want you to take them to the polls, we want them to get the provisional ballot, we want them to be told the ballot would not be counted because they—we do not have early voting and same-day registration, and then we want you to get that name so next year, when we're in court, we can present a list of names of people who have been denied their right to vote ... because of the denial of same-day registration and early voting.

(Def. Ex. 67 at 37–38.) Dr. Barber testified at trial that he hoped those at the banquet followed his instructions. (Doc. 329 at 123–24.) Under federal and State law, voters who know they are not registered and not eligible to vote are not entitled to a provisional ballot. See52 U.S.C. § 21082(a) ; N.C. Gen. Stat. § 163–166.11. The NC NAACP later issued a press release modifying the request to urge only those who believe they are registered voters to request a ballot. (Def. Ex. 65 at 2.) In the end, this tactic may have increased the numbers of persons who tried to register during early voting as well as those who demanded provisional ballots and tried to vote OOP.

Turning to "young" voters, Plaintiffs presented the testimony of Dr. Levine, who relies in part on national studies of EDR and SDR but also fails to distinguish between the two. (Doc. 332 at 134.) As noted above, the two mechanisms have entirely different effects on political participation. Thus, the national research on which he relies is not sufficiently relevant to this case.

Dr. Levine also analyzed the use of SDR by young voters from 2008 to 2012. He presented the following use statistics, which are percentages of voters in a given election using SDR:

200820102012 Older voters 5.23% 1.97% 4.75% Young Voters 11.23% 9.21% 12.47% Total voters 5.49% 2.25% 5.45% using SDR

(Pl. Ex. 50A at 11–12.)¹⁰⁹ As noted in the early-voting discussion above, youth turnout and registration rates increased from 2010 to 2014. Dr. Levine did not produce any evidence of disparate impact other than his prior use numbers and his inapposite national studies.

109 In raw numbers, around four to five times more "older voters" than "young voters" used SDR in each of these general elections. (Pl. Ex. 50A at 11–12.)

Historically, the effect of a voting mechanism has been measured commonly by turnout. Prior to this case, Plaintiffs' experts measured SDR's effect that way. They further acknowledge that such analyses are possible here and would be probative. (See Def. Ex. 348 (Dr. Burden analyzing the effect of SDR on turnout); Doc. 332 at 151–52 (Dr. Stewart conceding that he has not conducted any statistical analysis on the effect of SDR on turnout or registration rates); Doc. 342 at 129–30 (Dr. Stewart conceding that a properly conducted cross-State analysis would be an appropriate way to measure the effect of a voting law on turnout, despite not having done so in this case); Doc. 331 at 88–90 (Dr. Burden conceding that he had done no analysis to determine whether competitiveness of 2014 and 2008 general elections affected turnout); id. at 96 (Dr. Burden conceding that, while he has conducted "many national analyses" of the effect on turnout from SDR and other election changes, he has never examined whether any election law affected turnout in North Carolina).) The failure of Plaintiffs' experts to conduct the kind of analyses that were possible, especially where it has been done in their scholarship under more rigorous standards, (e.g., Doc. 331 at 148–49 (Dr. Burden conceding that his North Carolina case study analysis in this case was less rigorous than his academic work)), impairs the persuasiveness of their testimony. SeeFed. Rule Evid. 702 advisory committee notes (counseling courts to be wary of an expert who is not "as careful as he would be in his regular professional work outside his paid litigation consulting") (quoting Sheehan v. Daily Racing Form, Inc., 104 F.3d 940, 942 (7th Cir.1997) ); see alsoKumho Tire Co. v. Carmichael, 526 U.S. 137, 152, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999) ("[The Daubert gatekeeping requirement] is to make certain that an expert, whether basing testimony upon professional studies or personal experience, employs in the courtroom the same level of intellectual rigor that characterizes the practice of an expert in the relevant field.").

There was only one reason given for Plaintiffs' failure to produce such highly probative evidence in this case. On cross-examination, Defendants had Dr. Stewart perform calculations from the 2014 general election data, and the results tended to negate an inference of disparity. (E.g., Doc. 332 at 95–101.) On redirect, Plaintiffs asked Dr. Stewart whether any of these calculations changed his opinions. He said no. When asked, "Why not?" he responded, "Because my conclusions were based on comparing the use of these—the three techniques [early voting, SDR, and OOP voting], [among] blacks and whites, and none of this changes the fact that African-Americans utilized the changed procedures more than whites did." (Doc. 332 at 152–53.)

In support of their claim that African Americans disproportionately need in-person assistance, Plaintiffs cited statistics from the SBOE's database of incomplete registration applications. For those individuals who submitted voter registration applications between 2012 and 2014, 21.25% (136,113) were African American, while 65.12% (417,053) were white. (Pl. Ex. 242 at 163–64 (stating that the total number of registrations across the two-year election cycle was 640,417).) As of November 2014, 34.74% of the registration applications in the incomplete queue were submitted by African Americans, while 51.53% were submitted by white applicants. (Pl. Ex. 633 at 5.) In addition, 33.40% of applicants placed in the incomplete queue for failure to check the citizenship box were African American, while 28.86% were white. (Id.) Of those submitting applications without a birth date, 59.05% were African American, while 22.28% were white. (Id.)

In addition, Plaintiffs presented evidence that African Americans are more likely to move between counties than white residents. (Pl. Ex. 42 at 30 n.37.) Because North Carolina organizes registration at the county level, more action is required by those who move between counties. For example, a voter who has moved within his county may update his registration information, including a change of address, during early voting or on Election Day, and vote. SeeN.C. Gen. Stat. § 163–82.6A(e) ; (Doc. 126–1 at 5). By contrast, individuals who move between counties more than thirty days before Election Day must, in the absence of SDR, re-register in their new county prior to the twenty-five day cut-off in order to be eligible to vote. (Doc. 354 at 97–98.) Accordingly, because they are more likely to move between counties, African Americans are more likely to need to re-register.

In sum, Plaintiffs staked their case largely on aggregate disproportionate SDR use over six years.¹¹⁰ While in some cases data on differential use are all that may be available, in this case there have already been three elections without SDR. Thus, the prior differential in use is not the only, or most probative, evidence. As with the change in the early-voting schedule, the 2014 election turnout data casts doubt on Plaintiffs' claims. While effectively relaxing a registration deadline may make registration marginally easier, there is simply no persuasive evidence in this case that adding SDR caused an increase (that would not have occurred otherwise) in turnout or registration rates overall or for any subgroup, just as there is no evidence that removing SDR decreased turnout or registration rates. This conclusion is supported by Plaintiffs' own academic literature on SDR. Finally, the evidence does not support the conclusion that African Americans, Hispanics,¹¹¹ or "young" voters were habituated to using SDR during the three general elections it was available. Plaintiffs did demonstrate, however, that African Americans are more likely to move between counties, and thus more likely to need to re-register, and that some small subset are more likely to end up in the incomplete registration queue.

110 As a three-judge panel in a preclearance case, applying the VRA's § 5 retrogression standard with the burden on the State, has explained,

[A] change is not retrogressive simply because it deals with a method of voting or registration that minorities use more frequently, or even because it renders that method marginally more difficult or burdensome. Rather, to be retrogressive, a ballot access change must be sufficiently burdensome that it will likely cause some reasonable minority voters not to register to vote, not to go to the polls, or not to be able to cast an effective ballot once they get to the polls.

Florida v. United States, 885 F.Supp.2d 299, 312 (D.D.C.2012).

111 Plaintiffs produced evidence that Hispanics disproportionately used SDR, but the evidence presented was much more limited than that for African Americans. Plaintiffs' expert Allan Lichtman, Ph.D., testified as to some disparate use numbers regarding Hispanics, (Doc. 333 at 118), and included some use statistics in a rebuttal report, (Pl. Ex. 245 at 23). He did not include any data from 2014. (Id.) By contrast, Defendants provided evidence that Hispanic turnout increased from 2010 to 2014. (See, e.g., Def. Ex. 309 at 62–67, 76–77.)

4. Elimination of OOP Provisional Voting

From 2005 until 2013, OOP voting allowed a registered voter on Election Day to vote in a precinct other than his assigned precinct, so long as he was still voting in his county of residence. OOP voting only applied to Election Day because early voting is offered at centralized sites and not at precincts. In that sense, early voting continues to permit OOP voting for the ten days of early voting.

OOP voting required the poll worker to give the wrong-precinct voter a provisional ballot to cast. However, because the provisional ballot was for a different (wrong) precinct, it often included races for which the voter was not eligible to vote and omitted races for which he was. This contrasts with voters during early voting, where computer voting machines can be programmed to produce the correct ballot even if the centralized location for early voting is a different precinct. Some Election Day precincts do not use electronic voting machines, however. Consequently, after the OOP ballot was cast, the CBOE had to individually review the full ballot to tabulate and record the races for which the voter was eligible to vote and invalidate votes cast in the races for which the voter was not eligible, or the CBOE had to transcribe the appropriate eligible votes onto a separate, proper ballot, which would then be recorded. Therefore, except for national and State-wide races, OOP voters were often effectively disenfranchised for some races. In this regard, to the extent OOP voting included voters who failed to vote in their proper precinct merely by neglect or ignorance (and there was evidence of this at trial), and not due to need, it can be criticized as having encouraged voters to unwittingly forego their full voting rights.

Some form of OOP voting is permitted in sixteen States plus the District of Columbia. (Def. Ex. 270 at 18.) North Carolina is among thirty other States, including many jurisdictions previously covered by VRA § 5, that prohibit the practice removed by SL 2013–381.¹¹² (Def. Ex. 270 at 18.) Two more States, New York and Missouri, count OOP ballots only if cast in the correct polling place (the rare occurrence where more than one precinct uses the same polling place). (Id. at 18.) Two others, Connecticut and Massachusetts, will count OOP ballots if cast in the correct town or city, which frequently is one and the same. (Id.) Even among States that permit OOP voting, there are various limitations. For example, Louisiana counts votes cast in the wrong precinct, but only for federal races. (Id. at 18.)

112 At trial Mr. Trende testified that Utah does not permit OOP voting. However, it appears that Utah will count ballots cast in the wrong precinct so long as they are cast in the correct county. Utah Code Ann. § 20A–4–107(1)(b)(iii), (2)(c) ; (Def. Ex. 2 (Trende's initial report: coding Utah as offering OOP)). However, it went undisputed at trial that the majority of States, many of which were previously covered by § 5, do not offer OOP. See, e.g., Ala. Code §§ 17–9–10, 17–10–2(b)(2) ; Fla. Stat. Ann. § 101.048(2)(b) ; Miss. Code Ann. § 23–15–571(3)(a), (d) ; S.C. Code Ann. § 7–13–830 ; S.D. Codified Laws § 12–20–5.1 ; Tex. Elec. Code Ann. § 63.011(a) ; Va. Code Ann. § 24.2–653(B).

The history of North Carolina's use of OOP voting is measured by the casting of provisional ballots that the SBOE labels as "incorrect precinct." (See, e.g., Pl. Ex. 689.) The counting of OOP provisional ballots began in 2006 and continued for three more general elections, until it was ended by SL 2013–381. Recognizing that every vote is important, it is nevertheless true that even for the years it was in place, OOP ballots constituted only a fraction of a sliver of the total ballots cast. In 2012, when there were more OOP provisional ballots at least partially counted than ever, the total provisional incorrect precinct ballots as a percent of all ballots cast amounted to only .19% of white ballots and .33% of African American ballots.¹¹³ (Pl. Ex. 42 at 98 (tbl. 14)¹¹⁴ ; Pl. Ex. 242 at 161 (App'x U).) Put another way, 99.67% of African American voters and 99.81% of white voters were able to cast ballots other than by OOP. Nevertheless, compared to their share of the electorate, African American voters were disproportionately more likely than whites to cast an OOP provisional ballot in the elections prior to SL 2013–381. (Pl. Ex. 42 (Ex. 49); id. at 98 (tbl. 14).)

113 For every year that OOP voting was permitted, white registered voters cast more incorrect precinct ballots than African American voters. (Pl. Ex. 42 (Ex. 49).) As with earlier analyses, the court does not consider this fact.

114 Because Plaintiffs' data on the number of OOP provisional ballots cast excluded 35.4% of the records in the provisional ballot file (the race of the voter was not indicated), the court followed Plaintiffs' instruction and multiplied each number in Table 14 by 1.55%. (Pl. Ex. 42 at 98 n.126 ("[T]he proper correction to apply is to multiply each number by 1/.646, or 1.55.").)

In 2014, after SL 2013–381, North Carolina stopped counting "incorrect precinct" provisional ballots, though it continued to offer provisional ballots to registered voters, as required by HAVA. In the 2014 general election, the number of "incorrect precinct" provisional ballots cast on Election Day dropped to 1,930. (Pl. Ex. 689.) For whatever reason, many of these were counted in whole or in part, leaving only 1,387 not counted. (Id.) Of these, 576 were cast by African American voters and 595 were cast by white voters. (Id.) Accordingly, African American voters disproportionately cast provisional ballots in the wrong precinct in 2014.

Plaintiffs and Defendants dispute the reason why the number of OOP provisional ballots dropped in 2014. Plaintiffs claim that voters were deterred, while Defendants claim the results show that former OOP voters were able to adjust to voting in their correct precincts. The explanations are not mutually exclusive, but Defendants' explanation is better supported by the evidence. In fact, Plaintiffs' own evidence indicates that the drop in OOP provisional ballots observed in 2014 was likely due in large part to voters going to their correct precinct in light of the elimination of OOP voting. For example, Plaintiffs presented the testimony of Susan Schaffer to support their assertion that requiring people to vote in the correct precinct results in a significant number of people being deterred from voting. (Pl. Ex. 796.) In 2014, Ms. Schaffer, who had moved to North Carolina from New Jersey in 2011, served as a poll observer for the advocacy group Democracy North Carolina during the general election on Election Day. (Id. at 12–15.) Democracy North Carolina was trying to measure the impact of SL 2013–381 on the 2014 election. (Id. at 16.) Ms. Schaffer began collecting the names of those who presented but left the precinct because they were assigned to vote in another precinct and reported fifty-nine persons. (Id. at 17–21.) Poll workers had advised these individuals of the location of their assigned precinct. (Id. at 20–21.) Plaintiffs argued that this evidenced the burden of eliminating OOP voting.

At trial, however, Defendants presented persuasive evidence as to what happened to these voters. Of the fifty-nine people on Ms. Schaffer's original list, only fifty-two could be identified as registered in North Carolina's voter database.¹¹⁵ (Def. Ex. 343 at 1.) Of these fifty-two voters, forty-nine did ultimately vote on Election Day. (Id.) That is, 94.2% of identified registered voters were able to travel from the wrong precinct to their assigned one on Election Day. There were three voters on Ms. Schaffer's list who did not vote in 2014. Two were white, apparently a married couple. (Id. at 140–43.) It is unknown why they never successfully voted. The third was an African American man. Based on his November 4, 2014 registration date (Election Day), it does not appear that he would have been eligible to vote in any precinct on Election Day, whether before or after SL 2013–381. (Id. at 132–34.)

115 To the extent that the seven individuals could not be identified because they were not registered, SDR could not have saved them because they were Election Day voters, and SDR has never been offered on Election Day in North Carolina (that would make it EDR).

This sampling of "affected voters" offered by Plaintiffs' own witness supports the court's conclusion that the number of wrong-precinct provisional ballots fell substantially in 2014 because of the ability of wrong-precinct voters to actually go to their correct precinct.

As with the data relating to other voting procedures, the provisional ballot data leaves many questions unanswered. The court could assume that the lack of OOP voting in 2014 disparately affected African American voters because they previously used OOP voting disproportionately. Defendants, however, have offered evidence to suggest that this may not be the case and that the provisional ballot data are missing context. For example, Dr. Thornton found that, among all OOP voters, 45% had voted in the correct precinct in the past. (Def. Ex. 309 at 89.) African Americans were even more likely, at 49%, to have voted at their assigned precinct in the past. (Id.) And, looking only at the 2014 general election, overall and among African American OOP voters, 74% had voted at their correct precinct in the past. (Id.) Further, the provisional data does not explain why any voter voted out of his or her precinct. Plaintiffs assert they did so because of need, but this conclusion is undermined by the fact that so few used the option in relation to the vastly larger number of poor, less educated, and vehicle-less voters. Nor does the use data tell us whether any voter could have easily voted in his or her correct precinct.

Thomas Hofeller, Ph.D., an expert who studies mapping for redistricting,¹¹⁶ observed the data for OOP voters and measured the distance between each OOP voter's correct precinct and the precinct at which he or she actually voted. He found that, in 2012, 60.3% of African American OOP voters voted at a precinct within five miles of their assigned precinct. (Def. Ex. 212A at 18.) That same year, only 49.1% of white voters cast an OOP ballot within five miles of their assigned precinct. (Id.) However, this data is of marginal value, as it does not address the distance from the voter's home or work.

116 Dr. Hofeller was proffered without objection "as an expert in demography, census geography, and database building involving voter registration and turnout information, voting and registration patterns and also the same patterns based upon race and partisan affiliation." (Doc. 340 at 140.)

Finally, Plaintiffs offered the testimony of voters, African American and white, who expressed frustration that they had cast an OOP provisional ballot in 2014 that was not counted. However, the vast majority had made no effort whatsoever to determine the location of their assigned precinct. (E.g., Doc. 330 at 40 (William Kittrell, college student: "I figured since I was a resident of North Carolina and I had registered to vote in North Carolina, that I would be able to vote in any county that I was in."); id. at 175–76 (Terrilin Cunningham: assumed she could vote anywhere in the county because in a previous county "I just voted down the street from our church, which we didn't live anywhere near"); Doc. 334 at 154–57 (Michael Owens: elected not to vote early but tried to vote OOP on Election Day at the precinct where he had previously voted and was unaware he must vote in his home precinct). Only Timothy and Yvonne Washington, a married couple, were unable to walk to their assigned precinct, which was farther away, due to disabilities. (See Pl. Ex. 679; Pl. Ex. 797); see also infra Part II.A.3.d. Thus, it is far from clear, indeed doubtful, whether the elimination of OOP voting was the cause of most voters' failure to successfully cast a ballot.

In sum, Plaintiffs have shown that African Americans disproportionately cast OOP provisional ballots both before and after SL 2013–381.¹¹⁷ There are substantial questions about the reason for the disparate use, which will be addressed more thoroughly infra.

117 Hispanics were also more likely to cast OOP provisional ballots prior to SL 2013–381. (Pl. Ex. 245 at 24). The OOP provisional use data for Hispanics did not cover 2014. (Seeid.) In the 2008, 2010, and 2012 general elections combined, young voters cast 919,246 total ballots, of which 3,221 (0.35%) were OOP ballots. (SeePl. Ex. 236 at 30–32, 35–38.) By comparison, older voters cast 10,651,288 ballots in those elections, of which 14,697 (0.138%) were OOP ballots. (Seeid.) In the 2014 general election, after SL 2013–381, young voters cast 102,775 votes, of which 196 (0.191%) were OOP ballots. (Seeid.) Older voters cast 2,737,540 votes in that election, of which 1,459 (0.053%) were OOP ballots. (Seeid.)

5. Elimination of Pre-Registration

Pre-registration permitted those under the age of eighteen to register earlier than would otherwise be permitted. Currently, seventeen-year-olds who will be eighteen by the time of the general election are able to register sixty days prior to the accompanying primary. N.C. Gen. Stat. § 163–59. Plaintiffs argue that the elimination of pre-registration disparately impacts African American and Hispanic youth and imposes a severe burden on all youth.

Eight States and the District of Columbia allow pre-registration by those age sixteen or older. (Def. Ex. 270 at 32; Doc. 340 at 18.) North Dakota is included as offering pre-registration because it has no registration system at all. (Def. Ex. 270 at 32.)

North Carolina's pre-registration law was in place for only two general elections, 2010 and 2012, before it was repealed by SL 2013–381. (Pl. Ex. 235 at 18.) When available, about 152,000 adolescents pre-registered, (id. at 13), although it is unknown, of course, how many of these adolescents would have eventually registered without pre-registration. When in place, pre-registration did not clearly benefit either Democrats or Republicans. (Id. at 14.) In some years, Republicans had more pre-registrants than Democrats, and in other years the reverse was true. (Id.) In all years, however, the number of unaffiliated pre-registrants was greatest. (Id.) In 2010, 23% of those who pre-registered were African American, and in 2012, 30% were African American. (Id. at 16.) Thus, overall, African Americans disproportionately pre-registered. Hispanics, by contrast, were 3.8% of pre-registered voters, which is less than their proportion in the population. (Pl. Ex. 245 at 23.)

One of Plaintiffs' experts, Sunshine Hillygus, Ph.D., Professor of Political Science at Duke University,¹¹⁸ studied the effect of pre-registration on young voter patterns nationally as well as the use of pre-registration in Florida.¹¹⁹ Based on a difference-in-difference model¹²⁰ using responses from the CPS, she found that States with pre-registration experienced a 13% increase in youth turnout. (Pl. Ex. 235 at 20.) A lagged model¹²¹ of the same data showed a 2% increase in youth turnout. (Id. at 21.) In addition to her difference-in-difference model,¹²² Dr. Hillygus examined the effect of pre-registration in Florida by comparing the voting patterns of "those marginally eligible to vote in the 2008 election (17 turning 18 by November 4, 2008) and those marginally ineligible" (17 not turning 18 by November 4, 2008). (Id. at 22.) The key difference between these two groups was that those who were marginally ineligible only had the opportunity to pre-register in the lead up to the 2008 election, while those who were marginally eligible had the opportunity to register traditionally and vote.¹²³ (Id.) She found that those who had the opportunity to pre-register were 8% more likely to vote in 2012 than those who had the opportunity to register traditionally. (Id. at 22–23.) Based on these studies of areas outside of North Carolina, she concluded that pre-registration helps increase young voter turnout. (Id. at 25.) Extrapolating her difference-in-difference and lagged models to North Carolina, she further projected that the removal of pre-registration would result in 8,000 to 50,000 fewer young voters in North Carolina in 2016. (Id. at 28.) She did not, however, find that pre-registration disproportionately benefits any particular race. Instead, she found that it is "equally effective for various demographic groups, including whites versus minorities." (Id. at 24.)

118 Dr. Hillygus was proffered without objection as an expert in American political behavior, including political behavior in connection with pre-registration laws and survey methods. (Doc. 331 at 188.)

119 Yet another example of Plaintiffs' experts relying on practices of other States to opine on effects in North Carolina.

120 A difference-in-difference model attempts to control for (hold constant) potential confounders that vary between States across time and across elections. (Doc. 331 at 192–93.)

121 A lagged model attempts to determine a lower bound and provide a more conservative estimate using stricter assumptions; Dr. Hillygus deemed it appropriate if one were to assume that States with already high youth voter turnout were more likely to adopt pre-registration laws. (Doc. 331 at 195.)

122 Because Dr. Hillygus's difference-in-difference model depends upon CPS data, it is at least in part vulnerable to the over-reporting problems that are well established to exist within the CPS survey. (Doc. 331 at 135–36.) However, this concern is ameliorated by Dr. Hillygus's Florida study, which was based upon voter files and is consistent with the results of the difference-in-difference model. (Pl. Ex. 235 at 23.)

123 By saying an individual had the "opportunity to pre-register," Dr. Hillygus appears to mean that, due to the individual's birthday, pre-registration was the only available means of registration for that individual leading up the 2008 presidential election. (Pl. Ex. 235 at 22.) For example, Dr. Hillygus includes within the "opportunity to preregister" group individuals who were eligible to pre-register during the 2008 election but "waited until they were older to register traditionally," and includes within the "opportunity to register traditionally" group those who "pre-registered when they were younger, outside the context of a presidential election." (Id.) She used a "fuzzy regression discontinuity approach" to account for this factor (Id.)

At trial, Plaintiffs featured Nadia Cohen, who could have pre-registered but was unable to do so because of SL 2013–381. Even without pre-registration, Ms. Cohen could have registered long before the 2014 general election, given that she turned eighteen before that election. She did not register for the 2014 general election because no one told her to do so and because, in her words, "honestly, voting is not my top priority throughout the year." (Doc. 331 at 173.) At the time of trial she was enrolled to start college at the University of North Carolina at Chapel Hill in the fall of 2015. The elimination of pre-registration did not cause Ms. Cohen to be unable to vote in 2014.

In addition, Plaintiffs presented the testimony of individuals involved in assisting persons to register to vote. For example, Reverend Maria Palmer, mentioned above, is an Hispanic-American member of "Hotties,"—short for "Hispanic Outreach Team"—which assists Hispanics in voting. (Doc. 329 at 145.) She testified that the elimination of pre-registration was "big" because it affected Latino youth, who are often a family's first-generation voter (presumably because they are the first generation eligible to vote) and "not familiar with how things are done." (Doc. 329 at 148.)

There is no evidence that a voter can be habituated to pre-registration, since it is usually a one-time event. Moreover, because a segment of pre-registrants are mobile, they have an increased likelihood of facing an additional barrier of re-registering upon turning eighteen. That is either because the pre-registrant and his family change addresses between the time of pre-registration and age of majority (in which case he fails statutory mail verification, which is not initiated until the pre-registrant is eligible to vote) or because the pre-registrant subsequently moves out of the county (e.g., by going to college). In either case, the pre-registrant who moves will need to re-register in his new county of residence.

In sum, the evidence shows that pre-registration increases youth turnout. However, although African Americans used pre-registration disproportionately compared to whites in North Carolina, the evidence also establishes that pre-registration does not disproportionately benefit one race over the other. (Pl. Ex. 235 at 24.) In addition, while the evidence explains why pre-registration increases turnout, it does not explain why African Americans are more likely to pre-register or why the other means of registration are less available to African Americans than other groups. As noted below, North Carolina continues to offer substantial opportunity for seventeen-year-olds to register ahead of any election for which they are eligible, including primaries when they are age seventeen.

6. Other Challenged Provisions

In addition to these four mechanisms, Plaintiffs also challenge SL 2013–381's expansion of poll observers and ballot challenges, as well as the transfer of discretion to extend Election Day poll hours from CBOEs to the SBOE, as either disproportionately impacting minorities or imposing an otherwise unjustified burden on the right to vote.

North Carolina law permits the chair of each political party in every county to "designate two observers to attend each voting place at each primary and election." N.C. Gen. Stat. § 163–45(a). SL 2013–381 allows the chair of each county party to "designate 10 additional at-large observers who are residents of that county who may attend any voting place in that county." 2013 N.C. Sess. Law 381, § 11.1 (codified at N.C. Gen. Stat. § 163–45(a) ). "Not more than two observers from the same political party shall be permitted in the voting enclosure at any time, except that in addition one of the at-large observers from each party may also be in the voting enclosure." Id. The list of at-large observers must be "provided by the county director of elections to the chief judge [for each affected precinct]." Id.(codified at § 163–45(b) ). In conjunction with the addition of at-large observers, the law now permits any registered voter in the county, rather than in the precinct, to exercise the right to challenge a ballot on Election Day. Id.§ 20.2 (codified at N.C. Gen. Stat. § 163–87 ). During early voting, any resident of the State may now file a challenge. Id.§ 20.1 (codified at N.C. Gen. Stat. § 163–84 ). Little, if anything, was said about the justification for these procedures during the legislative process. (See Pl. Ex. 202; Pl. Ex. 549; Pl. Ex. 550.)

Under North Carolina law, the polls on Election Day are to remain open from 6:30 a.m. until 7:30 p.m. N.C. Gen. Stat. § 163–166.01. Beginning in 2001, each CBOE had the power to "direct that the polls remain open until 8:30 p.m." in "extraordinary circumstances." 2001 N.C. Sess. Law 460, § 3 (codified at N.C. Gen. Stat. § 163–166 (2002) ). SL 2013–381 eliminates the discretion of the CBOEs by deleting the "extraordinary circumstances" clause. 2013 N.C. Sess. Law 381, § 33.1. The law now provides:

If the polls are delayed in opening for more than 15 minutes, or are interrupted for more than 15 minutes after opening, the [SBOE] may extend the closing time by an equal number of minutes. As authorized by law, the [SBOE] shall be available either in person or by teleconference on the day of election to approve any such extension.

N.C. Gen. Stat. § 163–166.01. The law thus vests discretion in the SBOE to the exclusion of CBOEs and conditions the exercise of discretion on a delay of fifteen minutes or longer.

Virtually no evidence was offered at trial by either side as to these challenged provisions. At the preliminary injunction hearing, Plaintiffs offered limited testimony about a concern that additional observers might be a basis for intimidation. For example, Senator Dan Blue testified about a concern that African American voters may be intimidated by the presence of a white observer who does not look familiar to them and that bringing in people from outside the precinct may create an intimidating environment. (Doc. 164 at 109–11.) But as he stated, individuals have a First Amendment right to stand outside the polling place, and SL 2013–381 does not address this. (Id. at 108.) Moreover, the intimidation he was most concerned with, he said, occurs outside the polling place, not inside the restricted area where observers from both parties would be present and limited under SL 2013–381. (Id. at 136–37.)

Defendants did not present evidence as to the justification for poll observers and challengers,¹²⁴ and the need is not readily apparent. By the same token, however, Plaintiffs have not offered any persuasive evidence that SL 2013–381 renders any intimidation likely. Plaintiffs apparently assume that white poll observers would be placed in African American precincts; but there was no indication that would be so. The law requires that "[a]n observer shall do no electioneering at the voting place, and shall in no manner impede the voting process or interfere or communicate with or observe any voter in casting a ballot," unless the chief judge of elections permits the observer to make observations and take notes. N.C. Gen. Stat. § 163–45(c). Plaintiffs have provided no evidence that any poll observer or any challenger has abused his or her statutory power during the several elections held under SL 2013–381, nor have they forecast any evidence that such is likely in the future. Of course, if any problem ever develops, the participants would be subject to appropriate legal action.

124 Plaintiffs criticize proponents of SL 2013–381 for not testifying at all at trial and invite the court to take note of their absence. However, members of the General Assembly enjoy a qualified legislative privilege. The parties engaged in significant proceedings during this litigation over the scope of that privilege, and the court allowed certain discovery to take place. (Doc. 93.) The court will note any absence of justification when the legislation was debated and passed, but it will not infer improper motive merely from legislators' legitimate reliance on privilege.

With respect to the discretion to extend polling hours, Plaintiffs have not demonstrated how the elimination of the "extraordinary circumstances" clause will cause any burden or lessened opportunity based on race or age. This is especially true because, as former SBOE Director Gary Bartlett testified at the preliminary injunction hearing, the SBOE retains the ability to make up significant losses in time by ordering the polls to remain open for equal time in the event of a delay.¹²⁵ N.C. Gen. Stat. § 163–166.01. The law also protects any voter in line to vote at the time the polls close. Id. On these provisions, therefore, Plaintiffs have not demonstrated any burden on any voter or any disproportionate impact on minorities.

125 According to former Director Bartlett, this SBOE discretion to keep the polls open satisfied the concern he had about the removal of discretion from CBOEs. (Doc. 160–3 at 151.)

7. 2014 Data

The data from the 2014 elections seriously undermine Plaintiffs' claims in these cases. Plaintiffs argue that such data are little or no evidence of the actual effect of eliminating the voting procedures at issue. Their arguments are simply not persuasive. Data from actual implementation of an election law are precisely the sort of electoral information that courts are encouraged to consider, because they permit an understanding of the effect of the law based on "historical facts rather than speculation." Purcell v. Gonzalez, 549 U.S. 1, 6, 127 S.Ct. 5, 166 L.Ed.2d 1 (2006) (Stevens, J., concurring).

Plaintiffs argue that the court cannot merely compare "aggregate turnout numbers" from 2014 to 2010 to determine the effects of SL 2013–381. (Doc. 346 at 104.) True. But Defendants have not asked the court to take such a myopic view, and this court is examining so much more information to reach its conclusions.

Next, Plaintiffs argue that the 2014 election data were affected by a number of different factors besides SL 2013–381.

First, they argue that 2014 involved a very competitive $110 million North Carolina Senate race, representing the highest spending of its type in the history of the Nation. (Id. at 104–05.) This too is true. Oddly, it came to light at trial that one of Plaintiffs' experts, Morgan Kousser, Ph.D., a California resident, contributed to this anomaly by making a personal campaign contribution to the incumbent Democratic Senator for North Carolina in 2014 even though he was supposedly a neutral expert in this case. (Doc. 330 at 85–86.) Dr. Kousser is free to express his political views, but doing so while claiming to be an unbiased expert affects his credibility. More importantly, if Plaintiffs believed that the competitiveness of that campaign distorted the turnout numbers, they could have had their experts demonstrate this quantitatively. Plaintiffs' experts have done this type of analysis in the past, yet no effort was made to do so here. (See Def. Ex. 348; Doc. 342 at 129–30; Doc. 331 at 88–90.) In addition, as noted, Plaintiffs' attempt to rely on an extraordinary Senate campaign to downplay high turnout in 2014 is contradicted by their attempt to ignore the unique nature of the 2008 and 2012 presidential elections (whose results benefit them).

Second, Plaintiffs argue that the contest for the seat of Congressional District 12, which had been held by Congressman Melvin Watt (an African American) skewed the turnout statistics. But in the general election, this race could not be deemed competitive in any sense of the word, with the Democratic candidate taking over 75% of the vote. (See Def. Ex. 364 at 4.)

Third, Plaintiffs point to their own "engagement efforts" and African American "anger about the enactment of HB 589" to increase turnout as skewing the 2014 numbers. (Doc. 346 at 105.) Again, no actual evidence of the effect of the purported skewing was offered, even though Plaintiffs' experts purport to measure those kinds of things quantitatively. Moreover, if "engaging" "angry" voters meaningfully moves turnout, then turnout is more likely a function of motivation than the availability of the voting mechanisms at issue here.

Finally, Plaintiffs argue that, if 2014 turnout data has any relevance, "the relevant comparison is between actual 2014 turnout and what turnout in 2014 would have been had HB 589 not been in effect." (Id. at 106.) This would be most probative. But Plaintiffs then fault Defendants' expert, Dr. Thornton, for not having performed such an analysis. (Id.) This would be more persuasive if this were a VRA § 5 case where Defendants bore the burden of proof, but it is not. Plaintiffs bear the burden in this trial.

In the end, Plaintiffs rely on aggregate turnout data when it is expedient, but eschew it when it is not. Plaintiffs are correct that the 2014 turnout data are not dispositive. But, as the Supreme Court has recognized, it is highly probative, and the court considers it along with all the other data offered into evidence in assessing the totality of the circumstances.

E. Testimony of Other Experts

Plaintiffs introduced the testimony of various other experts who discussed the history of official discrimination in North Carolina and various socioeconomic disparities. The court's findings as to their opinions and conclusions will be discussed infra where appropriate.

II. CONCLUSIONS OF LAW

A. Section 2 of the VRA

1. The Law of Vote Denial and Abridgement Claims

The right to vote is fundamental and preservative of all others. League, 769 F.3d at 229. Section 2 of the VRA prohibits any "voting qualification or prerequisite to voting or standard, practice, or procedure" that "results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color." 52 U.S.C. § 10301(a). A violation of § 2

is established if, based on the totality of circumstances, it is shown that the political processes leading to nomination or election in the State or political subdivision are not equally open to participation by members of a class of citizens protected by subsection (a) in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice. The extent to which members of a protected class have been elected to office in the State or political subdivision is one circumstance which may be considered: Provided, That nothing in this section establishes a right to have members of a protected class elected in numbers equal to their proportion in the population.

Id.§ 10301(b). Although the Supreme Court held in 1980 that a § 2 plaintiff had to show discriminatory intent, City of Mobile v. Bolden, 446 U.S. 55, 100 S.Ct. 1519, 64 L.Ed.2d 47 (1980), Congress amended the VRA in 1982 such that a § 2 plaintiff need only show that a particular voting practice produces a discriminatory result. 52 U.S.C. § 10301 (formerly 42 U.S.C. § 1973(a) ); seeThornburg v. Gingles, 478 U.S. 30, 35, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986) ("Congress substantially revised § 2 to make clear that a violation could be proved by showing discriminatory effect alone and to establish as the relevant legal standard the 'results test,' applied by this Court in White v. Regester, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973), and by other federal courts before Bolden.")

"The essence of a § 2 claim is that a certain electoral law, practice, or structure interacts with social and historical conditions to cause an inequality in the opportunities enjoyed by black and white voters to elect their preferred representatives." Gingles, 478 U.S. at 47, 106 S.Ct. 2752. Section 2 encompasses claims of vote dilution, as well as vote denial or abridgment. See52 U.S.C. § 10301(a). Vote dilution refers to a governmental practice that dilutes the voting strength of minorities in various ways, including, for example, "the dispersal of blacks into districts in which they constitute an ineffective minority of voters or from the concentration of blacks into districts where they constitute an excessive majority." Gingles, 478 U.S. at 46 n. 11, 106 S.Ct. 2752. The bulk of the § 2 case law involves vote dilution claims.

This case, however, involves a vote denial/abridgement claim, for which there is less developed law under § 2. SeeSimmons v. Galvin, 575 F.3d 24, 28 n. 2, 41–42 (1st Cir.2009). This is likely because, historically, many election changes in covered jurisdictions like North Carolina had to obtain preclearance under § 5 of the VRA—under which the State bore the burden of proving non-retrogression—rendering any § 2 claim either moot or perhaps less attractive. It may also be because, to date, the Supreme Court has yet to entertain a § 2 vote denial case. See Nicholas O. Stephanopoulos, The South After Shelby County, 2013 Sup. Ct. Rev. 55, 106–07 (2013).

In Plaintiffs' appeal of the preliminary injunction decision in this case, the Fourth Circuit articulated the § 2 inquiry as follows:

First, the challenged standard, practice, or procedure must impose a discriminatory burden on members of a protected class, meaning that members of the protected class have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice.

Second, that burden must in part be caused by or linked to social and historical conditions that have or currently produce discrimination against members of the protected class.

League, 769 F.3d at 240 (citations and internal quotation marks omitted).¹²⁶ In assessing these two elements, the court considers "the totality of circumstances." Id. The factors typically relevant to the circumstances in a § 2 case are those adopted by the Supreme Court in Gingles from a Senate Report on the 1982 amendment to the VRA, which are:

126 In stating this test, the Fourth Circuit agreed with the Sixth Circuit's opinion in Ohio State Conference of NAACP v. Husted, 768 F.3d 524, 554 (6th Cir.2014). On the same day the Fourth Circuit issued its decision, however, the Sixth Circuit vacated its opinion after the Supreme Court stayed its implementation. No. 14–3877, 2014 WL 10384647, at *1 (6th Cir. Oct. 1, 2014). Thereafter, the Fifth Circuit also adopted the same test. Veasey v. Abbott, 796 F.3d 487, 504 (5th Cir.2015), rehearing en banc granted, 815 F.3d 958 (5th Cir.2016). The Seventh Circuit considered the test but cast doubt on it. Frank v. Walker, 768 F.3d 744, 754–55 (7th Cir.2014) ("We are skeptical about the second of these steps, because it does not distinguish discrimination by the defendants from other persons' discrimination. ... But if we were to adopt this approach for the sake of argument, our plaintiffs would fail at the first step ....").

1. "the extent of any history of official discrimination in the state or political subdivision that touched the right of the members of the minority group to register, to vote, or otherwise to participate in the democratic process";

2. "the extent to which voting in the elections of the state or political subdivision is racially polarized";

3. "the extent to which the state or political subdivision has used unusually large election districts, majority vote requirements, anti-single shot provisions, or other voting practices or procedures that may enhance the opportunity for discrimination against the minority group";

4. "if there is a candidate slating process, whether the members of the minority group have been denied access to that process";

5. "the extent to which members of the minority group in the state or political subdivision bear the effects of discrimination in such areas as education, employment and health, which hinder their ability to participate effectively in the political process";

6. "whether political campaigns have been characterized by overt or subtle racial appeals";

7. "the extent to which members of the minority group have been elected to public office in the jurisdiction";

8. "whether there is a significant lack of responsiveness on the part of elected officials to the particularized needs of the members of the minority group"; and

9. "whether the policy underlying the state or political subdivision's use of such voting qualification, prerequisite to voting, or standard, practice or procedure is tenuous."

Gingles, 478 U.S. at 36–37, 106 S.Ct. 2752 (quoting S. Rep. No. 97–417, at 28–29 (1982), reprinted in 1982 U.S.C.C.A.N. 177, 206–07 (hereinafter, "Senate Report")).

These factors are drawn from the vote dilution context, where they have more obvious application. SeeGingles, 478 U.S. at 36 n. 4, 45, 106 S.Ct. 2752. In both the vote dilution and vote denial context, "there is no requirement that any particular number of factors be proved, or [even] that a majority of them point one way or the other." League, 769 F.3d at 240 (quoting Gingles, 478 U.S. at 45, 106 S.Ct. 2752 )). "[T]his list of typical factors is neither comprehensive nor exclusive. While the enumerated factors will often be pertinent to certain types of § 2 violations, particularly to vote dilution claims, other factors may also be relevant and may be considered." Gingles, 478 U.S. at 45, 106 S.Ct. 2752 (footnote omitted). For example, in the interlocutory appeal of the denial of the preliminary injunction in this case, the Fourth Circuit stated that the "success[ ]" of North Carolina's "previous voting practices" in "fostering minority participation" is "centrally relevant" under § 2, being a "critical piece of the totality-of-the-circumstances analysis [§] 2 requires." League, 769 F.3d at 242. In addition, the Supreme Court has noted that "the State's interest in maintaining an electoral system" is "a legitimate factor to be considered by courts among the 'totality of circumstances' in determining whether a § 2 violation has occurred." Houston Lawyers' Ass'n v. Attorney Gen. of Tex., 501 U.S. 419, 426, 111 S.Ct. 2376, 115 L.Ed.2d 379 (1991) (vote dilution).¹²⁷

127 Legitimate interests served by a decision or rule have been found to be relevant in other statutory race discrimination claims. Cf.Tex. Dep't of Hous. & Cmty. Affairs v. Inclusive Cmtys. Project, Inc., ––– U.S. ––––, 135 S.Ct. 2507, 2522, 192 L.Ed.2d 514 (2015) ("An important and appropriate means of ensuring that disparate-impact liability is properly limited is to give housing authorities and private developers leeway to state and explain the valid interest served by their policies. This step of the analysis is analogous to the business necessity standard under Title VII and provides a defense against disparate-impact liability.").

The totality of the circumstances analysis is "local in nature." League, 769 F.3d at 243. The court undertakes this fact-intensive inquiry to determine "upon a searching practical evaluation of the past and present reality whether the political process is equally open to minority voters." Gingles, 478 U.S. at 79, 106 S.Ct. 2752 (citations and internal quotation marks omitted). "This determination is peculiarly dependent upon the facts of each case, and requires an intensely local appraisal of the design and impact of the contested electoral mechanisms." Id.(citations and internal quotation marks omitted). Where multiple electoral mechanisms are challenged, the court considers evidence of "their cumulative effect on minority access to the ballot box," League, 769 F.3d at 242, taking a " 'functional' view of the political process," Gingles, 478 U.S. at 45, 106 S.Ct. 2752 ; see alsoClingman v. Beaver, 544 U.S. 581, 607–08, 125 S.Ct. 2029, 161 L.Ed.2d 920 (2005) (O'Connor, J., concurring in part and concurring in the judgment) ("A panoply of regulations, each apparently defensible when considered alone, may nevertheless have the combined effect of severely restricting participation and competition."). The ultimate factual finding of vote denial or abridgement rests on this court's "particular familiarity with the indigenous political reality." Gingles, 478 U.S. at 79, 106 S.Ct. 2752.

As noted above, in addition to bearing the burden of demonstrating that the challenged voting practice results in an inequality of opportunity, a § 2 plaintiff also bears the burden of demonstrating that the inequality of opportunity is "caused by or linked to 'social and historical conditions' that have [produced] or currently produce discrimination against members of the protected class." League, 769 F.3d at 240 (quoting Gingles, 478 U.S. at 47, 106 S.Ct. 2752 ). For this reason, the Sixth Circuit has observed, the few courts that have considered vote denial claims have noted that the Gingles factors have been expressed either as a " 'causation requirement,' or through statements that a plaintiff cannot establish a Section 2 violation merely by showing a disproportionate impact or burden." Ohio State Conf. of NAACP v. Husted, 768 F.3d 524, 554 n. 9 (6th Cir.2014) (collecting cases), vacated on other grounds, No. 14–3877, 2014 WL 10384647 (6th Cir. Oct. 1, 2014).

Other courts have made the same observation. See, e.g., Frank v. Walker, 768 F.3d 744, 753 (7th Cir.2014) ("Section 2(b) tells us that § 2(a) does not condemn a voting practice just because it has a disparate effect on minorities. (If things were that simple, there wouldn't have been a need for Gingles to list nine non-exclusive factors in vote-dilution cases.)"); Gonzalez v. Arizona, 677 F.3d 383, 405 (9th Cir.2012) (en banc) ("[A] § 2 challenge based purely on a showing of some relevant statistical disparity between minorities and whites, without any evidence that the challenged voting qualification causes that disparity, will be rejected.") (citations and internal quotation marks omitted), aff'd sub nom.Arizona v. Inter Tribal Council of Ariz., Inc., ––– U.S. ––––, 133 S.Ct. 2247, 186 L.Ed.2d 239 (2013) ; League of United Latin Am. Citizens v. Clements, 999 F.2d 831, 866–67 (5th Cir.1993) (en banc) (hereinafter, " LULAC") ("[T]he Senate Report, while not insisting upon a causal nexus between socioeconomic status and depressed participation, clearly did not dispense with proof that participation in the political process is in fact depressed among minority citizens."); Salas v. Southwest Texas Jr. Coll. Dist., 964 F.2d 1542, 1556 (5th Cir.1992) (finding that plaintiffs "offered no evidence directly linking [lower turnout] with past official discrimination"); Irby v. Virginia State Bd. of Elections, 889 F.2d 1352, 1359 (4th Cir.1989) (rejecting § 2 claim because "[t]he evidence cast considerable doubt on the existence of a causal link between the appointive system and black underrepresentation"); Brown, 895 F.Supp.2d at 1249 ("[I]n the Eleventh Circuit a plaintiff must demonstrate something more than disproportionate impact to establish a Section 2 violation.").

In conducting its analysis, the court is guided by the fact that "Congress intended to give the Voting Rights Act 'the broadest possible scope.' " League, 769 F.3d at 240 (quoting Allen v. State Bd. of Elections, 393 U.S. 544, 567, 89 S.Ct. 817, 22 L.Ed.2d 1 (1969) ). This is reflected by the Senate Report, which found that because "courts have recognized that disproportionate education, employment, income level and living conditions arising from past discrimination tend to depress minority political participation ... [w]here these conditions are shown, and where the level of black participation in politics is depressed, plaintiffs need not prove any further causal nexus between their disparate socio-economic status and the depressed level of political participation." S. Rep. 97–417 at 29 n. 114, reprinted in 1982 U.S.C.C.A.N. at 207 n. 114.

Turning to the types of voting changes at issue in the present case, although many courts, including the Supreme Court in Crawford v. Marion County Election Board, 553 U.S. 181, 128 S.Ct. 1610, 170 L.Ed.2d 574 (2008), have examined the burden imposed on voters by requiring photo ID, few have conducted this analysis within the context of a § 2 claim.

Texas:

In August 2015, the Fifth Circuit applied the two-step test from the Sixth and Fourth Circuits to Texas's photo-ID requirement and upheld the district court's finding that it violated § 2 of the VRA. Veasey v. Abbott, 796 F.3d 487, 504, 513 (5th Cir.2015), rehearing en banc granted, 815 F.3d 958 (5th Cir.2016). The district court found "that 608,470 registered voters, or 4.5% of all registered voters in Texas," lacked qualifying ID. Id. at 505. Of those without qualifying ID, 534,512 did not qualify for a disability exemption from Texas's ID requirement. Id. at 505–06. The Fifth Circuit also credited the district court's finding that those without qualifying ID were disproportionately African American, Hispanic, and poor, and that "hundreds of thousands of voters face[d] round-trip travel times of 90 minutes or more to the nearest location issuing" acceptable ID. Id. at 506–07. Based on these findings, the court found that "[t]he district court did not err in concluding that [Texas's photo-ID requirement] disproportionately impacts Hispanic and African- American voters." Id. at 509. In analyzing whether this burden could be linked to social and historical conditions, the court reviewed the district court's findings on the Gingles factors. Id. at 509–512. Most pertinent here, the court did not disturb the district court's finding that Texas's photo-ID requirement was tenuous. Id. at 511–12. The court found that "[w]hile in-person voting fraud is rare and mail-in fraud is comparatively much more common, [Texas's ID law] would only combat the former." Id. at 511–12. The court also credited the district court's finding that there was "no credible evidence" that Texas's ID requirement would increase voter confidence or increase turnout, and that the requirement would instead decrease turnout. Id. at 512. The district court also discounted public opinion polls showing high levels of support for a photo-ID requirement on the ground that the polls did not contemplate the associated burden on minority voters. Id. Considering the district court's findings as a whole, the court held that the district court did not clearly err in finding Texas's ID requirement to violate § 2 of the VRA. Id. at 513. However, on March 9, 2016, the Fifth Circuit announced that it would rehear the case en banc. 815 F.3d 958 (5th Cir.2016). Under Fifth Circuit local rules, this has the effect of vacating the panel's opinion and judgment. 5th Cir. R. 41.3 (effect of granting rehearing en banc); Comer v. Murphy Oil, USA, 607 F.3d 1049, 1053 (5th Cir.2010).

Wisconsin:

In Frank v. Walker, 768 F.3d 744 (7th Cir.2014), the Seventh Circuit found that Wisconsin's photo-ID requirement did not violate the Constitution or § 2 of the VRA. Id. at 755. The court's characterization of the relevant burden and State interest in its constitutional analysis tended to shape its § 2 analysis. Seeid. at 753. The court tethered its burden analysis closely to the Supreme Court's decision in Crawford and found that Wisconsin's photo-ID requirement was no more burdensome than the Indiana law upheld in Crawford. See, e.g., id. at 745–46, 48–49. The court found that acquiring acceptable photo ID was sufficiently easy in Wisconsin that any disparities that existed were the product of disinterest, rather than disenfranchisement. In the court's own words:

Plaintiffs describe registered voters who lack photo ID as "disenfranchised." If the reason they lack photo ID is that the state has made it impossible, or even hard, for them to get photo ID, then "disfranchised" might be an apt description. But if photo ID is available to people willing to scrounge up a birth certificate and stand in line at the office that issues drivers' licenses, then all we know from the fact that a particular person lacks a photo ID is that he was unwilling to invest the necessary time. And Crawford tells us that "the inconvenience of making a trip to the [department of motor vehicles], gathering the required documents, and posing for a photograph surely does not qualify as a substantial burden on the right to vote, or even represent a significant increase over the usual burdens of voting."

Id. at 748 (quoting Crawford, 553 U.S. at 198, 128 S.Ct. 1610 ). In light of the non-substantial burden imposed by Wisconsin's photo-ID requirement, the court reasoned that the only basis for not upholding it would be if it did not serve any important purpose. Id. at 749. For reasons similar to those articulated by the district court in Veasey, the district court in Frank had held that Wisconsin's voter-ID law did not serve any legitimate purpose. Id. The Seventh Circuit viewed the district court's analysis as being contrary to Crawford, which, in its view, had established that, notwithstanding the absence of voter fraud in a jurisdiction, voter-ID requirements serve legitimate State interests. Id. at 749–51. Specifically, the court viewed the Supreme Court's finding that "a photo ID requirement promotes public confidence in the electoral system" as a " 'legislative fact'—a proposition about the state of the world, as opposed to a proposition about these litigants or about a single state." Id. at 750. According to the court:

Photo ID laws promote confidence, or they don't; there is no way they could promote public confidence in Indiana (as Crawford concluded) and not in Wisconsin. This means that they are valid in every state—holding constant the burden each voter must bear to get a photo ID—or they are valid in no state. Functionally identical laws cannot be valid in Indiana and invalid in Wisconsin (or the reverse), depending on which political scientist testifies, and whether a district judge's fundamental beliefs (his "priors," a social scientist would say) are more in line with the majority on the Supreme Court or the dissent.

Id. The court asserted that it was bound by such legislative facts "even if 20 political scientists disagree with the Supreme Court." Id.

Turning to the § 2 inquiry, the court concurred with the Sixth and Fourth Circuits that a vote denial claim consists of two elements, but expressed skepticism about the second step of the test on the ground that "it does not distinguish discrimination by the [State] from other persons' discrimination." Id. at 755. This distinction, however, did not prove material, as the court said that even if it were to adopt the test in full, plaintiffs would fail the first step. Id. Despite the district court's finding that minorities were less likely to possess qualifying ID, id. at 746, the Seventh Circuit held that Wisconsin's law nevertheless provided "everyone ... the same opportunity to get a qualifying photo ID," id. at 755.

Despite the guidance offered by Veasey and Frank, the cases are of limited assistance in deciding Plaintiffs' § 2 claim in this case because neither Texas nor Wisconsin had a reasonable impediment exception. In fact, this court is not aware of any case where a photo-ID requirement with a reasonable impediment exception has been considered under § 2. This is significant because the reasonable impediment exception provides a fail-safe voting option for the very groups that Veasey and Frank centered on: socioeconomically-disadvantaged individuals without qualifying ID.

Only a small handful of cases have addressed a legislature's modification under § 2 of the other voting mechanisms in this case. Briefly, they are:

South Dakota:

In Brooks v. Gant, No. 12cv5003, 2012 WL 4482984 (D.S.D. Sept. 27, 2012), Native American voters living within Shannon County, South Dakota, sued for declaratory and injunctive relief based on the county's failure to establish early, in-person voting locations and days similar to those offered in virtually every other county. Id. at *1. Plaintiffs alleged that their six days of early voting during the 2012 primary and general elections, which was less than that to be offered in similar counties, would require those who missed the dates to travel between one and three hours to the next nearest early-voting site and was less than the opportunity provided to white voters. Id. at *7. Coupled with allegations that Native Americans distrust voting by mail and that early voting in the county in the past saw increased (nearly doubled) turnout, these claims were held sufficient to state a claim under § 2. Id. at *7–8. In other words, the court held, having decided to offer early voting, a State must offer it equally.

Florida:

In Jacksonville Coalition for Voter Protection v. Hood, 351 F.Supp.2d 1326 (M.D.Fla.2004), voter groups sought, under § 2, to enjoin Duval County, Florida's decision not to offer more early-voting sites for the 2004 election. Id. at 1328, 1334. Initially, only one site was planned, but officials agreed to add four more sites. Id. at 1330. Plaintiffs alleged that Duval County had the "largest percentage of African-American registered voters of Florida's most densely populated counties" and that these African American voters would have less opportunity to vote in the upcoming election and thus be disproportionately affected. Id. at 1334. The district court denied the requested injunction because the record showed that three of the four new sites were in African American communities, alleviating travel burdens; that the communities farther away were predominantly white; and that the lack of days before the election rendered the request infeasible. Id. at 1335–37. In dicta, the court expressed concern over the "far-reaching implications" of plaintiffs' argument; namely, that it would support an argument to require a State with a higher percentage of African American voters but without early voting to adopt it, or would require a State with a lower percentage of African American voters than Florida to offer more days of early voting. Id. at 1335–36. The court concluded: "This simply cannot be the standard for establishing a Section 2 violation." Id. at 1336.

Separately, in May 2011, Florida decided, as part of some eighty sets of changes to election law in HB 1355, to reduce early voting from a possible 12–14 days to 8 days, thus eliminating the Sunday before Election Day and granting supervisors discretion to allow as few as forty-eight hours of early voting. Florida v. United States, 885 F.Supp.2d 299, 308–09, 302 (D.D.C.2012) (per curiam) (panel comprised of Garland, Kollar-Kotelly, and Huvelle). The U.S. Attorney General refused to pre-clear the law under § 5, and Florida sued in the United States District Court for the District of Columbia. Id. at 302. The court also declined to pre-clear the law because Florida failed to prove that the changes were not retrogressive (the § 5 standard), but the court stated that it could likely do so if the new law required the same number of voting hours as the old one (ninety-six hours) and did so on a 7 a.m. to 7 p.m. basis. Id. at 357. Despite testimony from voting rights groups that the two-week period had been essential to GOTV efforts, the court found that the GOTV groups "could adjust to a redistribution of the total 96 hours over a different number of days, including weekend days and a 'souls-to-the-polls' Sunday." Id. at 337. In fact, the court found, "[e]xpanding convenient non-working weekday and weekend voting hours should ... help third-party efforts to provide transportation to the polls for such voters." Id. The court also noted that the implementation of such hours, which establishes non-retrogression, "will be a significant factor in favor of a finding of nondiscriminatory purpose." Id. at 350–51. Florida made the change, and DOJ agreed not to object to pre-clearance.

Thereafter, a group of African American leaders sued to enjoin the law under § 2. Brown, 895 F.Supp.2d at 1238. On motion for preliminary injunction, the court denied relief, finding a lack of substantial likelihood of success as to either discriminatory intent or discriminatory results. Id. at 1255–56. This, even though African Americans disproportionately used early voting; the law was introduced via a "strike-all" amendment process the day before it was to be taken up; public comment was limited to three minutes per person; one State senator made a floor comment that "he did not want to make it easier to vote"¹²⁸ ; and, as noted above, the law was part of some eighty changes to Florida's election rules. Id. at 1246–48. Proponents argued that the law was intended to "increase flexibility and make the early voting process more efficient," and to be a "compromise" in reducing days but granting greater flexibility as to hours. Id. at 1248. Plaintiffs argued that the changes would not actually reduce, but would more likely increase, costs, were not the result of any empirical study, and did not serve the voting public as well as a fifteen-day time period. Id. The court noted the distinction between the § 5 inquiry of retrogression and the present § 2 analysis:

128 Worse, the senator was the Senate President Pro Temporeand added that "it should be harder to vote—as it is in Africa." Brown, 895 F.Supp.2d at 1247. The court noted that the three-judge § 5 court found this statement insufficient proof of discriminatory legislative intent because the senator was not a bill proponent and had little authority, and likewise declined to find discriminatory intent. Id.at 1247–48.

Moreover, in contrast to the Section 5 case before the Florida court, this Court, as Plaintiffs conceded at the hearing, is not conducting a "retrogression" analysis, meaning, this Court is not comparing the new statute against the old to determine whether these voting changes will "worsen the position of minority voters in comparison to the preexisting voting standard, practice, or procedure." Rather, the task before this Court under Section 2 of the VRA is to conduct a "practical evaluation of the past and present reality" to determine whether, under the totality of the circumstances, application of the 2011 Early Voting Statute serves to deny African American voters equal access to the political process. The important distinction between a Section 5 and a Section 2 claim plays a significant role in the Court's decision in this case.

Id. at 1251 (citations and internal quotation marks omitted). Despite accepting the findings of experts (Drs. Gronke and Stewart) that the changes would disproportionately affect African American voters, id., the Brown court found that "[b]ecause [the new statute] allows early voting during non-working hours, as well as voting during the weekend, including one Sunday, voting times which are important to African American voters, as well as to GOTV efforts, the Court cannot find that [it] denies equal access to the polls," id. at 1255. In doing so, the court also found the experience of other States important (noting similar statistics to those presented in the present case) and cautioned that "acceptance of Plaintiffs' argument that the eight days of early voting allowed by the Florida legislature violates Section 2 could have far-reaching implications," citing the concerns raised in Hood, 351 F.Supp.2d at 1335–36. Id. at 1254. The court framed the issue as follows: "whether the State of Florida, having decided to allow early voting, has adopted early voting procedures that provide equal access to the polls for all voters in Florida." Id. at 1254–55.

Ohio:

In Ohio State Conference of N.A.A.C.P. v. Husted, 768 F.3d 524 (6th Cir.2014), vacated on other grounds,¹²⁹ No. 14–3877, 2014 WL 10384647 (6th Cir. Oct. 1, 2014), plaintiffs moved to enjoin Ohio's adoption of SB 238 and Directive 2014–17 under § 2 of the VRA and the Fourteenth Amendment. Id. at 529. At issue was Ohio's reduction of early voting¹³⁰ from thirty-five to thirty days, thus eliminating five days that, because they overlapped with ordinary registration, were known as "Golden Week." Id. at 532. Factually, the case is distinguishable: Ohio had only one early-voting site available per county, open only 8 a.m. to 5 p.m., id. at 539 ; the State provided no evidence that it could not verify Golden Week SDR-voters as easily as others who registered on the last day of regular registration, especially since all absentee voters were verified on Election Day and the State had thirty days to verify even those who registered during Golden Week, id. at 547 ; and there was no evidence that local boards of elections had struggled sufficiently to cope with early voting administratively, id. at 548 (quoting Obama for America v. Husted, 697 F.3d 423, 443–33 (6th Cir.2012) (holding that the State's asserted interest in reducing costs and administrative burdens did not justify the burdens on voters because there was "no evidence that local boards of elections have struggled to cope with early voting in the past")). Some of the same experts testified, including Dr. Gronke, Dr. Burden, and Mr. Trende. The district court enjoined the elimination, and the Sixth Circuit affirmed. Id. at 561.

129 It appears that the case was dismissed after some type of settlement that left in place the law's elimination of the overlap of five days of registration with early voting during Golden Week. A new set of plaintiffs have filed suit challenging various provisions. SeeOhio Democratic Party v. Husted, No. 2:15cv1802 (S.D.Ohio 2015).

130 Ohio had adopted early voting in 2005 in response to problems in the 2004 election that caused voters to wait from two to twelve hours to vote. Id.at 530–31.

On the § 2 claim, the Sixth Circuit articulated the two-part test set forth above. It then explained that "what is distinct between a Section 5 analysis and a Section 2 analysis is the role that prior law plays in the comparison," stating that "under the Section 2 analysis, the focus is whether minorities enjoy less opportunity to vote as compared to other voters. " Id. at 558. It was in this manner that disproportionate use of the old law was relevant. Id. The court held that plaintiffs presented evidence that the eliminated early-voting times "are those that African Americans disproportionately use, and that racial inequalities in socioeconomic status and other factors make it much more difficult for African Americans to vote at the remaining times or through the other methods now available under the status quo as compared to other groups." Id. at 558–59.¹³¹ As noted, the Sixth Circuit subsequently vacated its decision.

131 On the Anderson–Burdickclaim, the court affirmed the district court's finding that the burden was "significant" but not "severe" as not clearly erroneous "given the extensive evidence in the record of the burdens African American, lower-income, and homeless voters will face in voting, absent the times eliminated by SB 238 and Directive 2014–17." Husted, 768 F.3d at 545.

North Carolina:

In League of Women Voters of N.C. v. North Carolina, 769 F.3d 224, 242 (4th Cir.2014), the Fourth Circuit applied the two-part test from Husted to the record that existed at this case's preliminary injunction stage. The court found that Plaintiffs were likely to succeed on their § 2 claim as to the elimination of SDR and OOP. Id. at 247. In reaching this conclusion, the court reasoned that, on the record available, SDR and OOP "were enacted to increase voter participation, [ ] African American voters disproportionately used those electoral mechanisms, and [SL 2013–381] restricted those mechanisms and thus disproportionately impacts African American voters."Id. Viewed in light "of relevant 'social and historical conditions' in North Carolina," the court found that SL 2013–381 "looks precisely like the textbook example of Section 2 vote denial." Id. at 246.

As is plain from the court's findings above, the trial evidence contradicted many of the factual premises Plaintiffs presented at the preliminary injunction stage that underlay the Fourth Circuit's decision. For example, the evidence demonstrated that early voting actually depresses voter turnout, and SDR, because it is offered as a part of early voting (compared to EDR, which is offered on Election Day), has not been shown to increase turnout in a statistically significant manner. Further, Plaintiffs failed to show that OOP increases voter turnout. In addition, while disproportionate use was the principal evidence available to predict the effects of SL 2013–381 at the time, the 2014 election data is now available and provides this court with evidence of how minorities actually participate under SL 2013–381.

Accordingly, this court will apply the Fourth Circuit's articulated legal principles, including the two-prong test, and other relevant case law to the fully-developed, extensive trial record.

2. The Totality of the Circumstances & Gingles

The inquiry under the totality of the circumstances is whether the provisions of SL 2013–381, individually or cumulatively, "impose a discriminatory burden" on Hispanics and African Americans, "caused by or linked to social and historical conditions that have or currently produce discrimination," such that they have "less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice." League, 769 F.3d at 240 (internal quotation marks omitted). Although "there is no requirement that any particular number of factors be proved, or [even] that a majority of them point one way or the other," League, 769 F.3d at 240 (citing Gingles, 478 U.S. at 45, 106 S.Ct. 2752 ), the Fourth Circuit has directed that the Gingles factors "may shed some light on whether the two elements of a Section 2 claim are met," id. Accordingly, this court turns to an analysis of the Gingles factors as reflected by the court's findings of fact.

a. The Success of the Prior Practices in Fostering Minority Political Participation

As noted, the Fourth Circuit directed that the "success[ ]" of North Carolina's "previous voting practices" in "fostering minority participation" is "centrally relevant" under § 2, being a "critical piece of the totality-of-the-circumstances analysis Section 2 requires." League, 769 F.3d at 242.

Despite this early appellate guidance, Plaintiffs offered no evidence that the old early-voting schedule, SDR, or OOP voting "foster[ed] minority participation."¹³² In fact, as noted above the evidence showed that early voting depresses voter turnout and SDR has not been shown to increase turnout in a statistically significant manner.

132 Although Plaintiffs showed that pre-registration increases turnout generally, the evidence demonstrated that its benefits are race-neutral. There is no evidence in this case that the system of signature attestation that preceded the voter-ID requirement fostered minority political participation.

Instead, Plaintiffs' experts rely on disproportionate use statistics for select elections to argue that African Americans were able to exceed parity with whites in terms of registration and turnout because of the eliminated procedures. This evidence is certainly relevant, but, as the court's findings of fact show, it does not justify such an inference. It cannot be that the Fourth Circuit's directive was too burdensome on Plaintiffs, because their experts admitted that they were capable of analyzing how such procedures affected political participation and have done so in the past. For whatever reason, however, they declined to produce such evidence in this case.

Plaintiffs did present evidence that African Americans used early voting as part of GOTV and "souls-to-the-polls" efforts, largely as a component of targeted campaigns to capitalize on the early voting and late registration opportunity. Such evidence of prior, disproportionate use of the electoral mechanisms is evidence of, but does not equate to, fostering minority participation. As a three-judge panel has cautioned as to early voting in the § 5 context:

[A] change is not retrogressive simply because it deals with a method of voting or registration that minorities use more frequently, or even because it renders that method marginally more difficult or burdensome. Rather, to be retrogressive, a ballot access change must be sufficiently burdensome that it will likely cause some reasonable minority voters not to register to vote, not to go to the polls, or not to be able to cast an effective ballot once they get to the polls.

Florida, 885 F.Supp.2d at 312. Instead, where there have been subsequent elections without the desired mechanisms, the more probative evidence is the participation data from the later elections without the desired mechanisms. SeePurcell, 549 U.S. at 6, 127 S.Ct. 5 (Stevens, J., concurring) (preferring "historical facts rather than speculation"). Here, this latter evidence suggests strongly that the eliminated mechanisms were not the cause of increased minority political participation because African American and Hispanic turnout rates (compared to the previous midterm) and registration rates (compared to either the previous midterm or even the previous presidential election) increased without the mechanisms. In addition, the parties all agree that a campaign strategy of targeting the removed mechanisms was a driver of minority use in 2008 and 2012.

In Shelby County v. Holder, ––– U.S. ––––, 133 S.Ct. 2612, 186 L.Ed.2d 651 (2013), the Supreme Court struck down as unconstitutional § 4(b) of the VRA, which was the formula used to determine which States and political subdivisions were subject to pre-clearance by the DOJ. Id. at 2631. The Supreme Court's decision in Shelby County was unmistakably premised on the fact that minority turnout and registration rates in most jurisdictions covered by § 5 were at or above parity. Id. at 2618–19 ("Census Bureau data indicate that African–American voter turnout has come to exceed white voter turnout in five of the six States originally covered by § 5, with a gap in the sixth State of less than one half of one percent.").¹³³ Likewise, there can be little dispute that minority turnout and registration rates are important to the purposes of the VRA, given that the pre-clearance coverage formula was based on those precise metrics. Id. at 2619 ("At the time of the Act's passage, these 'covered' jurisdictions were those States or political subdivisions that had maintained a test or device as a prerequisite to voting as of November 1, 1964, and had less than 50 percent voter registration or turnout in the 1964 Presidential election."); id. at 2620 ("In 1970, Congress reauthorized the Act for another five years, and extended the coverage formula in § 4(b) to jurisdictions that had a voting test and less than 50 percent voter registration or turnout as of 1968."); id.("In 1975, Congress reauthorized the Act for seven more years, and extended its coverage to jurisdictions that had a voting test and less than 50 percent voter registration or turnout as of 1972."); id. at 2625 ("Those conclusions are not ours alone. Congress said the same when it reauthorized the Act in 2006, writing that '[s]ignificant progress has been made in eliminating first generation barriers experienced by minority voters, including increased numbers of registered minority voters, minority voter turnout, and minority representation in Congress, State legislatures, and local elected offices.' " (quoting Fannie Lou Hamer, Rosa Parks, and Coretta Scott King Voting Rights Act Reauthorization and Amendments Act of 2006, § 2(b)(1), 120 Stat. 577)); id.("The House Report elaborated that 'the number of African–Americans who are registered and who turn out to cast ballots has increased significantly over the last 40 years, particularly since 1982,' and noted that '[i]n some circumstances, minorities register to vote and cast ballots at levels that surpass those of white voters.' " (quoting H.R. Rep. No. 109–478, at 12 (2006), reprinted in 2006 U.S.C.C.A.N. 618, 627)).

133 SeealsoShelby Cty., 133 S.Ct. at 2621 ("[T]hings have changed in the South. Voter turnout and registration rates now approach parity. Blatantly discriminatory evasions of federal decrees are rare. And minority candidates hold office at unprecedented levels." (quoting Northwest Austin Mun. Util. Dist. No. One v. Holder, 557 U.S. 193, 202, 129 S.Ct. 2504, 174 L.Ed.2d 140 (2009) )); id.at 2627 ("When upholding the constitutionality of the coverage formula in 1966, we concluded that it was 'rational in both practice and theory.' The formula looked to cause (discriminatory tests) and effect (low voter registration and turnout), and tailored the remedy (preclearance) to those jurisdictions exhibiting both." (quoting South Carolina v. Katzenbach, 383 U.S. 301, 330, 86 S.Ct. 803, 15 L.Ed.2d 769 (1966) )); id.at 2627–28 ("Coverage today is based on decades-old data and eradicated practices. The formula captures States by reference to literacy tests and low voter registration and turnout in the 1960s and early 1970s. But such tests have been banned nationwide for over 40 years. And voter registration and turnout numbers in the covered States have risen dramatically in the years since. Racial disparity in those numbers was compelling evidence justifying the preclearance remedy and the coverage formula. There is no longer such a disparity." (citations omitted)); id.at 2628 ("In 1965, the States could be divided into two groups: those with a recent history of voting tests and low voter registration and turnout, and those without those characteristics. Congress based its coverage formula on that distinction. Today the Nation is no longer divided along those lines, yet the Voting Rights Act continues to treat it as if it were."); id.("[Between 1965 and 2006], largely because of the Voting Rights Act, voting tests were abolished, disparities in voter registration and turnout due to race were erased, and African–Americans attained political office in record numbers.").

Historically, when courts have found § 2 violations, they have frequently grounded that decision in part on lagging minority turnout and registration rates. See, e.g., Veasey, 796 F.3d at 510 ; Miss. State Chapter, Operation Push, Inc. v. Mabus, 932 F.2d 400, 413 (5th Cir.1991) ; Gomez v. City of Watsonville, 863 F.2d 1407, 1416 n. 4 (9th Cir.1988) ("Low voter registration and turnout have often been considered evidence of minority voters' lack of ability to participate effectively in the political process."); Jeffers v. Tucker, 847 F.Supp. 655, 660 (E.D.Ark.1994) ; Ward v. Columbus Cty., 782 F.Supp. 1097, 1104 (E.D.N.C.1991) ; Neal v. Coleburn, 689 F.Supp. 1426, 1428 (E.D.Va.1988). And when courts find no § 2 violation, they also rely on the electoral mechanism's effect on minority success in turnout and registration rates, or else find the failure to produce such evidence fatal. See, e.g., Frank, 768 F.3d at 747, 751, 753–54 ; Smith v. Brunswick Cty., 984 F.2d 1393, 1395, 1400–02 (4th Cir.1993) ; LULAC, 999 F.2d at 867 ; id. at 914 n. 16 (King, J., dissenting); Salas, 964 F.2d at 1550, 1555–56 ; Solomon v. Liberty Cty., 957 F.Supp. 1522, 1563–65 (N.D.Fla.1997), aff'd sub nom.Solomon v. Liberty Cty. Comm'rs, 221 F.3d 1218 (11th Cir.2000) (en banc). As Justice Thomas has explained, turnout and registration rates are the primary metrics of whether electoral procedures have "served to suppress minority voting." Holder v. Hall, 512 U.S. 874, 926–27, 114 S.Ct. 2581, 129 L.Ed.2d 687 (1994) (Thomas, J., concurring) ("[I]n some circumstances, results in recent elections might also be relevant for demonstrating that a particular practice concerning registration or polling has served to suppress minority voting. Better factors to consider would be figures for voter registration or turnout at the last election, broken down according to race.").

The point is not that the 2014 turnout and registration rates are the only or dispositive evidence of causation in this case; they are not. SeeFrank, 768 F.3d at 754. It is merely that Plaintiffs have not carried the burden of showing that the former electoral provisions "foster[ed] minority participation," League, 769 F.3d at 242, and the State has offered evidence that they did not. Though capable of producing such evidence, and if the mechanisms had such an effect, Plaintiffs did not introduce it. Rather, they offered earlier disproportionate use figures, which are less probative on this "critical piece of the totality-of-the-circumstances analysis" than the 2014 data. League, 769 F.3d at 242.

In sum, when increased minority participation correlates with the availability of so-called convenience voting procedures,¹³⁴ it is some evidence that they are fostering the increased participation. But when minority participation increases despite the unavailability of such mechanisms, the causal inference is rebutted, especially where Plaintiffs have failed to show what turnout and registration rates would have been in 2014 had the conveniences been available. Such is the case here. The court therefore finds that this consideration does not favor the Plaintiffs.

134 As noted in the findings of fact, the correlation itself is far from clear, and significant evidence, besides the 2014 data, tends to rebut the inference of causation.

b. History of Official Discrimination

The first Gingles factor has the court examine "the extent of any history of official discrimination in the state or political subdivision that touched the right of the members of the minority group to register, to vote, or otherwise to participate in the democratic process." 478 U.S. at 36–37, 106 S.Ct. 2752. As the text of the factor itself expresses, the relevant history of discrimination must be "official."

All evidence of discrimination is relevant, and North Carolina has a sordid history dating back well over a century that the court fully considers. But, as with most evidence, "contemporary examples of discrimination are more probative than historical examples." Veasey, 796 F.3d at 509 (citing Shelby Cty., 133 S.Ct. at 2628 ). The Supreme Court has cautioned, "[P]ast discrimination cannot, in the manner of original sin, condemn governmental action that is not itself unlawful. ... More distant instances of official discrimination in other cases are of limited help ...." City of Mobile v. Bolden, 446 U.S. 55, 74, 100 S.Ct. 1519, 64 L.Ed.2d 47 (1980), superseded on other grounds by statute, Voting Rights Act Amendments of 1982, Pub. L. No. 97–205, 96 Stat. 131; accordWesley v. Collins, 791 F.2d 1255, 1261 (6th Cir.1986) ; Riddick v. Sch. Bd. of City of Norfolk, 784 F.2d 521, 539 (4th Cir.1986) ; cf.Shelby Cty., 133 S.Ct. at 2629 ("The [Fifteenth] Amendment is not designed to punish for the past; its purpose is to ensure a better future."); Freeman v. Pitts, 503 U.S. 467, 495–96, 112 S.Ct. 1430, 118 L.Ed.2d 108 (1992) ("In one sense of the term, vestiges of past segregation by state decree do remain in our society and in our schools. Past wrongs to the black race, wrongs committed by the State and in its name, are a stubborn fact of history. And stubborn facts of history linger and persist. But though we cannot escape our history, neither must we overstate its consequences in fixing legal responsibilities.").

Finally, the relevant history of official discrimination must be brought "home to this case." LULAC, 999 F.2d at 847. It must interact with the challenged provisions of SL 2013–381. League, 769 F.3d at 239 ; accordSalas, 964 F.2d at 1556 ; NAACP v. City of Columbia, 850 F.Supp. 404, 421 (D.S.C.1993) ("In this case, no credible evidence of present day discrimination was presented. There was, however, an abundance of uncontroverted testimony about South Carolina's past history of discrimination affecting voting, and the plaintiffs have unquestionably prevailed as to this part of the first Senate Report factor. Nevertheless, the court finds that there are no vestiges of past discrimination that significantly interact with present political structures to deny access to the political system.").

With this in mind, the court turns to the testimony of James L. Leloudis, II, Ph.D., Professor of History at the University of North Carolina at Chapel Hill.¹³⁵ Dr. Leloudis was Plaintiffs' primary expert on the history of official discrimination in North Carolina.¹³⁶ He divides the history of official discrimination against African Americans in North Carolina into three chapters: Civil War to Jim Crow; Jim Crow to the civil rights movement; and the civil rights movement to the present day. (Doc. 338 at 16.)

135 Dr. Leloudis was proffered without objection as an expert in the history of the United States and North Carolina. (Doc. 338 at 15–16.)

136 As analyzed further infra, Charles T. Clotfelter, Ph.D., testified concerning the history of educational disparities in North Carolina's public schools.

After the close of the Civil War, the government of North Carolina instituted the Black Codes, which sought to reduce the rights, including suffrage, of newly-emancipated slaves. (Pl. Ex. 230 at 4–5.) Federal Reconstruction intervened and caused North Carolina to revise its State Constitution, ensuring political liberties regardless of race. (Id. at 5–7.) As a result of the expansion of the franchise, in 1868, the bi-racial Republican Party took control of the governorship, legislature, and congressional delegation. (Id. at 7.) By 1880, just fifteen years after the Civil War, voter participation in North Carolina was 78% for whites but an astounding 90% for African Americans. (Pl. Ex. 237 at 13.) An alliance of whites and African Americans continued to field candidates and win elections, despite political violence by the growing Ku Klux Klan. (Pl. Ex. 230 at 7–8.)

In 1894, this alliance controlled the General Assembly and enacted legislation with what Dr. Leloudis describes as "the aim of guaranteeing full and fair access to the franchise." (Id. at 10.) These progressive changes, relevant to this case, included the creation of the precinct system, and a requirement that citizens register and vote in their precinct of residence. Act of Mar. 8, 1895, ch. 159, §§ 5, 10, 14, 1895 N.C. Sess. Laws 211. Registrars and election judges for each precinct were to be appointed from "each political party of the state," rather than the election system being controlled entirely by one political party, as was the previous practice. Id.§ 7; (Pl. Ex. 230 at 10). Ballot challenges could be made by any voter, without regard to the residence of the challenger. Act of Mar. 8, 1895, ch. 159, § 12. There was also introduced one day on which everyone would vote. Id.§ 3. On this day, the polls were to open at 7 a.m. and run "until sunset of the same day, and no longer." Id.§ 17. Felons, women, and those below the age of twenty-one were disqualified from voting. Id.§§ 13–14. The voter was also only qualified if he had lived in the State for twelve months and the county for ninety days preceding Election Day. Id.§ 14.¹³⁷ Against this historical backdrop and even under the conditions at the time, turnout among African Americans, many or most of whom were former slaves, rose remarkably from 60% to 90%. (Pl. Ex. 230 at 11.)

137 There were various other provisions of the law, including a prohibition on voter intimidation and public disclosure of campaign financing. (SeePl. Ex. 230 at 10–11.)

Things changed in the run-up to the 1898 election. Democrats engaged in a campaign full of racial violence and racial appeals. (Id. at 13.) For example, one prominent Democratic newspaper published a cartoon bearing the caption "The Vampire that Hovers over North Carolina" that portrayed an African American face with "Negro Rule" on the wings. (Id.) As Election Day 1898 arrived, Klansmen engaged in violence and voter intimidation at the polls. Democrats won the election by a slim margin. (Id. at 16.)

Now in control of the General Assembly, the Democrats enacted a literacy test and other laws that had the effect of suppressing the vote of African Americans and supporters of minority political parties. (Id. at 16.) At this time, the SBOE and CBOEs were created, having control over local registrars and election judges, and all board members were appointed by the party in control of the General Assembly. Act of Mar. 6, 1899, ch. 507, §§ 4–5, 1899 N.C. Sess. Laws 658, 659. The new laws also required every North Carolinian to re-register, which would clearly make it easier for Klansmen to intimidate African Americans. Id.§ 11.

In 1900, the Democrats again succeeded at the polls and would remain in power for over 100 years, until the 2000s. (Pl. Ex. 230 at 18.) The following decades saw the introduction of Jim Crow laws and other forms of segregation. (Id. at 19–20.) They also saw new discriminatory laws in the areas of education and labor. (Id.)

The period from 1917 to 1950 saw gains for African American political participation. African Americans engaged in political organization and successfully managed to register, despite the literacy test. (Id. at 20–22.) Around 1950, segregation and the civil rights movement began to move to the forefront of American politics. (Id. at 22–25.) The era saw racial appeals at the center of campaigns. But in the 1960s, Terry Sanford was elected governor, and he began to attack what he called the "poverty-segregation complex" in the State. (Id. at 25.) At the federal level, the Civil Rights Act was passed in 1964 and the VRA in 1965. (Id. at 25–26.) The General Assembly, meanwhile, instituted multimember legislative districts that tended to dilute African American voting strength, which continued into the 1980s. (Id. at 27–28.)¹³⁸Dr. Leloudis does not catalog any official discrimination after the 1980s. By the turn of that decade, African Americans were making significant headway in political strength. By 1989, there were nineteen African Americans in the General Assembly, more than ever before. (Id. at 28.) By 1991, the office of Speaker of the House was held by Representative Dan Blue, an African American. (Id.) Redistricting under his leadership resulted in new majority-minority districts. (Id.) These successes paralleled local successes for African Americans. (Id. at 29.) By 2001, 22% of the governor's appointees to State agencies and commissions were minorities, which matched the State's demographics. (Id. at 29–30.) This recent period saw the election of Henry Frye, the first African American chief justice of the North Carolina Supreme Court, and the elevation of Representative Henry Michaux as the first African American leader of the House Appropriations Committee. (See, e.g., Doc. 338 at 41–42.)

138 Dr. Leloudis' testimony and expert report do not focus on legislative districts after Gingles. The court notes, however, that North Carolina's district maps continued to be a source of litigation after the 1980s, most notably with regard to North Carolina's Twelfth Congressional District. See, e.g., Easley v. Cromartie, 532 U.S. 234, 257–58, 121 S.Ct. 1452, 149 L.Ed.2d 430 (2001) (holding that plaintiffs failed to produce sufficient evidence to show that one version of the Twelfth District had been drawn predominantly on the basis of race); Shaw v. Hunt, 517 U.S. 899, 906, 116 S.Ct. 1894, 135 L.Ed.2d 207 (1996) (holding that an earlier version of the Twelfth District, which appeared to have been drawn to "assure black voter majorities," was an unconstitutional racial gerrymander); Bartlett v. Strickland, 556 U.S. 1, 8, 129 S.Ct. 1231, 173 L.Ed.2d 173 (2009) (holding that the State's creation of certain "crossover" State districts to enhance minority voting power was not required by the VRA and therefore violated the "whole county" provision of the State constitution). In addition, though again not raised by the parties, the court notes that similar issues have arisen in three lawsuits involving North Carolina's current State and congressional districts, two of which remain pending in this district. SeeDickson v. Rucho, 368 N.C. 481, 781 S.E.2d 404, 441 (2015) (finding no federal constitutional violation for several congressional districts, including the First and Twelfth Districts, as well as several State districts); Harris v. McCrory, 159 F.Supp.3d 600, 627–28, No. 1:13cv949, 2016 WL 482052, at *21 (M.D.N.C. Feb. 5, 2016) (invalidating maps for the First and Twelfth Congressional Districts as racial gerrymanders); Covington v. North Carolina, No. 15cv399 (M.D.N.C.) (challenging various State districts, many of which were held valid in Dickson).

During the period from 1992 to 2009, no-excuse early voting, no-excuse absentee voting, SDR, OOP voting, and pre-registration were enacted. (Pl. Ex. 230 at 30.) Dr. Leloudis admittedly did not review the legislative record associated with these changes. (Doc. 338 at 39–41.) Nevertheless, he asserts that "the net effect of these reforms was steady improvement in voter turnout." (Pl. Ex. 230 at 30.) Neither his report nor his testimony substantiates such a claim. Nor has any of the Plaintiffs' other experts. In fact, during trial, Dr. Leloudis confessed he was "not comfortable with that kind of speculation." (Doc. 338 at 45–47.) In short, he was contradictory: he stood by his opinion that 2012 turnout would have been lower absent the election reforms at issue, but he refused to answer whether he would have predicted that the turnout would have gone down in 2014 as being unwarranted speculation. (Id.)

Finally, Dr. Leloudis opined on the challenged provisions of SL 2013–381. Here, his opinions lost their moorings in history and became part advocacy. For example, he failed to consider the entirety of SL 2013–381, did not meaningfully consider the justifications for the provisions, and noted that while it was possible that partisans can have policy differences without being racist, his "preponderance of associations" led him to conclude otherwise. (Id. at 36–38)

Overall, the court credits Dr. Leloudis' historical account up to the introduction of the voting laws at issue. But from there, he carries little credibility because his findings ignore or cursorily discount relevant facts inconsistent with his opinions and are not based on "sufficient facts." Fed. R. Evid. 702(b). These opinions are less those of a detached expert and more those of an advocate. Va. Elec. & Power Co. v. Sun Shipbuilding & Dry Dock Co., 68 F.R.D. 397, 406 (E.D.Va.1975) ("It is evident from the definitions that an expert is expected to owe his allegiance to his calling and not to the party employing him. In order to be an expert one must be in a position to testify as an expert and not as a partisan.").

In sum, the court has considered the history of official racial discrimination in North Carolina (and unofficial discrimination, to the extent relevant and offered), including the opinions of Charles T. Clotfelter, Ph.D., Professor of Law, Economics, and Public Policy at Duke University, that are discussed below. There is significant, shameful past discrimination. In North Carolina's recent history, however, certainly for the last quarter century, there is little official discrimination to consider.

c. Racially-Polarized Voting

The second Gingles factor considers "the extent to which voting in the elections of the state or political subdivision is racially polarized." 478 U.S. at 37, 106 S.Ct. 2752.

The Supreme Court has noted that while "[s]ome commentators suggest that racially polarized voting is waning—as evidenced by, for example, the election of minority candidates where a majority of voters are white..., [s]till, racial discrimination and racially polarized voting are not ancient history." Bartlett v. Strickland, 556 U.S. 1, 25, 129 S.Ct. 1231, 173 L.Ed.2d 173 (2009) (per Kennedy, J.).¹³⁹ In North Carolina, African Americans overwhelmingly vote for Democratic candidates; for example, at the presidential level they voted 85% for John Kerry in 2004 and 95% for President Obama in 2008. (Pl. Ex. 231 at 24.) By contrast, 27% of white North Carolinians voted for John Kerry in 2004 and 35% for President Obama in 2008. (Id.) That is, white support for the preferred candidate of African Americans increased when there was an African American running for President. (Id.) In 2008, President Obama carried North Carolina. Defendants admit in their answer in case 1:13cv861 that "past court decisions in the voting rights area speak for themselves and that racially polarized voting continues to exist in North Carolina." (Doc. 19 at 13.)

139 Justices Thomas and Scalia concurred in the judgment only, adhering to their opinion that § 2 does not countenance vote dilution claims. Bartlett, 556 U.S. at 26, 129 S.Ct. at 1250 (Thomas, J., concurring in the judgment).

The court finds that polarized voting between African Americans and whites remains in North Carolina, so this factor favors Plaintiffs. Plaintiffs have not presented any evidence regarding Hispanics and racially polarized voting. (See, e.g., Pl. Ex. 231 at 20–26; Pl. Ex. 229 at 7–8.)

d. Enhancing the Opportunity for Discrimination

The third Gingles factor has the court consider "the extent to which the state or political subdivision has used unusually large election districts, majority vote requirements, anti-single shot provisions, or other voting practices or procedures that may enhance the opportunity for discrimination against the minority group." 478 U.S. at 37, 106 S.Ct. 2752.

Plaintiffs do not claim that any of these procedures are currently used in North Carolina. Rather, in support of this factor, Plaintiffs refer to voting restrictions imposed in the late nineteenth century, (Pl. Ex. 229 at 9); a poll tax that lasted until 1920, (id.); a literacy test that is currently part of the North Carolina State Constitution, though rendered unenforceable by the VRA since 1965, (id. at 9–10); and various objection letters issued by the DOJ when North Carolina was subject to § 5 preclearance, (id. at 10). Plaintiffs also cite to various legal decisions from the 1980s that collect this history. (Doc. 346 at 128–29.) Yet, no evidence was introduced showing how these discontinued practices have, or could have, interacted with SL 2013–381 to lessen minority opportunity to participate in the political process.

Plaintiffs also argue that SL 2013–381 itself enhances the opportunity for discrimination because the eliminated mechanisms "were effective tools in helping to reverse the persistent disparities in registration and participation." (Id. at 129.) Plaintiffs offered no such evidence in this case. And, in any event, that would be a rather circular method of employing the Gingles factors. This factor does not favor the Plaintiffs.e. Candidate Slating Process

The fourth Gingles factor asks, "if there is a candidate slating process, whether the members of the minority group have been denied access to that process." 478 U.S. at 37, 106 S.Ct. 2752. This factor does not appear to be relevant to any claim in this case, and, at any rate, no evidence was offered on it.

f. Continuing Effects of Discrimination Hindering Participation

The fifth Gingles factor, upon which Plaintiffs rely most heavily, examines "the extent to which members of the minority group in the state or political subdivision bear the effects of discrimination in such areas as education, employment and health, which hinder their ability to participate effectively in the political process." 478 U.S. at 37, 106 S.Ct. 2752.

In essence, this requires Plaintiffs to show both that (1) African Americans and/or Hispanics bear the effects of earlier discrimination, and (2) that those effects presently hinder their ability to participate effectively in the political process. Defendants contest both points. In support of their claim, Plaintiffs presented the following.

The evidence showed that African Americans and Hispanics are more likely to be unemployed and more likely to be poor¹⁴⁰ than whites.¹⁴¹ For the third quarter of 2014, 10.3% of African Americans and 8.1% of Hispanics were unemployed, compared to 5.3% of whites. (Pl. Ex. 229 at 11.) Similarly, as of 2013, 27% of African Americans and 43% of Hispanics were living below the poverty line, compared to 12% of whites. (Id.; see also Pl. Ex. 45 at 2, 6–7 (showing that African Americans and Hispanics are disproportionately likely to be poor).) These disparities are expressed in a slightly different way in the following table:

140 The poor obtain income from multiple sources, including wages and various welfare programs; there are "few racial differences in welfare receipt." (Pl. Ex. 45 at 14–15.)

141 With regard to those who are employed, Dr. Burden claimed that North Carolina law does not require employers to give employees time off to vote. For support, he cites a website of the League Plaintiffs. (Pl. Ex. 229 at 12 & n.54.) However, Dr. Burden and the website both ignore a North Carolina statute criminalizing retaliation against employees who wish to vote. SeeN.C. Gen. Stat. § 163–274(a)(6). At least one secondary source has specifically recognized this statute. SeeEugene Volokh, Private Employees' Speech and Political Activity: Statutory Protection Against Employer Retaliation, 16 Tex. Rev. L. & Pol. 295, 336–37 (2012).

Proportions of North Carolina's Adult Poor by Race & Ethnicity Race or Number of Percent of Percent of Adult Ethnicity Adult Poor Adult Poor Population White 559,397 50% 68% African American 351,262 31% 21% Hispanic 144,942 13% 7% Other 69,777 6% 4% Total 1,125,378 100% 100%

(Pl. Ex. 45 at 6–7.)

With regard to transportation, African Americans and Hispanics are less likely than whites to have access to a vehicle. Only 2.4% of whites live in homes without access to a vehicle, compared to 10.7% of African Americans and 6.4% of Hispanics. (Id. at 13–14.) That disparity becomes larger among the poor, as 27% of poor African Americans live in households without access to a car, compared to 8.8% of poor whites. (Id. at 14.)

As discussed above, African Americans and Hispanics are more likely to move than whites. (Id. at 17 ("For instance, 13.6 % of non-Hispanic whites reside in a different house than they did last year, compared to 18.5% for both non-Hispanic blacks and Hispanics."); Pl. Ex. 46 at 52 ("[A]n average of 17.1% of blacks moved within the state ... compared to only 10.9% of whites.").) This is consistent with the finding that minorities are disproportionately likely to be poor and the poor are more than twice as likely to move as the non-poor. (Pl. Ex. 45 at 17.)

African Americans and Hispanics in North Carolina also fare worse than whites in terms of health outcomes. (Pl. Ex. 229 at 11 (stating that 24% of African Americans and 29% of Hispanics have "fair" to "poor" health, compared to 16% of whites in North Carolina).) Further, the poor in North Carolina are more likely to be disabled, and among the poor, minorities are more likely to be disabled. (Pl. Ex. 45 at 2.)

African Americans and Hispanics are more likely to experience disparate educational outcomes than whites. Dr. Burden reported that among the 2011–2012 cohort, high school graduation rates in North Carolina were 85% for whites, 75% for African Americans, and 73% for Hispanics.¹⁴² (Pl. Ex. 229 at 12.) Plaintiffs' expert Allan Lichtman, Ph.D., Distinguished Professor of History at American University, put these figures at 88.5% for whites and 80.9% for African Americans. (Pl. Ex. 231 at 11.) By age 25, that gap narrows slightly, as 15.7% of African American North Carolinians lack a high school degree, compared to 10.1% of whites. (Pl. Ex. 45 at 10.) African Americans and Hispanics also drop out of school at higher rates and are suspended more frequently than whites. (Id. at 11.) They also have lower scores in both reading and mathematics on standardized tests. (Pl. Ex. 229 at 11.) Dr. Burden notes that "educational attainment is often the single best predictor of whether an individual votes." (Id. at 12.)

142 The poor have less education than the non-poor, and African Americans are more likely to be poor, but among the poor without college degrees, there are no disparities by race. (Pl. Ex. 45 at 10–11.) As Cynthia Duncan, Ph.D., a sociologist who studies poverty, testified at trial, "[T]he educational achievement between [poor] blacks and whites only differs statistically significantly at the college level." (Doc. 337 at 93.)

At trial, Dr. Clotfelter was tendered as an expert in the history and economics of education in North Carolina. He examined disparities between African American and white students in North Carolina in terms of educational resources and outcomes, past and present. (Doc. 330 at 127–28.) The court finds him to be a credible witness.

Dr. Clotfelter testified that historically North Carolina has contributed to its educational disparities in the following ways: slavery, segregated schooling, and purposeful differences in funding favoring whites during segregation.¹⁴³ (See Pl. Ex. 237.) Disparities in educational expenditures existed through at least the 1950s, although by some metrics earlier disparities had reversed course; for example, African American teachers were more highly paid than white teachers throughout the 1950s. (Id. at 15–16.) Compared to other southern States, North Carolina was "comparatively progressive" in its educational expenditures. (Id. at 16.)

143 Dr. Burden claims that North Carolina discriminates against minorities through its criminal justice system. (Pl. Ex. 229 at 13.) His observations about disparate prison statistics are not linked to discrimination or to any proof of discriminatory law enforcement, nor does he explain how such statistics interact with SL 2013–381. He also charges that North Carolina's felon disenfranchisement law disproportionately affects African Americans, (id.), but he fails to note that this law was passed by a bi-racial Republican Party as part of an omnibus election bill designed to help African Americans vote. Act of Mar. 8, 1895, ch. 159, § 13, 1895 N.C. Sess. Laws 211, 217; (Pl. Ex. 230 at 9–11.)

Nevertheless, disparities continue to the present day in various ways. Dr. Clotfelter found that disparities in education funding have continued after the end of de jure segregation. (Id. at 3–4.) For example, he observes that a child's educational success correlates with his parents' educational levels, which tends to perpetuate historic patterns of official discrimination even after official discrimination has ended. (Id. at 6–12.) An achievement gap still persists in North Carolina, though this gap compares favorably to the majority of other States. (See, e.g., Def. Ex. 268 at 49–50.)

Therefore, Plaintiffs have demonstrated that African Americans and Hispanics in North Carolina are disproportionately likely to move, be poor, less educated, have less access to transportation, and experience poor health. There was no showing that Hispanics suffer these as a result of historical discrimination. However, there was a showing that the socioeconomic disparities experienced by African Americans can be linked to the State's disgraceful history of discrimination. That history is more distant than it once was, and things have clearly improved, but the effects of historical discrimination against African Americans are assuredly linked by generations, and thus achievement in education and wealth can be tied to the success of one's parents and grandparents. Accordingly, this court finds that African Americans experience socioeconomic factors that may hinder their political participation generally. This is consistent with the Senate Report itself, which contemplates that socioeconomic disadvantage makes political participation more difficult.

There are, however, questions as to how these socioeconomic disadvantages endured by African Americans, which make political participation more difficult as a general matter, make, or are linked to making, it more difficult under SL 2013–381. Is it possible that if a State otherwise makes it easy enough to register and vote, existing socioeconomic disparities in this 21st century will prove immaterial? If not, could African Americans ever have had an equal opportunity prior to SL 2013–381, or in fact under any voting system? In addition, if socioeconomic disparities suggest African Americans will have a more difficult time participating, what is the court to make of the fact that African American participation increased in 2014 under SL 2013–381?¹⁴⁴ Plaintiffs did not account for this last question, but they did offer several witnesses in an attempt to link existing socioeconomic disparities to the experience under SL 2013–381.

144 White turnout still exceeded African American turnout in 2014 by .8% (40.5% to 39.7%), but that represents the smallest midterm voting differential for the years the court has been provided data. (Pl. Ex. 242 at 161 (App'x U).) The differential was 10.7% in 2006 (34.2% to 23.5%) and 3.4% in 2010 (39.6% to 36.2%)—when the removed mechanisms were in place. (Id.) Of course, African American turnout exceeded white turnout by 5.5% in 2008 (68.5% to 63%) and 6.8% in 2012 (67.2% to 60.4%). (Id.)

At trial, Plaintiffs presented Cynthia M. Duncan, Ph.D., a sociologist who studies poverty. Her testimony was credible but of limited helpfulness. She analyzed poverty in North Carolina by analyzing Census data for the State,¹⁴⁵ combined with information she gleaned from interviews with forty-seven poor North Carolinians of various races.¹⁴⁶ (Pl. Ex. 45 at 2.) Although Dr. Duncan explored the facts of poverty in North Carolina with reports from her interviews, she did not provide strong opinions on the causes of poverty in North Carolina. In fact, she explained about poverty scholarship very generally: "Scholars now recognize that larger structural factors such as discrimination, lack of jobs and racial and class segregation are intertwined with cultural factors such as failure to stay in school, having children young and out of wedlock, or getting involved with drugs and criminal activity." (Id. at 3.) In any case, she did not offer any quantitative analysis on any interaction between voting under SL 2013–381 and poverty. (Doc 337 at 90–91.)

145 Many of the poverty statistics set forth above were provided by Dr. Duncan.

146 The interviewees were not randomly selected. On cross-examination, Dr. Duncan preferred not to disclose how the interviewees were selected because her work was in conjunction with "some nonprofits" who helped select the interviewees, (Doc. 337 at 94–96), and the court did not require her to do so.

Plaintiffs presented the video deposition of Lynne Vernon-Feagans, Ph.D., a developmental psychologist and linguist tendered as an expert in non-urban poverty in North Carolina. She is the principal investigator for the Family Life Project and analyzed its data to see whether African American families in the data set would face greater obstacles "in meeting the requirements" of SL 2013–381 than non-African American adults in the study. (Pl. Ex. 240 at 1, 4.)

The Family Life Project is a study of families living in three rural North Carolina counties. (Id. at 8.) Families were recruited into the project upon the birth of a child (the researchers visited hospitals every day) from fall 2003 to fall 2004. (Id. at 8–9.) Data were collected through interviews conducted at eleven home visits over a seven-year period. (Id. at 9.)

Based on her analysis of the project's data, Dr. Vernon-Feagans opined that poor rural African American families were poorer than poor rural non-African American families. (Id. at 11.) She also opined that the African American families she studied were disproportionately likely to experience factors preventing them from participating in "civic life." (Id. at 5.) These include less access to transportation; less access to computers and internet services; non-standard work hours; racial discrimination; and "a greater number of negative life events (divorce, death in the family, major illness)." (Id. at 5–6.) They possessed demographic factors making participation in civic life more difficult: "single parenthood, lower levels of literacy, or multiple child care arrangements." (Id. at 6.) They also experienced more instability in living conditions—"more frequent residential moves"—and "less predictability in their daily lives" than non-African American families. (Id.) Ultimately, she concluded: "Together these multiple factors can disproportionately hinder African American adults from participating in community life and create barriers that can also prevent them from meeting the demands of North Carolina House Bill 589." (Id.) Specifically, she opined on what would be required of the studied women "in order to obtain a valid picture ID, to vote in a correct precinct, to come to a correct polling place during restricted hours, and to register to vote 25 days before an Election." (Id. at 5.) She opined that the addition of the voter-ID requirement and the elimination of early voting, SDR, OOP, and pre-registration would make it harder for African American families to vote. There are multiple problems with her opinion.

First, her opinion is based on SL 2013–381's version of the photo-ID requirement, which has been significantly amended by SL 2015–103 and the reasonable impediment exception. Second, she does not actually seem to understand SL 2013–381. It is unclear what Dr. Vernon-Feagans meant by "restricted hours" since SL 2013–381 has a same-hours requirement and the effect of increasing the number of hours available for each day of early voting and producing more convenient hours for lower-income persons. Third, Dr. Vernon-Feagans has never studied election law, voting behavior, or political participation rates. (Pl. Ex. 681 at 56.) The data she analyzed to reach her opinions was not designed to study any aspect of voting and includes no information about political participation, including the acts of registering and voting. (Id. at 56–57.) The court has no basis for believing that Dr. Vernon-Feagans understood the law before or after SL 2013–381, and therefore has no basis for accepting her opinions about the "obstacles" imposed by the law. Fourth, Dr. Vernon-Feagans' dataset and opinions are limited. Since she studied only rural counties, her data cannot be extrapolated to urban areas. Her families also were a limited subset: single mothers¹⁴⁷ and their children.¹⁴⁸

147 As Dr. Duncan explained, poverty is strongly correlated with marital status, with the unmarried more likely to be poor than the married. (Doc. 337 at 97–98.) This fact is true nationally. (Id.at 99.) No Plaintiffs' witness linked the burdens of this subset to official discrimination.

148 (E.g., Pl. Ex. 681 at 44 ("[S]o if we're looking at just poor families, there were no differences between our African-American moms and our non-African-American moms."); id.at 48 ("So what we have here as far as just comparing, again, African-American moms and non-African-American moms overall ...."); id.at 54 ("So all of these things together [the four provisions of SL 2013–381 she addressed] are going to make it harder especially for the poor African-American moms, at least that's what we have here....") (emphasis added)); Pl. Ex. 509 (examining access to vehicle for "all mothers" by race); Pl. Ex. 510 (examining access to technology for "all mothers" by race); Pl. Ex. 511 (examining literacy level for "mothers" by race); Pl. Ex. 512 (examining residential instability for "mothers" by race).

Fifth, Dr. Vernon-Feagans' definition of "civic life" was vague, meaning "activities that families may want to have access to in their community." (Pl. Ex. 681 at 64.) For her, this would include going to a parent-teacher conference or a doctor's appointment. (Id. at 16.) It did not appear to focus on how a citizen participates in and contributes to a political or social community's common welfare. Cf. Webster's Third New International Dictionary 412 (1986) (defining civic to refer to activities "forming a component of or connected with the function, integration, and development of a civilized community ... involving the common public activities and interests of the body of citizens").

Finally, Dr. Vernon-Feagans' opinion was merely predictive, it was not based on the actual registration and voting experiences of any of the subjects of her study under SL 2013–381 or any other voting procedure. In other words, neither she nor Plaintiffs presented any data on voter turnout and registration rates for her project participants or for the studied counties more generally, although such information is presumably available because the participants are known and Plaintiffs have access to North Carolina's voter database. It is certainly available for the rural areas studied. In summary, the court finds Dr. Vernon-Feagans' testimony of limited probative value.

Plaintiffs presented Gerald Webster, Ph.D., Professor of Geography at the University of Wyoming, as an expert in geography, political geography, and spatial analysis. Dr. Webster opined on the effect of the elimination of OOP voting on North Carolina's two largest counties, Wake and Mecklenburg. (Pl. Ex. 241 at 2.) There were a total of 842 OOP provisional ballots¹⁴⁹ cast in those two counties in the 2014 general election. (Id. at 43.) Dr. Webster examined the average distance between the precinct at which each of these voters voted on Election Day and their assigned precinct. (Id.) He found that, on average, an OOP voter in Mecklenburg County voted at a precinct that was 6.6 miles away from his or her assigned precinct; for Wake County, the average was 6.8 miles. (Id.) Since African Americans disproportionately cast OOP provisional ballots statewide, he assumed that African Americans disproportionately cast the OOP provisional ballots from Wake and Mecklenburg Counties. He therefore concluded that African Americans, who tend to have less household vehicle access, were disproportionately harmed by the elimination of OOP voting. (Id.)

149 There were only 1,930 incorrect precinct provisional ballots cast on Election Day in 2014. (Pl. Ex. 689.) Thus these 842 OOP provisional ballots from two counties represent 43.63% of all such ballots.

Dr. Webster's opinion is limited. He did not know the racial breakdown of the OOP voters he studied, nor did he study the voting history of any of the OOP voters (even though he was capable of doing so). (Doc. 334 at 190, 198.) Further, he did not consider whether these individuals had voted at the correct precinct in the past, nor did he know why these individuals voted at the incorrect precinct. (Id. at 198–99.) Most critically, he did not know whether any individual's correct precinct was closer to the voter's home or workplace, a key question in the burden analysis. (Id. at 198.)

In contrast, Defendants offered the opinion of Dr. Hofeller, who also studied the distance between the precinct where OOP voters cast their ballot and their correct precinct. (Def. Ex. 212A at 18–19.) In contrast to Dr. Webster, Dr. Hofeller examined statewide data and broke them down by race. (Id. at 18–19 & tbl. 23.) He found that 60.3% of African American OOP voters cast a ballot within five miles of their correct precinct. (Id. at 18.) By contrast, only 49.1% of white OOP voters cast a ballot within five miles of their correct precinct. (Id. at 18 & tbl.23.)

The court finds neither expert to provide very persuasive evidence on the issue. But Dr. Webster's approach was myopic, even relative to Dr. Hofeller's.

Finally, Plaintiffs presented the testimony of Kathryn Summers, Ph.D., a professor at the University of Baltimore, as an expert in literacy and voter system usability. She opined that SL 2013–381, especially the elimination of SDR, would burden low literacy voters, of which a disproportionate share are African American.

To reach her conclusion, Dr. Summers designed a usability study in Baltimore, Maryland. (Doc. 331 at 16.) Dr. Summers non-randomly recruited twenty African American, low-literacy participants, all from Baltimore, except for one, who was from Pennsylvania; some she recruited off the streets, others at a train station. (Id. at 38–39.) Each participant was paid $80 per hour and a half. (Id.) The participants were placed in a room with a computer, and Dr. Summers sat with them. (Id. at 39–40.) She then asked them to pretend they had just moved to North Carolina and to pretend there was an upcoming election in which they wished to vote, though no campaign issues were discussed. (Id. at 40.) The participants would then try to figure out how to register and vote. (Id. at 41–42.) Dr. Summers would role play, pretending to be someone at the SBOE, for example. (Id.)

The problems with the study are myriad. The most obvious is that Dr. Summers was studying Marylanders, not North Carolinians. Unsurprisingly, low literacy residents of Baltimore are not familiar with the electoral system of North Carolina.¹⁵⁰ Dr. Summers admitted that she could have conducted her study in North Carolina on actual North Carolinians, but choose not to do so because of the "time frame and the cost." (Id. at 42.) Additionally, she only studied twenty individuals and did not employ any control group of average-literacy participants in order to isolate the alleged impact of low literacy on voting.¹⁵¹ (Id. at 34.)

150 An additional problem is that such Marylanders cannot be said to have suffered from the effects of historical discrimination in North Carolina.

151 Further, although of lesser importance, Dr. Summers pretended to mimic various types of North Carolina government employees, but she had absolutely no experience working in any of those roles or knew what they provided. (Doc. 331 at 42–43.)

The details about the participants themselves are most telling. Dr. Summers opined at trial: "Specifically, the thing that would help—absolutely be most helpful above anything else would be an opportunity to do same-day registration." (Id. at 31.) But the question for the court is equality of opportunity, and, interestingly, nineteen of the twenty participants in her study were already registered Maryland voters. (Id. at 45.) Seventeen of the twenty had voted in the last midterm. (Id.) That these low-literacy participants were able to register and vote despite the fact that Maryland does not offer SDR significantly undermines her opinion. (Id.) Dr. Summers did not explain whether the other participants had in fact tried and failed to register or vote in Maryland.

Dr. Summers also criticizes North Carolina's registration system because several of her participants struggled to register to vote in North Carolina or never realized they needed to register. In particular, Dr. Summers criticizes the registration form itself and how it can be accessed. (Pl. Ex. 239 at 33–34.) Of course, SL 2013–381 did not change (nor do Plaintiffs challenge) the registration form—which is a simple, seven question, single page document similar to the federal form. (Doc. 341 at 159.) Even after the elimination of SDR, a citizen may still register at the CBOE and enlist the aid of staff. One may also register by mail and have a number of persons assist in completing the form. Finally, one may register after the twenty-five day cut-off; that registrant just cannot simultaneously vote in the upcoming election.

As Dr. Hood observes, it makes little sense for Dr. Summers to assume that low-literacy citizens cannot figure out how to register in time for an election but can figure out how to vote early. (Def. Ex. 268 at 52.) Early-voting sites are less numerous compared to precinct sites on Election Day, when interest in voting peaks. (Id.) A registered voter can determine where to vote because his precinct is listed on his voter registration card. (Id.) By contrast, how would a motivated citizen with low literacy be able to determine when and where to vote early if he cannot figure out how to register on time?¹⁵² The court does not accept Dr. Summers' conclusions. She was not credible, and her opinions are not well-founded.

152 Dr. Thornton makes a similar point. To the extent low literacy voters tend to be poor, they tend to include those who apply for government benefits. Based on her expertise as a labor economist, she explained the process and documentation needed for a person to apply for welfare benefits. (Def. Ex. 309 at 49–52.) While the court sustains Plaintiffs' objection to her testimony on any qualitative comparison of the processes, it is commonsense that North Carolina's one-page voter registration form cannot seriously be more intellectually demanding than that of gathering the relevant documents and applying for governmental benefits. Dr. Summers, in response, admitted as much but explained that low literacy voters can overcome the "complex and burdensome" welfare application process because welfare tends to provide a greater level of motivation than voting, whose benefits are "less tangible." (Pl. Ex. 251 at 7.)

Finally, on the issue of voter ID, Dr. Stewart presented socioeconomic data related to his first no-match list. (Doc. 408 and 53.) He analyzed the socioeconomic conditions of the zip codes and counties in which individuals on the no-match list disproportionately live and found that they were disproportionately in areas with lower socioeconomic conditions. (Pl. Ex. 242 at 52–54, 58.) Because Dr. Stewart did not actually have the wealth, income, or educational attainment of each individual voter, he "imputed" the socioeconomic characteristics of the voter's ZIP code to each voter. (Id. at 62 n. 76.) Based on this analysis, he found that the median household income for whites on the no-match list was $47,840, compared to $41,766 (- $6,074) for African Americans. (Id. at 62 (tbl. 13).) Per capita income for whites on the no-match list was $25,305, compared to $22,038 (-$3,267) for African Americans. (Id.) Median house value was $114,222 for whites, compared to $99,055 (-$15,167) for African Americans. (Id.) The percentage with at least some college was 57.9% for whites, compared to 54.2% (-3.7%) for African Americans. (Id.)

Dr. Stewart's data is of limited probative value for several reasons. First, it is based upon his first no-match list, which contained 397,971 individuals. (Id. at 38 (tbl. 7).) His second no-match list, which he relied upon at this court's January trial and described as his "best estimate," contains 173,108 fewer individuals. (Pl. Ex. 891 at 19 (tbl. 11).) Dr. Stewart did not provide socioeconomic data for the no-match list on which he actually relies. Second, while his analysis shows socioeconomic disparities, Plaintiffs failed to show that these disparities would make it materially more difficult to vote under the reasonable impediment exception. Plaintiffs did not offer expert opinion establishing what level of literacy is necessary to acquire ID at the DMV or to complete the reasonable impediment process. Thus, Dr. Stewart's finding that 3.7% more whites on the no-match list have "some college" than African Americans is not very helpful. Surely, someone with far less than "some college" could vote under the reasonable impediment exception, particularly in light of the substantial assistance available. Third, and finally, Dr. Stewart did not offer any opinion on the burden that remains after the enactment of the reasonable impediment exception. (Doc. 408 at 59.) Accordingly, his analysis fails to consider whether the reasonable impediment exception ameliorates any alleged burden that may have been imposed by SL 2013–381's photo-ID requirement, much less proves otherwise.

In sum, Plaintiffs established that some segment of the State's African Americans endure socioeconomic disparities that can be linked to State discrimination and this may make it more difficult for them generally to participate in any electoral system. Plaintiffs, however, failed to show that such disparities will have materially adverse effects on the ability of minority voters to cast a ballot and effectively exercise the electoral franchise after SL 2013–381 within the multitude of voting and registration options available in the State, especially given that the 2014 turnout data show increased participation among African Americans under SL 2013–381. SeeSalas, 964 F.2d at 1556 (finding that "the high incidence of Hispanic registration in the District is persuasive evidence that Hispanic voters are not deterred from participation in the political process because of the effects of prior discrimination, including unemployment, illiteracy, and low income"). For these reasons, this factor only favors Plaintiffs in a general and limited way.

g. Racial Appeals in Campaigning

The sixth Gingles factors asks "whether political campaigns have been characterized by overt or subtle racial appeals." 478 U.S. at 37, 106 S.Ct. 2752. Plaintiffs presented relatively little recent evidence on this factor.¹⁵³ (Pl. Ex. 46 at 53; Pl. Ex. 238 at 17–19; Pl. Ex. 230 at 34.) The evidence that was presented was often disconnected from actual campaigning. (See, e.g., Pl. Ex. 238 at 17 (Western Carolina University student's looped sign around neck of dead black bear cub); 17–18 (self-proclaimed patriot from Duplin County strung up effigies of President Obama and other State political leaders).) Nevertheless, Plaintiffs did present one undeniable racial appeal that is quite recent. (Pl. Ex. 230 at 34.) In 2010, the North Carolina Republican Party's Executive Committee distributed a mailer in which an incumbent Democrat was portrayed wearing a Sombrero with his skin darkened by photo editing. (Id.) The mailer had the Democratic candidate exclaiming "Mucho taxo," an apparent reference to policies Republicans claimed were driving away jobs. (Id.) However, the passage and enforcement of SL 2013–381 was not and has not been marked by subtle or overt racial appeals.

153 Plaintiffs point to one ad from Governor McCrory's gubernatorial campaign in 2012. In that ad, former sheriff of Wilson County, Wayne Gay, stated "Once a sheriff, always a sheriff. Once a Democrat, always a Democrat. Never voted any other way. Until now. North Carolina is a mess. Not getting better. Our only hope is Pat McCrory. Outsider. Fresh blood. Did great as mayor of Charlotte. He'll do great for our state. Or I'll hunt him down." Mark Binker, Black Caucus head criticizes McCroy ad featuring former sheriff, WRAL.COM (Sept. 25, 2012), http://www.wral.com/news/state/nccapitol/blogpost/11587710/. Plaintiffs claim that Gay's statement that "[o]ur only hope is Pat McCrory" begs the question of who is "our." (Pl. Ex. 238 at 18–19.) Plaintiffs have failed to show that "our" means anything other than North Carolina voters generally.

Even considering the 2010 incident, and even if the acts of Jesse Helms while running for senator in 1990 can be considered recent racial appeals, (Pl. Ex. 238 at 10–16), SL 2013–381 does not meaningfully interact with any of the alleged racial appeals to affect minority political participation. This factor and the scant anecdotal evidence offered in support of it are not very probative here. Veasey, 796 F.3d at 511 (finding evidence of racial appeals "not highly probative here," noting "it is not clear how such anecdotal evidence of racial campaign appeals combines with [the law] to deny or abridge the right to vote"). Accordingly, this court finds that this factor does not favor the Plaintiffs.

h. Minority Electoral Success

Under § 2, the "extent to which members of a protected class have been elected to office in the State or political subdivision is one circumstance which may be considered" among the totality of the circumstances. 52 U.S.C. § 10301(b) ; accordGingles, 478 U.S. at 37, 106 S.Ct. 2752. However, neither this factor nor any other provision of the VRA "establishes a right to have members of a protected class elected in numbers equal to their proportion in the population." 52 U.S.C. § 10301(b).

Plaintiffs argue that African Americans and Hispanics are underrepresented in elected offices in North Carolina, compared to their proportion of the population. They acknowledge that African Americans have recently approached parity with whites, though Hispanics remain underrepresented. (Pl. Ex. 229 at 14.)In North Carolina's General Assembly, there are eleven African American members of the Senate (22% of the Senators) and twenty-three members of the House (19% of the Representatives). (Id.; Pl. Ex. 682.) Plaintiffs note that earlier, in 1989, African Americans constituted only 8% of the Senate's members and 11% of the House's members. (Pl. Ex. 229 at 14.) Among North Carolina's elected constitutional members of the executive branch, there has only been one African American in North Carolina history. (Id.) Among North Carolina's congressional delegation, two members (15%) are African American. (Id.) Plaintiffs' evidence further showed that African Americans constitute the following proportions of other elected offices: 19.4% of the State's superior court judges; 23.8% of school board seats; and 12% of sheriffs. (Pl. Ex. 682.)

In response, Defendants note that African Americans constitute 62.5% of the Democratic Senators and 50% of the Democratic Representatives. (Def. Ex. 268 at 47.) Moreover, after SL 2013–381 was enacted, and under a Republican redistricting plan, the House gained an additional African American member. (Doc. 331 at 127–28, 144.)¹⁵⁴ Among the State's congressional delegation, African Americans represent two of the three Democratic-controlled House districts. (Def. Ex. 268 at 47.) In 2014, at the county level, African Americans held 19% of the total commissioner seats. (Id.) Regarding Hispanics, there is one member of the Senate and one of the House. (Id. at 48.) Therefore, compared to their shares of the registered voters in North Carolina and of the citizen voting age population, Hispanics are only slightly underrepresented in the North Carolina General Assembly. (Id. at 47–48 ("As of 2014, Hispanics comprised 2.7% of the citizen voting age population and 1.9% of the registrant population in North Carolina.").)

154 As noted above, North Carolina's redistricting scheme is currently subject to three lawsuits. Seecases cited supranote 138.

In sum, while minorities have historically been underrepresented in North Carolina, today African American electoral success, at least outside of State-wide races, approaches parity with their prevalence in the electorate. Hispanics remain underrepresented, but only slightly so in the General Assembly. Based on this evidence, and given that § 2 does not give minorities any claim to proportional representation, to the extent that this factor favors Plaintiffs, it does so only weakly.

i. Responsiveness of Elected Officials

The eighth Gingles factor asks "whether there is a significant lack of responsiveness on the part of elected officials to the particularized needs of the members of the minority group." 478 U.S. at 37, 106 S.Ct. 2752.

Plaintiffs suggest that any lingering socioeconomic disparities between whites and minorities proves a lack of responsiveness. (Doc. 346 at 6–7.) Plaintiffs point to no State policy for this assertion. If a significant lack of responsiveness can be established by pointing to socioeconomic disparities generally, without accounting for specific State policies, then every State would be significantly unresponsive because socioeconomic disparities are unfortunately widespread. The court finds Plaintiffs' reliance on other Gingles factors unpersuasive evidence on this factor.

Plaintiffs also assert that the enactment of SL 2013–381 demonstrates a lack of responsiveness. (Doc. 346 at 129.) Citing Veasey, 796 F.3d at 511, they argue that legislators enacted SL 2013–381 by refusing to consider ameliorative amendments offered by the Democratic opponents of the bill. (See id. at 52, 129.) This, of course, overlooks that Republican supporters of the bill accepted the same-hours amendment, a central aspect of the new early-voting schedule, from a Democratic Senator who opposed all of SL 2013–381. Plaintiffs also ignore that SL 2013–381 exempts curbside voters (of which African Americans make up a disproportionate share, (Doc. 332 at 110–12)) from compliance with the photo-ID requirement by permitting them to vote with certain documents showing their name and address, such as a utility bill. N.C. Gen. Stat. §§ 163–166.12(a)(2), 163–166.13(a)(1).

The court declines to infer a significant lack of responsiveness on this record, especially where there is evidence of historical policy disagreements on several provisions. Moreover, besides this law, Plaintiffs have failed to point to any specific "particularized" need of African Americans and Hispanics in North Carolina about which the State has been unresponsive. Indeed, Dr. Leloudis, when pressed at trial, was unable to articulate anything the General Assembly needed to do to be more responsive to the needs of minorities, other than a generalized statement of "equitable inclusion in the political process." (Doc. 338 at 47.)

j. Tenuousness of the State's Justifications

The last Gingles factor requires the court to assess "whether the policy underlying the state or political subdivision's use of such voting qualification, prerequisite to voting, or standard, practice or procedure is tenuous." 478 U.S. at 37, 106 S.Ct. 2752. Tenuous is generally defined as "having little substance or strength." Webster's Third New International Dictionary 2357 (1986). In LULAC, the Fifth Circuit defined a non-tenuous law within the context of the VRA as one that is "not a pretext masking discriminatory intent." 999 F.2d at 870. When "determining whether a § 2 violation has occurred," the Supreme Court has held, "[a] State's justification for its electoral system is a proper factor for the courts to assess." Houston Lawyers' Ass'n, 501 U.S. at 426–27, 111 S.Ct. 2376 ; see alsoLULAC, 999 F.2d at 871 ("The weight, as well as tenuousness, of the state's interest is a legitimate factor in analyzing the totality of circumstances."). Each provision of the law will be addressed in turn.

i. Voter ID

The stated purpose of North Carolina's voter-ID law is to ensure that the in-person voter and the registrant are the same person. 2013 N.C. Sess. Law 381, preamble ("An Act ... to provide photo identification before voting to protect the right of each registered voter to cast a secure vote with reasonable security measures to confirm voter identity as accurately as possible without restriction"). Because North Carolina does not require the address on a voter's ID to match the address of registration, the voter-ID law is exclusively designed to prevent in-person voter impersonation fraud.

Nineteen States currently have some form of a photographic identification requirement for voting. (Def. Ex. 270 at 17.) At the time North Carolina enacted its voter-ID requirement, the Supreme Court had upheld a challenge to Indiana's ID law in Crawford. In Crawford, the Supreme Court found that Indiana's photo-ID law furthered the State's legitimate interest in modernizing elections, deterring voter fraud, and safeguarding voter confidence in the integrity and legitimacy of elections. 553 U.S. at 192–97, 128 S.Ct. 1610. The legislative history of SL 2013–381 and the law itself indicate that the North Carolina General Assembly sought to achieve these same ends. (Pl. Ex. 549 at 2–4 ("North Carolina is one of the last in the Southeast to introduce [photo ID] for honesty and integrity in the electoral process and we believe it will go a long way to building confidence back in our voters and our citizens.")); 2013 N.C. Sess. Law 381 (stating that its purpose was "to protect the right of each registered voter to cast a secure vote with reasonable security measures that confirm voter identity as accurately as possible without restriction"); 2015 N.C. Sess. Law 103 (stating that its purpose was to "authorize voters who suffer from a reasonable impediment preventing the voter from obtaining photo identification to complete reasonable impediment declarations when voting").

Plaintiffs argue that there was no evidence of voter impersonation fraud in North Carolina. The same can be said of Indiana—there being no evidence of in-person voter impersonation fraud occurring at any time in its history, yet the Supreme Court recognized in Crawford that voter fraud has occurred in other jurisdictions and that "not only is the risk of voter fraud real but ... it could affect the outcome of a close election." 553 U.S. at 194–96, 128 S.Ct. 1610. In reaching its conclusion, the Court in part relied upon the findings and recommendations of the Commission on Federal Election Reform. Seeid. at 193–97, 128 S.Ct. 1610.

In 2005, the Commission on Federal Election Reform, chaired by President Jimmy Carter and former Secretary of State James A. Baker, III, recommended that States implement "a uniform system of voter identification" as one of its five pillars of "modernizing" and "[b]uilding confidence in U.S. elections." See Commission on Federal Election Reform, Building Confidence in U.S. Elections iv (Sept. 2005) (hereinafter, "Carter-Baker Report"). Even though the Commission found that there was not "evidence of extensive fraud in U.S. elections or of multiple voting," it found that such fraud does occur and that "it could affect the outcome of a close election." Id. at 18. The Commission reasoned that "[t]he electoral system cannot inspire public confidence if no safeguards exist to deter or detect fraud or to confirm the identity of voters." Id. Likewise, the Commission recognized that changes in American life require our electoral systems to be modernized.

In the old days and in small towns where everyone knows each other, voters did not need to identify themselves. But in the United States, where 40 million people move each year, and in urban areas where some people do not even know the people living in their own apartment building let alone their precinct, some form of identification is needed.

Id.

Despite the bipartisan Commission's finding that ID is needed and the Supreme Court's ruling that photo ID serves legitimate interests, Plaintiffs nevertheless claim that North Carolina's voter-ID requirement is tenuous. Plaintiffs do not hide the fact that this is in essence a challenge to Crawford itself.

In support, Plaintiffs presented Lorraine Minnite, Ph.D., Associate Professor of Public Policy and Director of the Urban Studies Program in the Department of Public Policy and Administration at Rutgers University, who asserts that voter impersonation fraud "has been statistically nonexistent in North Carolina since 2000." (Doc. 409 at 21.) Between 2000 and 2014, Dr. Minnite found, the SBOE had only referred two cases of voter impersonation to district attorneys. (Id. at 20.) There is no evidence that either of these referrals resulted in a prosecution or conviction. (Id. at 21.) Two additional referrals have been made since 2014. (Doc. 410 at 138–39.) These referrals remain pending, but no charges had been filed at the time of this court's January trial. (Id.) Even Dr. Minnite conceded at trial, however, that the risk of voter fraud is "real in the sense that it could happen" and while "[t]here is no evidence of extensive fraud in U.S. elections or of multiple voting, ... both occur and it could affect the outcome of a close election." (Doc. 337 at 49–51.) She nevertheless discounts it, however, "in the sense that it likely happens." (Id. at 49.)

But even if there is no evidence of voter impersonation fraud in North Carolina, there are two problems with Plaintiffs' argument. The first is that the signature attestation process that North Carolina relied on prior to its ID law provided poll workers with limited tools to detect fraud. Dr. Minnite lauded signature attestation as a "very good system because a signature is a biometric identifier." (Doc. 409 at 37.) The problem is that poll workers did not have access to voter's signatures so that a comparison could be made. (Doc. 414 at 123–24.) Unless the poll book indicated that someone had already voted under the name provided, (see Doc. 410 at 90), the only way the poll worker could recognize an impersonator was if they either knew the person presenting or knew the person whose name was provided, (see Doc. 414 at 123–24). This of course assumes a small world that, as the bipartisan Commission recognized, no longer exists. Carter-Baker Report, at 18. Back when all voting occurred on Election Day at precincts with small numbers of voters, poll workers' knowledge of the community may have been sufficient to meaningfully guard against impersonators. See id. But as precincts have become larger and early voting at centralized locations has become more pervasive, the risk that voter fraud may go undetected has increased. (See Pl. Ex. 983 at 63 n.91 (Government Accountability Office report, stating that "[l]ike other crimes, instances of in-person voter fraud may occur that are never identified by or reported to officials. This is due, in part, to challenges associated with identifying this type of fraud, as both successful fraud and deterred fraud may go undetected.").)

Second, and most significantly, Crawford found photo ID to serve legitimate State interests even though there was "no evidence of any [in-person voter impersonation] actually occurring in Indiana at any time in its history." 553 U.S. at 195, 128 S.Ct. 1610. The only plausible reading of Crawford is that there need not be evidence of voter impersonation within a State for that State to have a legitimate interest in having a photo-ID requirement. Seeid. at 192–97, 128 S.Ct. 1610.

It is clear from Dr. Minnite's testimony that she simply disagrees with the Supreme Court. She did not even reference Crawford in her report, instead citing a Missouri Supreme Court case predating Crawford. (Doc. 337 at 47–48; Pl. Ex. 232 at 20.) At the July trial in this case, she testified that the United States Supreme Court's discussion of fraud "doesn't constitute an informed opinion or an informed knowledge about voter fraud" because "it doesn't sort of meet my standards of having a correct understanding about the evidence." (Doc. 337 at 48.) Further, at trial in January, despite recognizing that Crawford is still good law, Dr. Minnite explained that she views circumstances since Crawford as having cast a "pall over [it] as being something that we would cite to." (Doc. 409 at 24.) Principally, Dr. Minnite pointed to a change of position by Judge Richard Posner of the Seventh Circuit Court of Appeals, who drafted that court's majority opinion that was affirmed by the Supreme Court in Crawford. (Id. at 25.) After the Seventh Circuit upheld Wisconsin's voter-ID law in Frank v. Walker, 768 F.3d 744 (7th Cir.2014), Judge Posner requested a vote on whether to rehear the case en banc. Frank v. Walker, 773 F.3d 783, 783 (7th Cir.2014). Rehearing en banc was denied by an equally divided court. Id. In his dissent from the denial of rehearing en banc, Judge Posner asserted that Wisconsin's interest in deterring fraud was insufficient to justify the burden imposed by the ID requirement. Id. at 788, 797. This opinion was in part based on Judge Posner's finding that "voter-impersonation fraud is essentially non-existent in Wisconsin." Id. at 788. Dr. Minnite testified that she found Judge Posner's reasoning to imply that he believes Crawford was wrongly decided, but Judge Posner stops short of saying so. Seeid. at 784–86. In fact, Judge Posner went to significant lengths to distinguish Crawford. Id. In his view, the structure of Wisconsin's ID law was likely to place a heavier burden on voters than the Indiana law that was sustained in Crawford. Id.

There are significant problems with Dr. Minnite's reliance on Judge Posner's dissent. First, to the extent that Judge Posner's views were shaped by the specific stringency of Wisconsin's ID requirement, North Carolina's reasonable impediment exception makes its ID requirement less burdensome than either Indiana's or Wisconsin's—a fact that Dr. Minnite does not address. Seeid. at 784–85 ; (Doc. 409 at 24–26). Although indigent Indiana voters were permitted to vote without an ID so long as they completed an affidavit confirming their identity and indigence, neither Indiana's nor Wisconsin's ID laws had a reasonable impediment exception. SeeFrank, 773 F.3d at 784–86. Second, and most obviously, Judge Posner's views did not prevail, and Frank remains good law in the Seventh Circuit. SeeFrank, 768 F.3d 744. Third, while Judge Posner's views may shape the Supreme Court's approach in a future case, Crawford remains the law of the land and is binding on this court.

Plaintiffs next assert tenuousness on the ground that North Carolina's photo-ID law does not apply to absentee mail voting. According to Dr. Burden, voter fraud is more likely to occur with absentee mail voting than with in-person voting. (Doc. 407 at 54–55.) Dr. Burden thus reasoned that if North Carolina was truly concerned with voter fraud, it would apply its photo-ID requirement to absentee mail ballots rather than in-person voting. (Id. at 100–01.) Dr. Burden's testimony did not provide clarity on how practical it would be to require photo ID of absentee mail voters. He did, however, claim that his home state of Wisconsin requires absentee mail voters to show photo identification. (Id. at 101.) This is true, but not entirely accurate.

Wisconsin requires individuals requesting absentee ballots to present photo ID by mailing "in a photocopy of an acceptable photo ID with his or her request." Frank v. Walker, 17 F.Supp.3d 837, 844 (E.D.Wisc.2014) (citing Wis. Stat. §§ 6.86(1)(ar), 6.87(1), 6.87(1) ), rev'd on other grounds, 768 F.3d 744 (7th Cir.2014). This requirement, however, does not apply every time the voter casts an absentee mail ballot. See id. Wisconsin's photo-ID requirement does not apply to "absentee voters who have previously supplied acceptable photo IDs and whose names and addresses have not changed." Id.(citing Wis. Stat. § 6.87(4)(b) ). Nor does it apply to absentee voters who are in the military or oversees, or those "who are elderly, infirm, or disabled and indefinitely confined to their homes or certain care facilities." Id.(citing Wis. Stat. §§ 6.87(1), (3), 6.86(2), 6.875 ). Indeed, if Wisconsin's photo-ID requirement applied to every absentee voter every time he voted, the law might be even more burdensome on voters. Judge Posner concluded as much in his dissent in Frank, 773 F.3d at 785 (finding that the law's application to absentee voters as one reason for it being more burdensome that the Indiana law upheld in Crawford). In any case, Indiana's exception for absentee mail voters did not prevent the Supreme Court from finding that its photo-ID law served legitimate State interests. Crawford, 553 U.S. at 185–86, 192–97, 128 S.Ct. 1610 ; see alsoid. at 209, 128 S.Ct. 1610 (Scalia, J., concurring) ("That the State accommodates some voters by permitting (not requiring) the casting of absentee or provisional ballots, is an indulgence—not a constitutional imperative that falls short of what is required.").

States sometimes tread a difficult path, as they balance their interest in secure elections with protecting a voter's right to participate equally. The balancing act inherent in determining which requirements to impose and which to omit is why legislatures are given substantial leeway in determining when and how to address and remedy various challenges. Cf.Williamson v. Lee Optical of Okla., Inc., 348 U.S. 483, 489, 75 S.Ct. 461, 99 L.Ed. 563 (1955) ; Common Cause/Georgia v. Billups, 504 F.Supp.2d 1333, 1381–82 (N.D.Ga.2007) ("[T]he legislature has wide latitude in determining the problems it wishes to address and the manner in which to address them (quoting Indiana Democratic Party v. Rokita, 458 F.Supp.2d 775, 829 (S.D.Ind.2006) ), vacated on other grounds and order reentered, 554 F.3d 1340 (11th Cir.2009).

Moreover, while certain mechanisms may be more secure than others, SL 2013–381 and SL 2015–103 raised the security of both in-person and absentee mail voting. As noted above, North Carolina relied upon a system of signature attestation for in-person voting prior to SL 2013–381. (Doc. 410 at 83.) For voter impersonation fraud to occur, the impersonator only needed to acquire the voter's name and address of registration. (Doc. 410 at 89–90.) Both of these pieces of information—plus the voter's registration status, county of registration, and polling location—can be acquired by using the "NC Public Voter Search" feature on the SBOE's Website. NC Public Voter Search, N.C. State Bd. of Elections, https://enr.ncsbe.gov/voter_search_public/ (last visited April 11, 2016). The impostor also had to sign an ATV form attesting to his or her identity and address under penalty of perjury, (Doc. 410 at 83), but poll workers did not have the voter's registration signature to compare with the impostor's signature, (Doc. 414 at 123–24).

After SL 2013–381 and SL 2015–103, the at least 96.5% of North Carolinians who have a qualifying photo ID will be required to present it in order to vote in person. (Pl. Ex. 891 at 19 (tbl. 11).) Curbside voters, although exempt from the photo-ID requirement, must still present either a qualifying photo ID or a HAVA document bearing the voter's name and address. N.C. Gen. Stat. §§ 163–166.13(a)(1), 163–166.9, 163–166.12(a). Those who have an impediment to acquiring qualifying ID must, in addition to attesting to their identity and address, provide a HAVA document, their registration card, or their date of birth and SSN4. Id.§ 163–166.15(c). Therefore, even if an impostor seeks to use the reasonable impediment exception to commit fraud, he will at least need to acquire an acceptable HAVA document, the voter's registration card, or two pieces of personal information on the voter: the voter's date of birth and SSN4. Id.

Session Law 2013–381 also raised the security of absentee mail ballots. To request an absentee ballot today, absentee voters must provide either (1) their driver's license, learner's permit, or provisional license number; (2) their special identification card for nonoperators card number; or (3) their SSN4. N.C. Gen. Stat. § 163–230.2(a)(4). This requirement did not exist prior to SL 2013–381.¹⁵⁵See 2013 N.C. Sess. Law 381, § 4.3. In addition, absentee mail voters must complete their ballot and the absentee process in the presence of two witnesses who are at least eighteen years of age. N.C. Gen. Stat. § 163–231(a). The two witnesses must provide their name, address, and sign under penalty of perjury. Id.; (see also Doc. 410 at 101). In the alternative to the two-witness requirement, absentee voters may have their ballot witnessed by a notary public. N.C. Gen. Stat. § 163–231(a). Prior to SL 2013, the absentee mail voting process only required one witness. 2013 N.C. Sess. Law 381, § 4.4; (see also Doc. 410 at 101).

155 Both before and after SL 2013–381, voters who "registered to vote by mail on or after January 1, 2003, and [have] not previously voted in an election that includes a ballot item for federal office in North Carolina" must, "in order to cast a mail-in absentee vote," provide either (1) a valid photo-ID or (2) a HAVA document. N.C. Gen. Stat. § 163–166.12(b). This requirement, however, only applies to those who register by mail and who have not voted previously in certain elections. Seeid.The requirements of N.C. Gen. Stat. § 163–230.2(a)(4) apply each time an absentee ballot is requested and regardless of the method of registration.

In sum, despite making significant accommodations for voters, SL 2013–381 and SL 2015–103 nevertheless raised the level of security for both in-person and absentee mail voting. This has provided poll workers and CBOEs with a larger toolset with which to detect and deter voter fraud. Accordingly, North Carolina's ID law is not tenuous and serves legitimate State interests articulated by the Supreme Court in Crawford.

ii. Early Voting

Plaintiffs contest the General Assembly's justifications for reducing the number of early-voting days from seventeen to ten.

The General Assembly made a number of alterations to early voting with SL 2013–381. While deciding to reduce the early-voting period to ten days, it still required that CBOEs offer the same number of aggregate hours of early voting as offered in 2010 for future non-presidential elections and as offered in 2012 for presidential elections. N.C. Gen. Stat. § 163–227.2(g2). The General Assembly also permitted counties to seek a waiver from the aggregate-hours requirement but required a unanimous vote in favor of the waiver from a bipartisan CBOE and a bipartisan SBOE. Id.§ 163–227.2(g3). Considered together, a reduction in hours can be obtained only if there is agreement among five members of the majority party and three members of the minority party. N.C. Gen. Stat. § 163–19. If there is any dissent, the waiver request is vetoed. (See Doc. 340 at 204–05.) Moreover, while both before and after SL 2013–381, a CBOE could, by unanimous vote of its three members, open additional early-voting satellite sites, N.C. Gen. Stat. § 163–227.2(g), SL 2013–381 ensured greater uniformity, and perhaps fairness, by requiring that every satellite site "provide for the same days of operation and same number of hours of operation on each day for all [satellite] sites in that county for that election," id.

Defendants offered multiple justifications for the changes to the early-voting schedule. Proponents of the reduction of days of early voting, even before the same-hours requirement was introduced, stated an intent to free up resources so that CBOEs could offer more early-voting sites across each county. (See, e.g., Pl. Ex. 202 at 31–32 ("[T]here is some savings in [the removal of the seven days] to open up additional sites ... [y]ou don't have to go across town if it's not in your neighborhood."); Pl. Ex. 549 at 2–8.) Proponents expressed a concern that the first seven days of early voting were the least used days, (Pl. Ex. 202 at 31), which was also true. There was also a stated concern that early-voting sites had been located based on political gamesmanship, so that having more sites spread more evenly across counties, open for longer hours, would negatethis influence. (See, e.g., Pl. Ex. 202 at 31–32, 75; Pl. Ex. 549 at 2–8; Pl. Ex. 550 at 55–57.) This was a legitimate concern. (See, e.g., Def. Ex. 212A at 14–16, 19 (finding that the placement of and hours offered for early-voting sites favored Democrats over Republicans).) In fact, as part of the SBOE's 2008 report on SDR, former SBOE Director Gary Bartlett, a twenty-year veteran SBOE Executive Director and Democratic appointee, recommended that the legislature "[d]evelop an equitable plan to allot county one-stop sites that diffuses any claims of partisanship." (Pl. Ex. 56 at 7.)

However, while SL 2013–381 mandated uniformity in the hours of satellite early-voting sites, without the same-hours requirement the increase in the early-voting sites would have been entirely at the discretion of each CBOE. Senator Stein, a Democrat, proposed the same-hours amendment, which he believed would compel CBOEs to offer more early-voting sites and extended hours. (Pl. Ex. 549 at 16–18; Pl. Ex. 550 at 11.) He said on the floor of the Senate, "So if you all want to ensure that the citizens of this state have the same access to early voting that they have today with ten days as they have for seventeen days, you will support this amendment and I encourage you to do so." (Pl. Ex. 549 at 17.) The proponents apparently agreed with Senator Stein that they wanted to ensure the same amount of access to early voting, while also increasing the convenience of early voting for each day of the early-voting period, because the amendment passed by a vote of forty-seven to one. (Id. at 49.)

And, in fact, the modified early-voting schedule worked just this way: in 2014, not only were there more early-voting sites than in 2010, the previous midterm, but even more than in 2012, the previous presidential election. (See, e.g., Doc. 340 at 205–06.) There were also new, more convenient hours offered outside the typical 9 a.m. to 5 p.m. window. (See id.; Doc. 335 at 80; Def. Ex. 13; Pl. Ex. 242 at 167–68.)

Plaintiffs frame the law as no more than a thinly-veiled effort to provide less access to the polls by reducing the number of days of early voting. But this oversimplifies the nature of the changes in the law. Without the same-hours amendment, the court may well have had serious concerns about the change, particularly as early voting appears to have gained growing popularity among the electorate. However, the new law offers the same aggregate hours as before; provides for more polling places rather than fewer, so that different parts of a county are more equally served; establishes longer, more convenient hours of early voting for each day of early voting, to accommodate typical work schedules; and promotes bipartisanship in the placement and hours of early-voting sites. It does this by reallocating the resources devoted to the first seven days of early voting—those farthest from Election Day and the least popular period of early voting. In a world of budgets and limited resources, the General Assembly's stated concerns and the means chosen to address them had a reasoned basis at the time; and, as history has demonstrated, the elections of 2014 have confirmed the proponents' contentions and rebutted Plaintiffs' fears.

Reasonable minds may differ as to the most desirable early-voting system.¹⁵⁶ Some may urge even more early voting availability; and, as the evidence at trial revealed, there are non-frivolous reasons why a State may wish to curtail or even eliminate early voting. Substantial scholarship, by Plaintiffs' own experts, supports the conclusion that early voting actually dilutes interest in elections and depresses turnout. There was also no evidence presented that the earlier system was implemented following any empirical or quantitative analysis of perceived needs. The new schedule may also prove to have flaws, although it clearly has advantages over the old one. But such policy choices are for elected bodies; the question here is tenuousness. In light of the trade-off between days and hours, especially where at least one three-judge federal court endorsed a nearly identical reduction and the Attorney General did not object, seeBrown, 895 F.Supp.2d at 1241–42, the court does not find the rationale for the new system to be tenuous. See alsoid. at 1256 (finding no § 2 denial of equality of opportunity in Florida's reduction of early-voting days from 12–14 to 8).

156 For example, New York, one of the Nation's most populated States containing former § 5 jurisdictions, has no early voting whatsoever.

Plaintiffs' arguments otherwise fall short. Plaintiffs describe the aggregate-hours amendment as a "freezing" of the early-voting hours offered in a county. (Doc. 346 at 67.) This is a mischaracterization of the law. As described by its sponsor and Plaintiffs' own witness, Senator Stein, the law is nothing more than a "floor," (Pl. Ex. 549 at 48), or "baseline minimum," (Pl. Ex. 550 at 11), on the number of hours counties must offer. Counties, however, "can clearly offer more [hours]." (Id.)

Plaintiffs criticize the cost savings justification originally offered for the reduction in early-voting days. (Doc. 346 at 67–68.) But, by "savings," the proponents' reasons included an opportunity to reallocate resources, not just a net monetary savings. (Pl. Ex. 202 at 30–31 ("There is probably a savings in the sense that by going from seventeen to the ten days you actually have more opportunity to open up new sites ....").) As noted above, the primary justification for the change in the early-voting schedule was more polling places and greater parity and uniformity in early-voting site accessibility. (See, e.g., id.) If the proponents' aim was to decrease overall costs, it is doubtful the General Assembly would have adopted the same-hours requirement by a vote of forty-seven to one. However, given the actual justifications offered by the law's proponents, it is not surprising that Senator Stein's amendment was so overwhelmingly approved. When combined with the ten-day schedule, the same-hours requirement furthered the interests of the law's proponents by de facto requiring more polling places and more convenient hours.

For these reasons, because the ten-day early-voting schedule with the same-hours requirement promotes a significant State interest by making voting hours more convenient and polling locations more numerous and evenly distributed, it is not tenuous.

iii. SDR

Plaintiffs argue that Defendants' justification for repealing SDR is tenuous because, they say, there was no evidence that SDR caused any significant administrative problem; thus, the attempt to fix a non-existent problem was pretextual. (Doc. 346 at 68.) Defendants argue that the General Assembly had ample evidence before it that SDR created several administrative problems that ultimately resulted in ineligible voters having their votes counted, effectively disenfranchising other eligible voters.

The evidence before both this court, and the General Assembly back in 2013, makes clear that SDR—which effectively moved the twenty-five day registration cut-off to up to three days before Election Day—brought with it several administrative problems and increased the likelihood that the State's verification process could not be implemented. Indeed, Plaintiffs do not seriously dispute this latter point. Rather, at trial they acknowledged that SDR voters failed the State's verification system, but Plaintiffs argue that SDR voters did so at rates comparable to, if not less than, that of ordinary registrants. Thus, Plaintiffs argue that the law's sponsors must have had ulterior motives in eliminating SDR. To properly resolve the contentions, it is necessary to understand North Carolina's system for verifying voter registrants, about which there was considerable evidence and debate at trial.

Under North Carolina law that predates SDR and its repeal, CBOEs are required to verify that only qualified voters are registered to vote. N.C. Gen. Stat. § 163–82.7. To qualify to vote in North Carolina, a person must (1) be born in the United States or a naturalized citizen, (2) be eighteen years old, and (3) "have resided in the State of North Carolina and in the precinct in which the person offers to vote for 30 days next preceding an election." N.C. Gen. Stat. § 163–55(a).¹⁵⁷ North Carolina law mandates a "mail verification" process to confirm a voter's residence and thus precinct, a process that is interwoven into multiple sections of the General Statutes and one which Plaintiffs do not challenge in this case. (Doc. 336 at 114; Doc. 340 at 207.) Mail verification must be completed before a person is considered "registered" to vote in North Carolina. (Doc. 340 at 207–08.)

157 North Carolina law disqualifies felons from voting unless their rights of citizenship have been restored as prescribed by law. N.C. Gen. Stat. § 163–55(a)(2).

The mail verification process begins when a CBOE receives a registration application. SeeN.C. Gen. Stat. § 163–82.6. Upon receipt, the CBOE first makes a preliminary determination whether the applicant is qualified to vote. N.C. Gen. Stat. § 163–82.7(a). North Carolina law does not require a CBOE to make this preliminary determination by a specific time, but SBOE Director Strach testified that it should begin "as soon as [the CBOE] receive [s]" the application. (Doc. 336 at 104.)¹⁵⁸ If the CBOE determines that the applicant is not qualified to vote, it sends, by certified mail, a notice of denial of registration to the applicant. N.C. Gen. Stat. § 163–82.7(b). This initial decision is appealable. Id.

158 When SDR was in place, CBOEs had to make a preliminary determination of an SDR applicant's qualifications within two business days of receiving the registration. 2007 N.C. Sess. Law 253, § 1.

If the CBOE tentatively determines that the applicant is qualified, it initiates the "mail verification notice procedure." Id.§ 163–82.7(a)(2). This begins with the CBOE sending "a notice to the applicant, by nonforwardable mail, at the address the applicant provides on the [voter registration] application form." Id.§ 163–82.7(c). The notice includes the applicant's assigned precinct and voting place and states that the county will register the applicant "if the [United States] Postal Service does not return the notice as undeliverable to the [CBOE]." Id. If the notice is not returned as undeliverable, the CBOE registers the applicant to vote. Id.§ 163–82.7(d). Under SBOE policy, the CBOE must do so if the notice is not returned as undeliverable within fifteen days after mailing. (Doc. 336 at 111, 121–22.)

If the notice is returned as undeliverable, however, the registrant's status remains unverified, and the statute requires the CBOE to repeat the process by sending a second notice by nonforwardable mail to the same address. N.C. Gen. Stat. § 163–82.7(e). If the second notice is not returned as undeliverable, the applicant is registered to vote. Id. If, however, the second notice is returned as undeliverable, the applicant's registration is denied. Id.§ 163–82.7(f) ; (Doc. 336 at 126; Doc. 340 at 208). As before, SBOE policy requires the CBOE to register the applicant to vote if the second mail verification notice is not returned as undeliverable within fifteen days of mailing. (Doc. 336 at 112.)

Voting complicates the mail verification process. Under the statute, "If the [CBOE] has made a tentative determination that an applicant is qualified to vote ..., then that person shall not be denied the right to vote in person in an election unless the Postal Service has returned two notices as undeliverable."¹⁵⁹ N.C. Gen. Stat. § 163–82.7(g)(1). SDR changed the calculus. Once the CBOE has made a tentative determination that an applicant is qualified to vote and the applicant votes, the process of denying the applicant's vote is complicated. Before the applicant votes, his registration can simply be denied. Id.§ 163–82.7(f). After the registrant votes, his vote counts unless it is challenged on Election Day under N.C. Gen. Stat. § 163–89, even if the first notice has been returned undeliverable at the time an applicant votes.¹⁶⁰Id.§ 163–82.7(g)(2) ; (Doc. 341 at 163; Doc. 336 at 113). Accordingly, the statutory scheme—developed before SDR, when the State had at least twenty-five days to discharge its statutory obligation to mail verify each traditional registrant—was designed to accommodate voting while verification was ongoing, but contemplated resolving verification questions before a vote was cast and certainly before it was counted.

159 If a notice has been returned undeliverable "within 25 days before the election," then the applicant "may vote only in person in that first election and may not vote by absentee ballot except in person [as provided under N.C. Gen. Stat. § 163–227.2 ]" under procedures developed by the CBOE to obtain the correct address and assure that the person votes in the proper place and in the proper contests. N.C. Gen. Stat. § 163–82.7(g)(2).

160 A first or second notice returned undeliverable after an applicant votes requires that the applicant be treated as an inactive registered voter and undergo a separate confirmation mailing process. N.C. Gen. Stat. § 163–82.7(g)(3). The confirmation mailing process is part of "list maintenance," which entails confirming that a voter on North Carolina's voting rolls is still eligible to vote. (Doc. 340 at 209–10.) Unlike with the mail verification process, a confirmation mailing is forwardable mail. N.C. Gen. Stat. § 163–82.14(d)(2) ; (see alsoDoc. 340 at 209). The confirmation mailing process, however, requires a voter to return the mailing to the appropriate CBOE, not simply that the mailing be delivered. N.C. Gen. Stat. § 163–82.14(d)(2) ; (see alsoDoc. 336 at 115). An alternative to the confirmation mailing process for returning a voter to active status is for the voter to appear to vote in a subsequent election and give oral or written affirmation of continuous residency within the county. N.C. Gen. Stat. § 163–82.14(d)(3) ; (see alsoDoc. 336 at 115). This confirmation mailing process is not at issue in this case.

Mail verification is admittedly "not a precise verification system" for determining an applicant's residency. (Doc. 336 at 112 (Director Strach acknowledging that mail verification has imperfections).) For example, just because the postal service delivers a verification does not mean that the applicant lives there. In addition, eligible voters, such as those living in school dormitories and apartments, can fail mail verification because sometimes those institutions fail to properly deliver nonforwardable mail to such residents. Though imperfect, however, North Carolina's mail verification process provides some assurance that the registrant resides where he or she claims. More importantly, it is the method the State adopted years before any of the voting changes at issue were introduced.

The North Carolina General Assembly expressed several reasons for eliminating SDR. Generally, SDR caused problems with the processing and verification of voter registrations during early voting. These problems were noted four years earlier in a March 31, 2009 SBOE memorandum¹⁶¹ and in debate on HB 589, and they went unrebutted by the bill's opponents. (Pl. Ex. 202 at 39–41; Pl. Ex. 549 at 5.)

161 Opponents of the bill were apparently unaware of, or unwilling to acknowledge, this report. (See, e.g., Doc. 134–4 at 220 ("Same day registration, I don't know of a single problem we've had with that ....").)

For one, SDR imposed administrative burdens on CBOEs. During the Senate Rules Committee debate, Senator Rucho contended:

There's no way and there's no simple way to validate. What we're trying to do is give the Board of Elections an opportunity to do their job correctly, validate those individuals and be sure that the election is above board.

(Doc. 134–4 at 45.) Later, during the second reading, he added:

It also allows time for—to verify voters' information by repealing same day registration and which will ensure accuracy. It's been a challenge for the Board of Elections to be able to identify and validate everyone that has come there on the basis of one-day registration. ...

(Id. at 87; see also Pl. Ex. 56 at 5–6; Pl. Ex. 202 at 41 (noting that "a lot of [CBOEs] have a very hard time working their way through a system when someone comes up and registers to vote and votes at the same time"); Pl. Ex. 549 at 5.) The SBOE's 2009 memo reported that CBOEs "had to hire additional staff to process [SDR] registrations." (Pl. Ex. 56 at 5.) Moreover, CBOEs also had difficulty making a timely preliminary determination about the qualifications of a registration applicant using SDR within the two business days required. See 2007 N.C. Sess. Law 253, § 1. According to the SBOE, CBOE staff "worked long hours and workweeks to meet this two-business day requirement. Although staff worked as efficiently as they could, generally, it was not possible to process the number of voter registration applications received during one-stop [early voting] within this two-day period." (Pl. Ex. 56 at 5.) Similarly, "[d]ue to volume issues, [CBOEs] experienced minor [delays] in DMV validations, especially during the last few days of one-stop voting." (Id.) Thus, these actual administrative burdens on CBOEs, identified by the SBOE Executive Director in 2009, persisted even though CBOEs responded with increased staffing and hours. For a State with ten million residents, these are real-life concerns. Marston v. Lewis, 410 U.S. 679, 681, 93 S.Ct. 1211, 35 L.Ed.2d 627 (1973) (reversing district court injunction against implementation of fifty-day registration cut-off where State showed, among other things, that in the weeks preceding the general election "county recorders and their staffs [were] unable to process the incoming affidavits because of their work in the fall primaries").

Further, SDR created additional administrative burdens by requiring CBOEs to process "intersecting registrations." For example, in the words of former Director Bartlett,

There were issues with some voters who submitted a voter registration application to one county during the last few days before the registration deadline and then appeared to vote in another county or actually registered at one-stop in another county and voted. Similarly, there were voters who registered at a one-stop site and voted although they had been issued a mail-in absentee ballot in a previous county of registration.

(Pl. Ex. 56 at 7.) To address this issue, CBOEs were required to process the newest registration and then cancel and revoke the previous registration. (Id.) Because registrations are timely so long as they are postmarked by the registration deadline, "[m]any mailed-in registrations were processed after one-stop voting had begun." (Id.) Thus, there were situations where SDR-registrations were cancelled because "the voter submitted a new registration in another county that was belatedly processed." (Id.) Each of these situations required CBOEs to conduct research "on a case by case basis to ensure that there was no possible fraud being committed." (Id.)

The North Carolina General Assembly also articulated that it eliminated SDR to better ensure that the State's mail verification process could be completed before a registration applicant's vote was counted. (Pl. Ex. 202 at 41; Pl. Ex. 549 at 5.)

North Carolina's mail verification requirement is most effective when a traditional registrant votes on Election Day, thus offering the maximum number of days to run its intended course. Early voting complicates that by advancing the day a ballot is cast and reducing the ability for the State to remove registrants who fail mail verification before voting. SDR further magnifies this problem. First, by making registration and voting contemporaneous, it further reduces the State's ability to remove registrants before they vote. Second, by placing registration closer to the point where votes must be challenged or counted, it makes it much more likely that those who fail mail verification will do so after it is too late to challenge their ballot. In this regard, SDR (and the removed seven days of early voting) conflicted with pre-existing law; in a practical sense, they are incompatible.

Ballots cast by same-day registrants, like all other absentee ballots, are counted on Election Day; under statute, they must be challenged, if ever, on Election Day.¹⁶²(Doc. 336 at 157, 225–26.) At the time of HB 589 and under the best case scenario, therefore, CBOEs had, at most, twenty days to verify same-day registrants before Election Day if they presented on the first day of the previous seventeen-day period and CBOEs were able to begin the verification process immediately. But, in truth, the majority of early voting occurred later in the early-voting period, closer to Election Day, and CBOEs struggled to begin the mail verification process that quickly. (Pl. Ex. 42 (Ex. 40); Def. Ex. 268 at 40.) Former Director Bartlett recognized this issue in the SBOE's 2009 memo, noting that SDR simply failed to provide a sufficient number of days to permit the mailings to run their course to verify same-day registrants. (Pl. Ex. 56 at 5–6.) Bartlett went so far as to urge that "SDR laws need to be revised" because of the problem. (Id. at 6–9.) In addition, George Gilbert, former director of the Guilford CBOE, acknowledged that a voter who registers even right before the "close of books," twenty-five days before Election Day, will have more time to pass the verification procedure than a voter who registered and voted during early voting. (Doc. 165 at 16.)

162 Former Director Bartlett claimed in the SBOE's 2009 memo that "[a]s long as the second notice is returned prior to canvass, then the one-stop registrant's registration can be denied and their in-person absentee ballot appropriately disapproved." (Pl. Ex. 56 at 6.) Bartlett's claim that the registrant's registration can be denied is incorrect. By the canvass day, the registrant has either voted or not. If the registrant has voted, then failing mail verification will merely make his registration inactive and subject him to the confirmation mailing process. SeeN.C. Gen. Stat. 163–82.7(g)(3) ; (Doc. 340 at 208–09). Bartlett's claim that CBOEs can simply reject the SDR absentee ballot on the canvass date is also contrary to Strach's testimony, (Doc. 336 at 157 (saying that absentee ballots not challenged by Election Day are not retrievable)), and N.C. Gen. Stat. § 163–89.
As a general rule, challenges to absentee ballots (including early voting and SDR ballots) must be made on Election Day. SeeN.C. Gen. Stat. § 163–89(a) ("The absentee ballot of any voter may be challenged on the day of any statewide primary or general election or county bond election beginning no earlier than noon and ending no later than 5:00 P.M., or by the chief judge at the time of closing of the polls as provided in G.S. 163–232 and G.S. 163–258.26(b).") Section 163–89(a) provides a later challenge deadline for the absentee ballots of voters received by CBOEs pursuant to N.C. Gen. Stat. § 163–231(b)(ii) or (iii). This appears to be mislabeled, as § 163–231(b) has no (ii) or (iii) subsections, and thus the reference to § 163–231(b) (ii) and (iii) appears to be to § 163–231(b)(2) b & c. Section 163–231(b) (1) provides, in sum, that the CBOE must receive absentee ballots by 5:00 p.m. on Election Day. Section 163–231(b)(2), however, permits certain absentee ballots that are received at a later time. Under § 163–231(b)(2) b, absentee ballots are to be accepted so long as they are postmarked by Election Day and received by the CBOE no "later than three days after the election by 5:00 p.m." Under § 163–231(b)(2) c, certain military absentee ballots are timely if received by the CBOE no "later than the end of business on the business day before the canvass."
Accordingly, absentee ballots postmarked by Election Day and certain military absentee ballots appear to be the only ballots that can be challenged after Election Day under § 163–89(a). Those ballots "may be challenged no earlier than noon on the day following the election and no later than 5:00 p.m. on the next business day following the deadline for receipt of such absentee ballots." Id.§ 163–89(a). Therefore, given the deadlines for receipt outlined above, absentee ballots postmarked by Election Day appear to be subject to challenge up to four days after Election day (three days after Election Day + the next business day) and certain absentee military ballots appear to be subject to challenge on Election Day (business day before the canvass + the next business day). Seeid.§§ 163–89(a), 163–231(b). There is no reason to believe that absentee ballots cast during early voting, including SDR ballots, fall into either one of these categories. In fact, the trial testimony indicated otherwise. (Doc. 341 at 163 (Director Strach: stating that an absentee early-voting ballot will count "unless that voter [is] challenged on Election Day").)
In sum, Bartlett's claim is contrary to the evidence presented in this case and to N.C. Gen. Stat. § 163–89. Although hearings for all challenges are set on the canvass date, N.C. Gen. Stat. § 163–89(e), challenges to absentee early voting ballots, including SDR ballots, must be filed before that date—on Election Day. But even if Bartlett were correct, there would still be insufficient time. The 2009 SBOE report authored by Bartlett assumed CBOEs could make sua sponte challenges at the canvass, yet still found that "there was not enough time between the end of one-stop voting (and SDRs) and the canvass day to ensure that verification mailings completed the mail verification process." (Pl. Ex. 56 at 5.)

Plaintiffs respond that "SDR registrants [mail-] verified at rates comparable to, and sometimes higher than, non-SDR registrants." (Doc. 346 at 32.) Defendants counter with evidence that, regardless of whether non-SDR registrants fail mail verification at higher rates, SDR registrants nevertheless vote at higher rates than non-SDR registrants despite failing mail verification.¹⁶³ (Doc. 347 at 42.) Defendants cite the testimony of Brian Neesby, a data analyst for the SBOE, and a memorandum he prepared for Director Strach, as evidence of registrants' mail verification rates. (Id.) Mr. Neesby examined the voted-but-failed mail verification rate of SDR versus non-SDR registrants and found that, for the 2012 general election, 2.44% of SDR registrants (2,361/97,373) failed mail verification after voting, compared to 0.34% of non-SDR registrants (2,306/680,904). (Def. Ex. 16 at 5.) In addition, he found that 95.6% of traditional (non-SDR) registrants completed mail verification before voting, whereas 96.2% of SDR registrants voted before mail verification could be completed. (Id. at 6.)

163 Plaintiffs rely on two SBOE reports on mail verification rates for the 2010 and 2012 general elections and a Guilford CBOE report on mail verification in the 2012 general election. (Id.)
Plaintiffs also rely on testimony Dr. Lichtman attempted to offer in Plaintiffs' rebuttal case on the last day of trial, in which he concluded that 3.63% of all 2014 non-SDR registrants voted but did not complete mail verification by April 2015. (Doc. 342 at 148.) Defendants objected to Dr. Lichtman's rebuttal testimony when it was offered. The court expressed reservations about allowing it but reserved ruling. (Id.at 146–47.) The parties have since briefed the motion. (Docs. 324, 325, 351.) The court now concludes that Dr. Lichtman's testimony should not be admitted.
There are several problems with Dr. Lichtman's late-disclosed testimony. First, Dr. Lichtman claims to respond to Mr. Neesby's analysis in the SBOE's May 2015 memorandum (Def. Ex. 16) which was available to Plaintiffs ahead of trial, and Plaintiffs decided to have Dr. Lichtman complete his analysis at least ten days before Dr. Lichtman's rebuttal testimony, yet Plaintiffs made no effort to disclose it to Defendants. (Doc. 342 at 185.) By attempting to offer it on the last day of trial, Defendants were severely prejudiced and had no opportunity to examine the accuracy of Dr. Lichtman's conclusions or the methodology he used. Second, the need for advance disclosure was made apparent during Dr. Lichtman's cross-examination, which revealed that he did not extract the data himself but retained a third party, David Ely, to do so. (Id.at 166–67.) Thus, Dr. Lichtman had no personal knowledge of how the data was extracted from SEIMS, (seeid.at 163–170), and Defendants had no opportunity to examine Mr. Ely, who was in California, on his methodology or to attempt to replicate it, (id.at 166–67). Defendants assert that Dr. Lichtman (or more accurately, Mr. Ely) omitted several SEIMS database proxies that other SBOE analyses have included. (Doc. 325 at 10 n.7.) Yet, given Dr. Lichtman's limited knowledge of SEIMS or the data extracted, and the limited nature of the cross-examination that was possible given the timing of the disclosure, this court is in a poor position to ascertain whether these claims are true. (Id.at 163–70.) Third, even though Dr. Lichtman purported to rebut the testimony of Mr. Neesby, the inferences that can be drawn from his analysis are more limited. In contrast to Mr. Neesby—who studied the vote-but-failed rate of 2012 voters—Dr. Lichtman purported to study the vote-but-did-not-passrate of 2014 voters. (Id.at 148–49.) As noted above and by the Fourth Circuit in League, whether a voter has failed to pass mail verification is not as probative as whether the voter has failed mail verification. See769 F.3d at 246. The latter is a much stronger proxy for the voter being ineligible than the former. Fourth, by comparing the vote-but-failed mail verification rates of SDR and non-SDR registrants within the same year, Mr. Neesby's analysis permits the court to infer which voting practice is more problematic. (Def. Ex. 16 at 5.) Because Dr. Lichtman only studied regular registrants in 2014 (when SDR was not available), his analysis does not indicate whether SDR registrants during 2014 would have voted despite failing mail verification at an even higher rate than the non-SDR registrants he studied. (Doc. 342 at 148.) The reliability of Dr. Lichtman's proposed testimony is also uncertain insofar as his 3.63% figure is an order of magnitude higher than the comparable figure for 2012, without explanation. (SeeDef. Ex. 16 at 5 (finding 0.34%.).) Finally, in their proposed findings of fact and conclusions of law, Plaintiffs merely cite Dr. Lichtman for the proposition that "3.63% of 2014 registrantshad not passed mail verification by April 2015." (Doc. 346 at 32 (emphasis added).) In other words, Plaintiffs do not claim that Dr. Lichtman's study shows that 3.63% of 2014 registrants voteddespite failing mail verification. (Id.) This may be an oversight, but the fact that even Plaintiffs do not clearly assert this critical fact only further cautions against reliance on Dr. Lichtman's analysis. Given all of these factors, the court, in its discretion, will exclude Dr. Lichtman's 2014 analysis. Plaintiffs failed to lay a sufficient foundation for Dr. Lichtman's testimony. Fed. R. Civ. P. 602. Moreover, even if the court were to consider it, the court would not find it credible for the reasons discussed above and those stated in Defendants' objection to his testimony. (SeeDef. Ex. 325.) The court finds the testimony of Mr. Neesby, who was highly credible and demonstrated intimate knowledge about the SEIMS database, to be the more probative and reliable measure of the comparative rates at which SDR and non-SDR registrants vote despite failing mail verification.

Plaintiffs have filed several motions to exclude Mr. Neesby as a witness and to strike his testimony, (Docs. 326, 327), to which Defendants have responded, (Doc. 350).¹⁶⁴ They contend his analysis constitutes expert testimony and that Mr. Neesby neither was disclosed as, nor is, an expert. (Doc. 326 at 9–11.) While Plaintiffs "do[ ] not object to Mr. Neesby's testimony to the extent that it constitutes the bare presentation of election data from the SBOE's database," they do object to certain statements they claim "provid[e] explanatory inferences and conclusions." (Id. at 14–15.) Consistent with Plaintiffs' objection, this court need not consider Mr. Neesby's opinions it regards to be expert testimony. However, it will consider Mr. Neesby's voted-but-failed-data, which is merely the presentation of election data from the SBOE's database. Further, the exhibit containing Mr. Neesby's data was admitted into evidence without objection, (Doc. 341 at 172), and was made available to Plaintiffs on June 6, 2015, over a month before trial and prior to Mr. Neesby's July 18, 2015 deposition, (Doc. 326 at 17–18). Accordingly, Plaintiffs' motion will be denied in that regard.

164 Plaintiff Intervenors' challenge to Mr. Neesby is limited to Exhibit BN-3 and his analysis on the mail verification failure rates of pre-registrants. (Doc. 328.) Plaintiff Intervenors do not seek to exclude Mr. Neesby's voted-but-failed analysis contained in Defendants' Exhibit 16. (Id.)

The important fact is that SDR's proximity to Election Day makes it much more likely that SDR registrants will be able to vote despite failing statutory mail verification.¹⁶⁵ This is evident apart from Mr. Neesby's data.¹⁶⁶ In fact, Plaintiffs' own exhibits are consistent with what the data demonstrate. (Pl. Ex. 56 ("Some same day registrants did not complete the mail verification cycle prior to the certification date. ... 2.4 percent of registrations were subsequently denied due to the inability of the county boards to verify the applicant's address through the mail."); Pl. Ex. 68A at 6.) These data merely confirm what logic reveals must be the case—SDR's proximity to Election Day, well inside the twenty-five day registration cut-off, simply does not provide a sufficient number of days for the mail verification process to work, and thus effectively frustrates—or negates — North Carolina's process for verifying a voter's residence. Moreover, by making registration and voting contemporaneous, SDR negated the State's ability to cancel registrations and made N.C. Gen. Stat. § 163–82.7(f) a ity. In this regard, the General Assembly was responding to a legitimate and obvious problem.

165 Plaintiffs have pointed out that this is possible even for those registrations submitted before the twenty-five day cut off (i.e., registration immediately before the twenty-five day cutoff under N.C. Gen. Stat. § 163–82.6(c) ). This is true—there are instances where mail verification does not run its course even under the twenty-five day schedule. But the year-round registration system is designed to avoid this result, and SDR's proximity to Election Day clearly made this problem much worse. This was not merely theoretical.

166 It does not appear that the data in Defendants' Exhibit 16 had been compiled at the time SL 2013–381 was adopted.

In League, the Fourth Circuit found the State's best fact to be "that a thousand votes that had not yet been properly verified had been counted in an election." 769 F.3d at 246 (emphasis added). The court found this to be tenuous because "nothing in the district court's opinion suggests that any of [the over one-thousand votes] were fraudulently or otherwise improperly cast." Id. This court's preliminary injunction opinion could have been clearer on this point. For example, this court stated that "over a thousand ballots were counted in recent elections by voters who were not (or could not be) properly verified." McCrory, 997 F.Supp.2d at 353 & n. 37 ("[I]n the 2012 general election, SBOE records show that approximately 1,288 ballots were counted despite being cast by voters who did not complete the verification process."). However, the trial evidence demonstrates that it was not just that votes were being counted without mail verification being completed, which would not reveal whether or not a vote was properly counted. Instead, votes were being counted despite voters failing mail verification (Def. Ex. 16)).

In fact, the evidence before the legislators in 2013 established that in the 2012 general election alone, SDR required CBOEs to count the votes of at least 1,264 (we now know to be 2,361) SDR registrants who failed the State's mail verification after Election Day and the canvass.¹⁶⁷ (Pl. Ex. 68A at 6; Def. Ex. 16 at 5.) Insofar as mail verification is aimed at determining whether or not the address given by the registrant is correct (an eligibility to vote requirement), failing mail verification is a proxy (albeit an imperfect one) for the registrant's given address being incorrect. Because residence determines what races a registrant is eligible to vote for,¹⁶⁸ a registrant who votes despite providing the wrong address will cast an otherwise improper and ineligible ballot. When voters cast a ballot in a local contest for which they are not eligible, eligible voters who also cast a ballot for that race are disenfranchised—regardless of whether the ineligible voters intended to commit fraud. Insofar as every affected voter's right is important, "[h]aving every ballot cast by every eligible voter is also of fundamental importance." League of Women Voters v. Blackwell, 340 F.Supp.2d 823, 829–30 (N.D.Ohio 2004). For, "the right of suffrage can be denied by a debasement or dilution of the weight of a citizen's vote just as effectively as by wholly prohibiting the free exercise of the franchise." Purcell, 549 U.S. at 4, 127 S.Ct. 5 (quoting Reynolds v. Sims, 377 U.S. 533, 555, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964) ).

167 Under State law, these persons are deemed "ineligible" to vote unless and until they correct their registration. Seesupranote 160 (discussing the process for voters on the confirmation mailing list to vote).

168 Under North Carolina law, voters must be registered for a precinctand may have their vote count only for candidates for that precinct. (Doc. 336 at 227–28.)

Accordingly, at trial Defendants provided what the Fourth Circuit appears to have sought at the preliminary injunction stage—evidence that suggests an important number of registrants cast improper ballots by voting with an incorrect address.

Plaintiffs nevertheless have made several additional challenges to the State's provided justification.

Plaintiffs first highlight the problems with North Carolina's mail verification process. (Doc. 346 at 73, 143.) Their criticisms are not unjustified. But, as with many of Plaintiffs' arguments in this case, such criticisms involve State policy considerations that have not been challenged in this case. Indeed, while Plaintiffs pointed out the weaknesses of mail verification, they neither challenged it as inappropriate nor demonstrated that other States have abandoned it.¹⁶⁹The question is whether proponents of SL 2013–381 acted with only a tenuous reason when, in light of all current registration options available, they chose to repeal SDR (which they had opposed in the first instance) after it had been superimposed over the existing mail verification system in a fashion that made compliance impossible in some cases. The current mail verification process may be imperfect, but the General Assembly's decision to reverse its decision in 2007 by removing SDR instead of debating ways to overhaul the pre-existing and more wide-ranging mail verification system in an effort to avoid having eligible voter's ballots disenfranchised was not unreasonable.

169 In fact, mail verification appears to be a staple of voter identification and election integrity employed by other States. See, e.g., Colo. Rev. Stat. Ann. § 1–2–509 ; Del. Code Ann. tit. 15, § 2014 ; Me. Rev. Stat. Ann. tit. 21–A, § 122 ; Mo. Ann. Stat. § 115.155 ; 17 R.I. Gen. Laws. Ann. § 17–9.1–25 ; W. Va. Code Ann § 3–2–16. How States accommodate mail verification in the context of SDR or EDR reflects States' policy choices. For example, Wisconsin, with expansive voter registration, appears to allow EDR in a fashion that counts the votes of those who later fail mail verification but simply precludes those voters from voting again until they address the problem; such persons may be referred for prosecution. SeeWis. Stat. Ann. §§ 6.29, 6.55, 6.56. Colorado, by contrast, does not consider a voter to be registered until mail verification occurs, apparently in order to comply with the NVRA's prohibition from removing a registered voter from the rolls absent certain notice. Colo. Rev. Stat. Ann § 1–2–509(3) ; see, e.g., Common Cause of Colorado v. Buescher, 750 F.Supp.2d 1259, 1279 (D.Colo.2010).

Plaintiffs next contend that the General Assembly's justifications are tenuous because they refused to take alternative routes to cure the problems, routes proposed by a local voter advocacy and government lobbying organization. According to Plaintiffs, Bob Phillips, the executive director of Plaintiff Common Cause North Carolina, "had shared with legislators three possible ways to change SDR to allow for more time for completion of the mail verification process." (Doc. 346 at 65; Pl. Ex. 12.)

This argument fails for a number of reasons. Most importantly, while relevant, just because the General Assembly chose one policy approach over another to address a real problem does not render the chosen policy approach tenuous. See, e.g., Askew v. City of Rome, 127 F.3d 1355, 1387–88 (11th Cir.1997).

Moreover, Phillips' proposals face their own obstacles. Phillips proposed eliminating SDR during the last five days of the early-voting period. (Pl. Ex. 12 at 6.) Of course, this conflicts with the relief demanded in this case by Plaintiffs. It also fails to acknowledge that it would still provide an inadequate number of days for the mail verification process to run its course, and Plaintiffs offered no evidence otherwise. Phillips' second proposal was to move the canvass date back six days. This fails to recognize that the canvass date has no bearing on the fact that challenges must be made for SDR ballots on Election Day. SeeN.C. Gen. Stat. § 163–89(a). It also ignores the fact that, even if the canvass was later,¹⁷⁰ and even if challenges could be made on that day, SDR would still deprive the State of its ability to deny registration applications under N.C. Gen. Stat. § 163–82.7(f) —in effect requiring the State to carry the burden of mounting a successful challenge under N.C. Gen. Stat. § 163–89. Phillips' third proposal was not to count a ballot if the first mail verification notice returns as undeliverable. (Pl. Ex. 12 at 6.) This only weakly addresses SDR's problems articulated by the General Assembly because, as noted above, much of SDR's use takes place toward the end of early voting (closest to Election Day), which provides little-to-no time for even the first mail verification notice to return undeliverable by Election Day, when election officials are already busy managing other aspects of the election. (Pl. Ex. 42 (Ex. 40); Def. Ex. 268 at 40.)¹⁷¹Plaintiffs suggest that the removal of SDR will create administrative burdens, specifically requiring the increased processing of provisional ballots. (Doc. 346 at 29; Pl. Ex. 811 at 31–32.) According to Plaintiffs, some number of unregistered individuals who would have cast a regular ballot through SDR will now cast a provisional ballot. This appears to be true in part, with the caveat that under federal and State law, voters who do not claim to be registered are not entitled to cast a provisional ballot. See52 U.S.C. § 21082(a) ; N.C. Gen. Stat. § 163–166.11. Thus, the only voters who will now cast a provisional ballot but could have cast a regular ballot with SDR in place are those who claim they are registered to vote. Nevertheless, the number of provisional ballots cast during early voting because of "no record of registration" increased by 871 from 2010 to 2014 (97 to 968).¹⁷² (Pl. Ex. 689.) This supports Plaintiffs' argument, albeit weakly. In addition, because the registrant is claiming to be registered, CBOEs evaluating "no record of registration" provisional ballots must research whether or not there was an attempt to register. (Pl. Ex. 800 at 34–39.) It is not clear, however, whether this process is more burdensome than that required to process an SDR application within the forty-eight hour period previously required by statute. 2007 N.C. Sess. Law 253, § 1. But even assuming the two processes involve similar burdens, there were vastly more SDR-registrants than there will be new provisional ballots for "no record of registration." (See Pl. 40 at 35 (Ex. 14); Pl. Ex. 689.) For example, even if all 871 of the additional "no record of registration" provisional ballots cast between 2010 and 2014 are attributable to the removal of SDR, there were at least 60,918 SDR registrants in 2010, each of which had to be processed within forty-eight hours. As such, any administrative burden caused by the lack of SDR is greatly outweighed by the administrative burdens caused by SDR. (See Pl. Ex. 56 at 5.)

170 Moving the canvass date back a sufficient distance would not be without its own challenges. (Doc. 336 at 156–57) (Strach: testifying that "in order for the mail verification system ... to work, I am not sure how far it [would have to be moved back]. You could possibly be into December or later, and then we would have other issues with seating elected officials ....").)

171 Further, to the extent a registrant can fail mail verification due to postal error, (Pl. Ex. 56 at 5), having only one mailing would increase the risk that such an error would erroneously disenfranchise the voter. With two mailings, this error must be repeated in order to affect the voter.

172 The total number of provisional ballots cast (including Election Day voting) for "no record of registration" actually decreased from 2010 to 2014 (9,927 to 7,765), but this disparity is attributable to a decrease in "no record of registration" provisional ballots being cast on Election Day. (Pl. Ex. 689.) These Election Day provisional ballots are not probative of the effect of SDR's removal given that SDR was never available on Election Day.

Finally, Plaintiffs contend that the removal of SDR is tenuous because North Carolina law permits the unverified-address-ballots of "unreported movers" (registrants who move to a different precinct within the same county more than thirty days before Election Day but fail to notify the CBOE) to be counted. (Doc. 346 at 143.) What Plaintiffs do not acknowledge is that North Carolina's accommodation of unreported movers is in part required by § 8(e) of the NVRA. 52 U.S.C. § 20507(e). North Carolina provides unreported movers with three voting options on Election Day. First, they may vote a regular ballot in their new precinct, so long as they give an oral or written affirmation of their new address upon presentation to vote. N.C. Gen. Stat. § 163–82.15(e). Second, they may vote a regular ballot at a "central location in the county to be chosen by the [CBOE]." Id. Third, they can cast a provisional ballot at their former precinct, and the CBOE is required to "count the individual's provisional ballot for all ballot items on which it determines that the individual was eligible under State or federal law to vote." Id. In each situation, even if the unreported mover is subject to the mail verification process after voting, seeid.§ 163–82.15(b) (requiring CBOE that receives notice of change of residence to initiate mail verification),¹⁷³ the same proximity-to-Election-Day problem as SDR exists. However, it results not from State law but from § 8(e) of the NVRA, which requires North Carolina to accommodate unreported movers. See52 U.S.C. § 20507(e)(2)(A),(B).

173 It is not entirely clear, and the parties have not established, whether an unreported mover's completion of an oath or affirmation of new address while voting triggers the form of mail verification set out in N.C. Gen. Stat § 163–82.15(b).

Under the NVRA, States have in effect two options. First, they can permit an unreported mover to choose between voting at his former precinct, a central location, or his new location.¹⁷⁴See id.§ 20507(e)(2)(A). Second, they can permit the unreported mover to vote in the current election at either the unreported mover's former or new precinct.¹⁷⁵See id.§ 20507(e)(2)(B). North Carolina accommodates unreported movers more than the NVRA requires—by giving unreported movers the option to vote in the current election at three locations (former, new, central), seeN.C. Gen. Stat. § 163–82.15(e), but the State could not prevent the unverified-address-ballots of unreported movers from being counted without violating the NVRA, see52 U.S.C. § 20507(e)(2). At a minimum, the State must provide at least one location for unreported movers to vote while updating their address, and at this location the State can do no more than ask the unreported movers to provide "oral or written affirmation" of their address. See id. Thus, the State cannot be faulted for permitting a voting practice that federal law requires.¹⁷⁶

174 If a State chooses this option, it is not actually required to permit the unreported mover to vote in the current election at his new precinct. See52 U.S.C. § 20507(e)(2)(A)(ii)(II). Instead, the State need only permit the unreported mover to "correct [his or her] voting records for purposes of voting in future elections." Id.Whether or not the unreported mover can actually cast a ballot in the current election at his new precinct is to be determined by State law. Id.

175 The NVRA provides that so long as "State law permits the registrant to vote in the current election upon oral or written affirmation by the registrant of the new address at" either their former or new precinct, then "voting at the other locations ... need not be provided as options." 52 U.S.C. § 20507(e)(2)(B).

176 Further, even if the State had the authority to address the issue, there are at least two logical reasons to remove SDR but permit unreported movers to update their address so close to Election Day. First, there were far more SDR-registrants than unreported movers. For example, in 2012, Plaintiffs' expert Dr. Gronke reports there were 246,895 SDR registrants, (Pl. Ex. 40 at 35 (Ex. 14)), compared to 9,720 unreported movers, (Pl. Ex. 42 at 95). Second, the unreported mover provision assists those who move, and African Americans move more frequently than whites. (See, e.g., Pl. Ex. 45 at 17.) African Americans cast a disproportionate share of unreported mover provisional ballots in 2010 and 2014. (Pl. Ex. 689.) In other words, there is absolutely no discriminatory purpose that can be inferred from removing SDR but leaving the unreported mover provision.

For all these reasons, the court finds that the trial evidence demonstrates that the State's basis for removing SDR was not pretextual or tenuous, but instead substantial.¹⁷⁷ It sought to repeal a voting mechanism that resulted in the counting of votes for thousands of voters who failed mail verification. The Supreme Court has recognized that a State "indisputably has a compelling interest in preserving the integrity of its election process." Purcell, 549 U.S. at 4, 127 S.Ct. 5 (quoting Eu v. San Francisco Cty. Democratic Cent. Comm., 489 U.S. 214, 231, 109 S.Ct. 1013, 103 L.Ed.2d 271 (1989) ). "Confidence in the integrity of our electoral processes is essential to the functioning of our participatory democracy." Id.

177 There was also evidence that certain ineligible SDR-voters tainted the outcome of a close municipal election in the town of Pembroke in November 2013 that caused the SBOE to order a new election. (Doc. 341 at 50–51.) According to Strach: "one of the reasons ... that the new election was called was because there was a concern that ineligible people were being brought to the one-stop sites to register and vote." (Id.at 51.) For example, a group of individuals presented a common lease and voted via SDR even though they were not actually residents at the address on the lease. (Id.at 50–51.) The poll worker should not have allowed the individuals to vote via SDR, given that their lease was not an acceptable HAVA document. (Id.) The incident, nevertheless, illustrates the value of mail verification, especially given that poll worker (often volunteers) error does occur. (Def. Ex. 816 at 43–44.) Since the ineligible voters did not actually live where they purported, (Doc. 341 at 50–51), they likely would have failed mail verification if only there had been time for it to be completed, (seeid.).

Interestingly enough, at this court's January trial, Dr. Burden recognized that States have a strong interest in a registration requirement. (Doc. 407 at 53.) In his words, "[m]andating voter registration at one point became a sort of obvious reform for many states. That was a new restriction, but it had a strong motivation behind it." (Id.) Of course, a registration requirement is diluted where the State is not able to verify the information provided. SDR presents just such an issue; it requires the State to accept the votes of several thousand individuals who failed mail verification.¹⁷⁸ In light of this fact, it is difficult to understand how Dr. Stewart can consider registration an "obvious reform," but consider the removal of SDR to be merely tenuous. (See Pl. Ex. 44 at 14–18.)

178 These SDR voters failed even though they were required to provide at least a HAVA document bearing their name and address, see2007 N.C. Sess. Law 253, § 1.

iv. OOP Voting

When introducing the revised HB 589 to the Senate Rules Committee, the repeal of OOP voting was described as "basically mov[ing] the law back to the way it was prior to 2005." (See Pl. Ex. 202 at 12.) Before 2005, North Carolina law required voters to vote in their assigned precinct.

Not until the 2004 general election did State election officials for the first time begin to count provisional ballots cast outside the voter's precinct. James v. Bartlett, 359 N.C. 260, 266, 607 S.E.2d 638, 641 (2005). The counting of these ballots changed the outcome in some elections, and the losing candidates filed suit. SeeIn re Election Protest of Fletcher, 175 N.C.App. 755, 756, 625 S.E.2d 564, 565 (2006) (noting the challenger's party affiliation). In James, a unanimous North Carolina Supreme Court held that the State's election statutes "clearly and unambiguously" prohibited the counting of such ballots. 359 N.C. at 267, 607 S.E.2d at 642. In so construing the statutes at issue, the Court avoided deciding whether the North Carolina State Constitution also prohibits the counting of OOP ballots. Id. at 266, 607 S.E.2d at 642.

Less than a month after the Republican challengers won in the North Carolina Supreme Court, however, the Democratically-controlled General Assembly amended the election laws and stated that it had intended to permit OOP ballots all along, 2005 N.C. Sess. Law 2, § 1(3), even though HAVA does not require that OOP provisional ballots be counted, see52 U.S.C. § 21082(a)(4) ; James, 359 N.C. at 267–68, 607 S.E.2d at 643.¹⁷⁹ The General Assembly, strictly along party lines, retroactively applied this rule in order to ify the Republican challengers' win at the Supreme Court. In 2013, after the Republicans took control of the General Assembly, they repealed OOP voting. See 2013 N.C. Sess. Law 381, § 49.1.

179 In 2003, the General Assembly enacted a law "to ensure" that North Carolina law complied with the requirements of HAVA. 2003 N.C. Sess. Law 226, § 1. This enactment had bipartisan support, but apparently not as to the interpretation the Democratically-controlled legislature later sought to attribute to it.

Plaintiffs claim that the repeal of OOP voting was tenuous. The trial evidence demonstrated otherwise.

Plaintiffs argue that there was no reason given for repealing OOP voting during the legislative process. The Supreme Court has explained that, in the Fourteenth Amendment context, "Although race-based decisionmaking is inherently suspect, until a claimant makes a showing sufficient to support that allegation the good faith of a state legislature must be presumed." Miller v. Johnson, 515 U.S. 900, 916, 115 S.Ct. 2475, 132 L.Ed.2d 762 (1995) (citations omitted). No legislator in the House or Senate openly opposed the removal of OOP voting during the legislative debate. (See, e.g., Doc. 335 at 208.) There appears to have been nothing said about OOP voting, whether for or against it.

Given the controversy in 2005, there was probably little need for a debate. When the Supreme Court unanimously interpreted the election laws to prohibit OOP voting, it provided extensive justifications for requiring voters to cast ballots in the correct precinct. The Court initially noted that the precinct-based "voting system is woven throughout the fabric of our election laws." James, 359 N.C. at 267, 607 S.E.2d at 642. It went on to explain how North Carolina's

statutory residency requirement provides protection against election fraud and permits election officials to conduct elections in a timely and efficient manner. The General Assembly recognized in ratifying N.C.G.S. § 163–55 that without a precinct residency requirement, there would be a generous magnification of the potential for mischief in the form of one person voting in numerous precincts.

Id. at 270, 607 S.E.2d at 644 (citations omitted). The Court observed that, "If voters could simply appear at any precinct to cast their ballot, there would be no way under the present system to conduct elections without overwhelming delays, mass confusion, and the potential for fraud that robs the validity and integrity of our elections process." Id.

The Court agreed with the findings of the Sixth Circuit as well, which provided a number of reasons for enforcing a precinct system:

The advantages of the precinct system are significant and numerous: it caps the number of voters attempting to vote in the same place on election day; it allows each precinct ballot to list all of the votes a citizen may cast for all pertinent federal, state, and local elections, referenda, initiatives, and levies; it allows each precinct ballot to list only those votes a citizen may cast, making ballots less confusing; it makes it easier for election officials to monitor votes and prevent election fraud; and it generally puts polling places in closer proximity to voter residences.

Id. at 270–71, 607 S.E.2d at 644–45 (quoting Sandusky Cty. Democratic Party v. Blackwell, 387 F.3d 565, 568–69 (6th Cir.2004) (per curiam)). In the Sixth Circuit case, that court had further explained,

The States long have been primarily responsible for regulating federal, state, and local elections. These regulations have covered a range of issues, from registration requirements to eligibility requirements to ballot requirements to vote-counting requirements. SeeStorer v. Brown, 415 U.S. 724, 730, 94 S.Ct. 1274, 39 L.Ed.2d 714 (1974) ("[T]he States have evolved comprehensive, and in many respects complex, election codes regulating in most substantial ways, with respect to both federal and state elections, the time, place, and manner of holding primary and general elections,

the registration and qualifications of voters, and the selection and qualification of candidates."). One aspect common to elections in almost every state is that voters are required to vote in a particular precinct. Indeed, in at least 27 of the states using a precinct voting system, including Ohio, a voter's ballot will only be counted as a valid ballot if it is cast in the correct precinct.

Sandusky Cty. Democratic Party, 387 F.3d at 568.

Several of the legislators who voted for SL 2013–381 had also voted against allowing OOP voting in 2005, opting instead for the system recognized in Sandusky County Democratic Party and later in 2005 by the North Carolina Supreme Court in James. (Compare Def. Ex. 168 (House and Senate votes on OOP voting in 2005 by legislator), with Pl. Exs. 124–25 (House and Senate votes on removal of OOP voting in 2013 by legislator).)

Defendants note an additional benefit in requiring in-precinct voting, which Plaintiffs fail to acknowledge: OOP voting actually partially disenfranchises voters. That is, when OOP was permitted, although OOP voters were permitted to vote in state-wide races, they were not permitted to vote in precinct-specific contests for which they would otherwise have been eligible had they only appeared at their assigned precinct. (Doc. 336 at 227–28.) This problem is only further aggravated by political organizations intentionally transporting voters to the wrong precinct, which in fact occurred before SL 2013–381 through, for example, the GOTV activities of one of the Plaintiffs in this case. (See, e.g., Pl. Ex. 9 at 5; Pl. Ex. 811 at 46.) Removing OOP voting thus ensures voters are fully enfranchised to vote in those contests in which they are eligible while also promoting the efficient and organized operation of the State's election system. (See Def. Ex. 132 at 2–3 (Mecklenburg CBOE director observing that OOP potentially caused a second primary in the County).)

Plaintiffs contend that there is no trial evidence of the administrative burdens caused by OOP voting or the abuse of the precinct system. (Doc. 346 at 68–69.) Defendants did, however, produce such evidence. First, SBOE Director Strach described the additional procedures required for administering OOP voting:

The county board of elections would determine ... what precinct [the voters] were properly registered in. So then they would have to go and research to see which offices they were eligible to vote for, and if those offices were different than the ballot that had been cast where they voted, then they would have to see which of the offices they voted for on that ballot they were eligible for, which ones they were not. ... [I]f that ballot was counted, it would either have to be hand counted or they would have to complete a new ballot with those [eligible] races voted for in the way that the voter voted so they could be put through the machine, the voting machine.

(Doc. 336 at 227–28; see also Pl. Ex. 817 at 43–46 (CBOE official describing the process of counting OOP provisional ballots under the old law).) Strach also testified that, after the removal of OOP voting, the review of provisional ballots to determine only whether ballots were cast in the correct precinct required minimal effort. (Doc. 336 at 228–29.)

It is true that the State's accommodation of unreported movers creates administrative burdens that could be avoided without violating the NVRA. See52 U.S.C. § 20507(e)(2). As noted above, North Carolina provides unreported movers with three Election Day voting sites, one of which is their former precinct. N.C. Gen. Stat. § 163–82.15(e). If an unreported mover elects that option, he is permitted to cast a provisional ballot and the CBOE is required to count it "for all ballot items on which it determines that the individual was eligible under State or federal law to vote." Id.§ 163–82.15(e)(iii). This means that, as with OOP ballots, the CBOE has to review the full ballot of unreported movers voting provisionally in their former precinct to tabulate and record the races for which the unreported mover is eligible to vote in his new precinct and invalidate votes cast in races for which the unreported mover is not eligible (i.e., those races for which only individuals residing in the unreported mover's old precinct are entitled to vote). See id.

For 2006 through 2012, unreported movers accounted for 25.4% of provisional ballots,¹⁸⁰ while OOP voters accounted for 14.7% of provisional ballots. (Pl. Ex. 42 at 95 (tbl. 12).) Thus, both OOP voters and unreported movers impose the same (or at least very similar) administrative burdens on CBOEs, and both involve partial disenfranchisement, yet unreported movers account for more provisional ballots than OOP voters. (Id.) The State could have removed the administrative burden caused by unreported movers, promoted full enfranchisement, and still complied with the NVRA by permitting unreported movers to vote only at their new precinct. 52 U.S.C. § 20507(e)(2)(B) (providing that a State need not permit voting at other locations if unreported movers are permitted "to vote in the current election upon oral or written affirmation" at the polling place for the voters' new address). If it had done so, CBOEs would not be required to conduct the research and hand tabulation efforts discussed above. But the fact that North Carolina expanded the precinct options for unreported movers is not sufficient to make the State's interest in administrative efficiency or full enfranchisement tenuous. First, the State could only have reduced the administrative burdens caused by unreported movers by removing voting opportunities for unreported movers, and African Americans cast a disproportionate share of unreported mover provisional ballots in 2010 and 2014. (See Pl. Ex. 689.) Second, surely the State's effort to reduce a problem is not stripped of all value simply because the State failed to completely eradicate it. Cf.Williamson, 348 U.S. at 489, 75 S.Ct. 461 ; Billups, 504 F.Supp.2d at 1381–82 ("[T]he legislature has wide latitude in determining the problems it wishes to address and the manner in which to address them (quoting Rokita, 458 F.Supp.2d at 829 )).

180 Under the statute, unreported movers voting at their new precinct or a central location could not have cast these provisional ballots, as the statute calls for them to receive regular ballots. N.C. Gen. Stat. § 163–82.15(e) ; (Def.. 368 at 59–60).

In any case, permitting unreported movers to cast provisional ballots does not undermine the State's asserted interest in a precinct-based system. As noted by the North Carolina Supreme Court in James, for voting to be timely and efficient on Election Day, officials need to be able to predict how many voters will vote in each precinct. 359 N.C. at 270–71, 607 S.E.2d at 644–45. Precincts seek to promote efficient voting on Election Day by capping this number. Id. In addition, unlike an OOP voter, an unreported mover voting a provisional ballot at the registrant's former precinct will not disturb this calculus. Because the unreported mover's change of address was not reported, the CBOE would have planned for the unreported mover to vote at his former precinct. In this sense, an unreported mover who votes at his former precinct is presenting where the State expected and planned for him to vote. Unlike OOP, there is no reason to think that unreported movers who vote at their new precinct will present in groups or disproportionately at one precinct. (Cf. Pl. Ex. 9 at 5 (demonstrating that GOTV efforts disturbed the precinct calculus by transporting groups of voters to the polls, irrespective of their correct precinct); Pl. Ex. 811 at 46.) In addition, while the number of unreported movers has an inherent consistency in that it is tied to actual moves, there is no similar cap on the number of OOP voters. For those elections from 2006 to 2012, the number of unreported movers fluctuated up and down, while the number of OOP voters increased in each election (even when moving from a midterm to a presidential election).¹⁸¹ (Pl. Ex. 42 at 95 (tbl. 12).) In this sense, only OOP had the potential to transform from exception to the rule.

181 There were 10,474 unreported movers in 2006, 11,064 in 2008, 7,410 in 2010, and 9,720 in 2012. There were 3,115 OOP voters in 2006, 6,032 in 2008, 6,052 in 2012, and 7,486 in 2012. (Pl. Ex. 42 at 95 (tbl. 12).)

For these reasons, the State's asserted interest in reducing the administrative burdens on CBOE staff is not tenuous, but not strong. The same is true of the State's interest in fully enfranchising voters. However, the State's interest in a precinct-based system, as articulated in James, is substantial.

v. Pre-Registration

Defendants offer two rationales for eliminating pre-registration. First, they argue that young people registered through pre-registration may become confused about their eligibility and registration status. (Doc. 347 at 29, 56.) Second, they contend that the vast majority of States did not offer pre-registration. (Id. at 29.)

The second justification is true, but it is unclear how much weight to give such a generic justification. The court's assessment under § 2 is local.

The first justification appears to be true to some degree. During the legislative debate, Senator Rucho noted that there was some confusion in his family when his son pre-registered, (Pl. Ex. 202 at 22), though the nature of this confusion was not clearly laid out in the legislative debates.

In other ways, however, the legislative debates do tend to highlight some of the confusion that naturally flows from the way pre-registration actually works. Under the old law, when a person under the age of eighteen "pre-registered," his pre-registration application was sent to the SBOE. 2009 N.C. Sess. Law 541, § 10.(a). When that person reached the age of eligibility, the SBOE would automatically process the registration application. Id.§ 7.(a). Thus, based on this system, one who pre-registers was not yet actually registered to vote, and until turning the right age, the pre-registration application sat in an electronic queue, waiting to be processed. (Doc. 336 at 205.) But when the young person became old enough to register, the registration application would be processed, ultimately initiating mail verification. 2009 N.C. Sess. Law 541, § 10.(a) (providing that mail verification of pre-registrants was to begin "[n]o later than 60 days prior to the first election in which the applicant will be legally entitled to vote"). Any pre-registrant who had moved between (or even within) counties since pre-registering would likely fail mail verification. Likewise, any pre-registrant who moved to and became a resident in a different county after being successfully registered would nevertheless have to register again.¹⁸²See generallyN.C. Gen. Stat. § 163–57 (defining residency for voting purposes). Of course, none of this would likely have been apparent to the pre-registrant either when pre-registering (perhaps a year earlier) or when the registration was actually processed (perhaps a year or more later). The pre-registrant would only have known that he or she had "preregistered" to vote.

182 Presumably, this is why Plaintiffs' witness, Nadia Cohen, discussed below, still had not registered to vote after she graduated from high school, since she was about to move to a different county for college.

Further, the former director of the Wake County Board of Elections, Cherie Poucher, testified that her office received calls from pre-registrants who were confused that they had never received a voter registration card. (Def. Ex. 368 at 17–18.) Ordinarily, registrants receive a voter registration card within a few weeks of submitting their registration. However, because, as shown above, pre-registrants are not actually registered until a later time, they will not receive a voter registration card in the usual time frame. (Def. Ex. 368 at 17–18.) Poucher's testimony does not establish that these type of inquiries were a significant burden on the Wake CBOE, but it does suggest that at least some pre-registrants were confused about the significance of their pre-registration.

Weighing against the justifications offered by Defendants is the fact that, in many ways, pre-registration is simpler than the current registration process. With pre-registration, an individual became eligible to pre-register on his sixteenth birthday. Without pre-registration, "eligibility to register to vote depends on age relative to Election Day; seventeen-year-olds who will turn 18 by Election Day are eligible to register." (Pl. Ex. 235 at 16.) As described by Dr. Hillygus, this connection between eligibility and Election Day creates the following complexity, given that the date of Election Day is variable:

[A] 17–year-old with a November 8th birthday was not eligible to register in 2014, while a 17–year-old with the same birthday would be able to register in 2016. More complicated still, each municipality in the state can set its own date for the general election in off-congressional election years, creating variability across the state in the specific eligibility date in odd years.

(Id.) Due to this complexity, for some time the North Carolina SBOE decided "that the best way to resolve these inconsistencies [was] to only offer voter registration to those who are 18 and older." (Id. at 16 n. 41.)

In sum, Defendants have offered some evidence that pre-registration created voter confusion, and Plaintiffs have offered evidence that the current system is more complicated. The tenuousness of the justification is examined with an eye for finding a pretext for racial discrimination. See, e.g., LULAC, 999 F.2d at 870–71. But having reviewed all the evidence in the case, this court cannot find that Defendants' proffered justifications are a tenuous pretext for racial intent. First, given the recent enactment of pre-registration, it is unlikely the General Assembly could have anticipated that pre-registration's removal would make registration more complex. After all, Plaintiffs have provided no evidence that, prior to pre-registration's enactment in 2009, the State struggled to register individuals once they reached the age of eligibility. A reasonable legislator could believe that whatever system registered generations of North Carolinians prior to 2009 would go back into place when pre-registration was removed. Second, Plaintiffs have not provided evidence that the General Assembly had statistics regarding the use of pre-registration, especially based on a racial breakdown. (Doc. 285 at 47–48 (claiming that the legislature had demographic data on SDR and early voting, but making no claim as to pre-registration); Doc. 286 at 48–50 (same).) Third, because pre-registration did not tend to favor one party over another and since most pre-registrants chose to be unaffiliated voters, racial use would have been difficult to infer from party affiliation.

Finally, the State is surely permitted to draw some lines, especially where the age of registration the State chooses is rationally tied to when individuals actually will be eligible to vote in the next general election. If the State does not have a non-tenuous interest in tying registration to when individuals become eligible to vote, then it is difficult to imagine any line that could legitimately be drawn. If sixteen-year-olds cannot be legitimately prohibited from registering, then why can fifteen-year-olds or fourteen-year-olds? Accordingly, while the evidence is mixed and the State's justifications are weaker than for the other provisions, the court nevertheless cannot find that Defendants' justifications for the repeal of pre-registration are tenuous.

* * *

One final consideration regarding tenuousness. The fact that the legislative bodies of a majority of States have not adopted the measures under consideration here is an indication that answers to these questions are far from clear-cut. No one has ever suggested that a legislature's debate over whether to adopt them would fail to reflect legitimate concerns and not involve conflicting policy considerations. If a legislature can have a good faith, legitimate debate about the wisdom of a law, then that says something about whether it can have a similar good faith dispute about its repeal.

3. Equality of Opportunity and Social and Historical Conditions

Having examined the practical impact of SL 2013–381, the Gingles factors, and other considerations in the totality of the circumstances, this court must now determine whether Plaintiffs have met the Fourth Circuit's two-element test: (1) whether the specific election changes, individually and cumulatively, "impose a discriminatory burden" on African Americans and Hispanics in North Carolina, such that they "have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice"; and (2) whether the "discriminatory burden" is "caused by or linked to social and historical conditions that have or currently produce discrimination against" African Americans and Hispanics. League, 769 F.3d at 240. "[T]he relative ability of [minority] voters to vote through the remaining options available" under SL 2013–381 is relevant to whether they have an equal opportunity to participate in the political process. Husted, 768 F.3d at 556 ; accordHolder, 512 U.S. at 880–81, 114 S.Ct. 2581 (noting that under § 2, the effect of a voting rule can be evaluated "by comparing the system with that rule to the system without that rule").

The question is not whether the voting law could be made more convenient—they virtually always can be. Rather, the question is whether the electoral system as applied treats protected classes the same as everyone else, determined by the totality of the circumstances. League, 769 F.3d at 243 (stating that the § 2 inquiry requires " 'an intensely local appraisal of the design and impact of' electoral administration") (citations and internal quotation marks omitted)); Frank, 768 F.3d at 754 ("Yet it would be implausible to read § 2 as sweeping away almost all registration and voting rules. It is better to understand § 2(b) as an equal-treatment requirement (which is how it reads) than as an equal-outcome command (which is how the district court took it)."). This is necessarily the case because § 2 cannot be read to emphasize "the word 'opportunity' at the expense of the word 'equally.' ... Section 2 does not guarantee minority voters an electoral advantage." Bartlett, 556 U.S. at 20, 129 S.Ct. 1231 (per Kennedy, J.).¹⁸³

183 By way of comparison, in measuring the nature of the burden, courts examining claims under § 5 have acknowledged that a burden must be material. Florida, 885 F.Supp.2d at 312 ("[A] change is not retrogressive simply because it deals with a method of voting or registration that minorities use more frequently, or even because it renders that method marginally more difficult or burdensome. Rather, to be retrogressive, a ballot access change must be sufficiently burdensome that it will likely cause some reasonable minority voters not to register to vote, not to go to the polls, or not to be able to cast an effective ballot once they get to the polls."); South Carolina, 898 F.Supp.2d at 39–40 ("A state voting law has a discriminatory retrogressive effect if the law disproportionately and materiallyburdensminority voters when measured against the pre-existing state law." (emphasis added)); Texas v. Holder, 888 F.Supp.2d 113, 126 (D.D.C.2012) ("Texas can prove that SB 14 lacks retrogressive effect even if a disproportionate number of minority voters in the state currently lack photo ID. But to do so, Texas must prove that these would-be voters could easily obtain SB 14–qualifying ID without cost or major inconvenience."), vacated and remanded on othergrounds, –––U.S. ––––, 133 S.Ct. 2886, 186 L.Ed.2d 930 (2013) ; see alsoNicholas O. Stephanopoulos, The South After Shelby County, 2013 Sup. Ct. Rev. 55, 110 (2013) ("Sometimes a plaintiff will be able to establish a statistical disparity between minorities and whites as well as a material burdenon voting—meaning that preclearance would have been denied ...." (emphasis added)). One court has applied the same reasoning in the § 2 context. Frank, 768 F.3d at 753 ("Although these findings document a disparate outcome, they do not show a 'denial' of anything by Wisconsin, as § 2(a) requires; unless Wisconsin makes it needlesslyhard to get photo ID, it has not denied anything to any voter.").
In similar fashion, within the context of § 2, courts have acknowledged that changes causing inconvenience are not all actionable. See, e.g., Glover v. S.C. Democratic Party, No. C/A 4–04–CV–2171–25, 2004 WL 3262756, at *6 (D.S.C. Sept. 3, 2004) ("This Court acknowledges that a second primary election may impose an inconvenience, and hence a difficulty, for those who vote a second time. However, the Court does not find that difficulty voting equates with a 'denial or abridgement' of the right to vote."), aff'd sub nom.Reaves v. S.C. Democratic Party, 122 Fed.Appx. 83 (4th Cir.2005) ; Hood, 351 F.Supp.2d at 1335 ("While it may be true that having to drive to an early voting site and having to wait in line may cause people to be inconvenienced, inconvenience does not result in a denial of 'meaningful access to the political process.' " (citations and footnote omitted)).

a. Voter ID¹⁸⁴

184 Notably, the United States does not make a § 2 results claim against the current version of North Carolina's ID law. (Doc. 419 at 73 n.26.) The United States' only ID-related claim is that SL 2013–381 was passed with discriminatory intent.

Plaintiffs' § 2 challenge of the photo-ID requirement with the reasonable impediment exception presents an issue of first impression. Only South Carolina has such a law, and it was upheld as non-retrogressive under § 5 by the three-judge panel in South Carolina, 898 F.Supp.2d 30. As noted above, § 2 and § 5 require different inquiries. SeeLeague, 769 F.3d at 240. Nevertheless, the three judge panel's examination of the burden imposed by South Carolina's very similar voter-ID law is useful to this court's examination of whether North Carolina's law deprives minority voters of an equal opportunity to participate in the political process.

Plaintiffs claim South Carolina is distinguishable because voter ID is easier to acquire in South Carolina. In South Carolina, voters can acquire qualifying ID "at each county's election office ... [and] at each county's DMV office." South Carolina, 898 F.Supp.2d at 32. Each of South Carolina's forty-six counties has at least one elections office and DMV office. Id. at 33–34. "[S]ome of the more populated counties" have "more than one DMV office." Id. at 34. One option for South Carolina voters is to acquire a voter registration card bearing a photo. (Pl. Ex. 1064 at 26–28.) This form of photo ID can only be acquired through county election offices, not the DMV. (Id.) If the voter is already registered, the only supporting document that is required to receive a photo registration card is the voter's registration card. (Id. at 29–30.) New voters are required to show a HAVA document bearing their name and address. (Id. at 30.) No-fee voter ID does not expire in South Carolina. (Id. at 29.) The record does not establish what type of supporting documentation is required to receive a qualifying ID from South Carolina's DMV offices.

In North Carolina, the DMV is the exclusive provider of a free voter ID. (Pl. Ex. 1044 at 142.) The DMV currently has 114 brick and mortar sites that provide drivers' license services. (Doc. 410 at 164, 168–69.) Some brick and mortar sites have limited hours, and sixteen of North Carolina's 100 counties do not have a brick and mortar site. (Id. at 203–04; Pl. Ex. 241 at 13; Pl. Ex. 1044 at 143–144, 166–167.) Eleven of these counties are serviced by five DMV mobile units, which currently appear at twenty-four mobile sites. (Doc. 410 at 198–99; Pl. Ex. 241 at 13.) No mobile site is serviced more than three days per month. (Doc. 410 at 204; Pl. Ex. 241 at 13 n.3.) The DMV plans to increase the number of mobile sites from twenty-four to seventy by the summer of 2016. (Doc. 410 at 211.) Although a broad number of documents are acceptable, the DMV requires voters to present two supporting documents proving the voter's age and identity. (Def. Ex. 533 at 2 (tbl. 4).) A document establishing only the voter's name and address, such as a HAVA document, does not qualify. (Id.) In addition, no-fee voter-ID cards do expire in North Carolina, but as a practical matter they can be used for twelve years for voters under seventy and permanently thereafter.¹⁸⁵

185 No-fee voter IDs follow the same expiration rules as other non-operator IDs. (Pl. Ex. 1047 at 31.) If the voter is sixty-five or older, the ID expires five years later on the voter's seventieth birthday. (Id.) If the voter is younger than sixty-five, the ID is valid for eight years. (Id.) Of course, the voter ID remains a valid voting credential for four years past its expiration. N.C. Gen. Stat. § 163–166.13(e)(1), (2). In addition, if a voter's no-fee ID expires after his seventieth birthday, the ID remain a valid voting credential for the rest of the voter's life. Id.§ 163–166.13(f).

While at first glance it appears that brick and mortar locations are more evenly distributed among counties in South Carolina, there was no evidence of the average drive time to an ID issuing location for voters in South Carolina, a largely rural State. South Carolina may have at least two ID issuing offices in each county, but it has less than half as many counties as North Carolina, despite being more than half the size. State Area Measurements and Internal Point Coordinates, United States Census Bureau, https://www.census.gov/geo/reference/state-area.html (last visited April 12, 2016).¹⁸⁶ In North Carolina, the DMV estimates that 98% of the DMV's "market population" (those age 15 and older) lives within a thirty-minute drive of a DMV license service station, be that a brick and mortar site or a mobile site. (Doc. 410 at 168–69.) With the addition of planned mobile units in the coming months, DMV will have a brick and mortar or mobile site within a twenty-minute drive of 98% of the DMV's market population. (Id. at 175–76.) In addition, while no evidence was presented on the hours for ID-issuing offices in South Carolina, North Carolina has an extended hours program that provides weekday hours until 6 p.m. at twenty-one locations and Saturday hours at eleven locations. (Id. at 168; Pl. Ex. 664 ¶¶ 80–83.) When the extended-hours program was offered at nineteen locations (now twenty-one), 86% of DMV's market population lived within a thirty-minute drive of an extended-hours office. (Doc. 410 at 168.)

186 Fed. R. Evid. 201.

Nevertheless, from January 2014 to January 2016, the DMV issued 2,139 no-fee voter IDs, (id. at 177–79; Def. Ex. 494), whereas South Carolina issued 31,156 no-fee IDs during the first two years of its ID law,¹⁸⁷ (Doc. 1055 at 1 (tbl. A)). There are at least two inferences that can be drawn from this disparity. The first, which Plaintiffs urge, is that a free voter ID is easier to acquire in South Carolina. But this is not apparent. As of January 2016, only thirty people had presented to the North Carolina DMV with their name, date of birth, and a SSN all matching the Social Security Administration's database but failed to receive a no-fee ID card. (Doc. 410 at 178–79; Def. Ex. 495.) And as of trial, that number was reduced. (Doc. 410 at 178.) Since April 2015, only three persons had presented to the DMV for a free voter ID but been unable to satisfy the requirement that their personal information match the Social Security Administration database. (Id. at 179.) Moreover, Plaintiffs presented evidence of only two sisters who sought a free ID but were unable to obtain one. By contrast, one of their witnesses, Mr. Alonzo Phillips, candidly testified that although he was told by poll workers he would need a photo ID in 2016, he "didn't really think they ... were serious," "figured they would let [him] still vote with [his] registration card," and thus "forgot all about it." (Pl. Ex. 1048 at 42–43.) Based on the record, the evidence in support of Plaintiffs' inference is far from strong.

187 South Carolina issued 6,808 free voter IDs in 2015 and 553 thus far in 2016, bringing the total issuances to date to 38,517. (Pl. Ex. 1055 at 1 (tbl. A).)

The second, and perhaps more reasonable, inference, is that fewer people need to acquire a qualifying ID in North Carolina. As noted above, even though Dr. Stewart had a unique identifier in South Carolina, he matched a higher percentage of active and inactive voters in North Carolina than he did active voters in South Carolina. (Pl. Ex. 891 at 19 (tbl. 11); Def. Ex. 504 at 36 (tbl. 4).) This is true when broken down by race, and the magnitude of the difference only becomes more pronounced when the percentage of active voters matched in North Carolina is compared to the percentage of active voters matched in South Carolina, which is the more accurate comparison. (Pl. Ex. 1063; Def. Ex. 504 at 36 (tbl. 4).) Most importantly, the estimate provided by Dr. Stewart, which this court finds overestimates the number of North Carolinians without acceptable photo ID, also exceeds the estimates the court ultimately relied upon in South Carolina.¹⁸⁸

188 Seesupranotes 57–58.

It is difficult to know with any reasonable assurance how many voters still lack a valid photo ID or, among them, those who lack one due to any burden in acquiring it. However, there is little evidence that the latter is so large a group as to overwhelm election officials if those voters seek to vote under the reasonable impediment exception. For example, as shown by Dr. Thornton, of the 172,098 individuals that the SBOE contacted from Dr. Stewart's most recent no-match list, 72.1% did not vote in the 2012 presidential election and 39.5% had never voted in any election. (Def. Ex. 511 at 18 (tbls. 6, 7).) This is likely because 33% of individuals on Dr. Stewart's most recent no-match list are inactive voters. (Doc. 416 at 50.) Dr. Hood's analysis of reasonable impediment voting in South Carolina only bolsters this conclusion. (See Def. Ex. 500 at 2 (finding 1.10 reasonable impediment ballots cast per 10,000 votes in South Carolina during the 2014 midterm general election).)

Any difference in the ease of acquiring photo ID is where material distinctions between North Carolina and South Carolina end. North Carolina's reasonable impediment exception is materially indistinguishable from South Carolina's exception that received judicial preclearance in South Carolina, 898 F.Supp.2d 30. South Carolina's reasonable impediment provision permitted voters to vote without photo ID "so long as they fill[ed] out an affidavit at the polling place and indicate[d] the reason that they [had] not obtained" a qualifying photo ID. Id. at 35. A key question in that case was whether the exception would be interpreted broadly or narrowly. Id. Although the law provided that a voter's statement of reasonable impediment must be accepted unless it is false, merely denigrating to the photo-ID law, or nonsensical,¹⁸⁹ the law did not provide examples of what might constitute a reasonable impediment or establish how the law was to be construed. Id. at 36 & n. 5. Accordingly, the court leaned heavily on representations by the South Carolina Attorney General and the Executive Director of the board of elections, who provided an official interpretation of the law and described how it would be implemented. Id. at 35–36. Both officials provided a broad interpretation of the provision and stated that a driving principle of its implementation would be "erring in favor of the voter." Id. at 36. The attorney general also provided examples of reasons that would constitute a reasonable impediment and convinced the court that, so long as the statement was not false, "[a]ny reason that the voter subjectively deems reasonable will suffice." Id. The court ultimately adopted this broad interpretation as a condition of preclearance. Id. at 37. The court required that "filling out [the affidavit] must not become a trap for the unwary, or a tool for intimidation or disenfranchisement of qualified voters." Id. at 40. Further, the court required the reasonable impediment form to "have separate boxes that a voter may check for 'religious objection'; 'lack of transportation'; 'disability or illness'; 'lack of birth certificate'; 'work schedule'; 'family responsibilities'; and 'other reasonable impediment.' " Id. at 41. Finally, the court mandated that the form may "require a further brief written explanation from the voter only if he or she checks the 'other reasonable impediment' box." Id. So implemented, the court found that filling out the form would not constitute a material burden, at least under the VRA. Id.

189 North Carolina's "merely denigrates," "obviously nonsensical," and "factually false" provisions track exactly the provisions in South Carolina's law precleared by the three-judge panel based on the broad interpretation provided by the State. South Carolina, 898 F.Supp.2d at 36–37 & n. 5.

Upon close examination, North Carolina's reasonable impediment provision is effectively a codification of the three-judge panel's holding in South Carolina. As noted above, a voter's reasonable impediment declaration can only be rejected if it is false, merely denigrating to the photo-ID requirement, or obviously nonsensical. N.C. Gen. Stat. § 163–182.1B(a)(1). The law does not permit a voter's declaration to be denied on the ground that it is not reasonable. Id.§ 163–182.1B(b)(6). Only the voter's subjective belief is relevant to the issue of reasonableness. See id. Finally, the law requires all reasonable impediment forms to, "at a minimum," contain practically the exact same categories required by the court in South Carolina. The only omission is that the law does not require a box for "religious objection," id.§ 163–166.15(e), but this is because a separate provision of North Carolina law grants an exception for those with religious objections to having their photo taken, id.§ 163–166.13(a)(2). In fact, the law goes a step beyond what was required in South Carolina by requiring that a box be listed for "[l]ost or stolen photo identification." Id.§ 163–166.15(e)(1) f. As in South Carolina, a voter need only provide a written explanation if one of the provided boxes does not apply. Id.§ 163–166.15(e)(1) h.

Plaintiffs have failed to show that the reasonable impediment voting process deprives any group of an equality of opportunity to participate in the political process. As noted above, voters need only complete two forms as part of the reasonable impediment voting process. (Def. Ex. 546 at 3.) Most voters will have significant parts of these forms pre-populated for them, and, in any case, voters can receive as much assistance as necessary from either a person of their choosing or a poll worker. (See id. at 3; Doc. 414 at 138–39, 189, 211 ("[T]he precinct official is supposed to do everything they can to try to provide as much explanation to [the reasonable impediment voter] as possible until they do understand it.").) Poll workers have been trained to provide assistance without inquiring into whether the voter is illiterate. (Doc. 414 at 189.) As with SDR, poll workers are trained to ensure that the declaration is complete before accepting it. (Id. at 211–12.) Moreover, although Plaintiffs failed to provide any expert testimony on the literacy level required to complete the reasonable impediment process, this court is convinced that it is no more demanding or intimidating than other voting forms. See supra Part I.D.1.e. As noted above, the first step of the reasonable impediment voting process—the provisional voting application form—must also be completed by all provisional voters, including OOP voters. (See Def. Ex. 546 (Ex.1).) In addition, all voters must complete a registration form and an ATV form. (Doc. 410 at 91–93.) To complete the ATV form, the voter must attest that the address he provided is correct and that he has not voted in the election. (Pl. Ex. 1056.) As noted above, the fact that so many minority voters have completed these forms in the past undermines Plaintiffs' argument that the reasonable impediment voting process is burdensome, especially in light of the substantial assistance that is available in completing the reasonable impediment declaration. Finally, Plaintiffs cannot claim that the alternative identification requirement of the reasonable impediment exception (HAVA document, registration card, or SSN4 and date of birth) is impermissibly burdensome because SDR, which plaintiffs claim is necessary to avoid a § 2 violation, required voters to present at least a HAVA document.

Plaintiffs have also failed to show that the reasonable impediment challenge process is likely to be applied in an arbitrary or discriminatory manner. The challenge process is designed to place every burden on the challenger and give every benefit to the voter. See supra Part I.D.1.e. The challenger must make a public records request for the declaration he seeks to challenge, (Def. Ex. 547 at 1.), and complete and submit the challenge on the SBOE's Evidentiary Challenge form "no later than 5:00 P.M. on the third business day following the election," N.C. Gen. Stat. § 163–182.1B(b)(2) ; (Def. Ex. 547 at 7). The challenge form must be notarized, and fraudulently or falsely completing it is a felony. (Def. Ex. 547 at 7.) At a hearing, the scope of the challenge is "strictly limited" to the facts the challenger alleged in the written challenge form, (id.), and the challenger bears the burden of proof and persuasion, (id. at 4). The CBOE must reject the challenge unless, "having considered all facts in the light most favorable to the voter, the challenger has shown by clear and convincing evidence that the stated impediment (1) merely denigrates the photo identification requirement, (2) is obviously nonsensical, or (3) is factually false." (Id.) If the voter has checked one of the template boxes (e.g., lack of transportation), then the challenger can only prevail under the factual falsity provision, as the SBOE considers the template boxes to be non-denigrating and not nonsensical as a matter of law. (Id. at 6.) Moreover, the ability to bring a successful challenge under the factual falsity provision is very limited, as a CBOE cannot "find a challenge valid if it provides only evidence regarding the reasonableness of the impediment." N.C. Gen. Stat. § 163–182.1B(b)(6). In addition, Director Strach's testimony revealed that the SBOE interprets many of the challenges that Plaintiffs view as testing the "falsity" of the impediment as only relevant as to reasonableness. See supra Part I.D.1.e. Tellingly, despite also permitting the rejection of "factually false" impediments, the Executive Director of the South Carolina State Election Commission was not aware of a single reasonable impediment affidavit being challenged, much less rejected. (Pl. Ex. 1064 at 37–39.) South Carolina's implementation of its ID requirement was deemed a "nonevent." (Pl. Ex. 1064 at 92.)

Moreover, Plaintiffs have failed to show that CBOEs are likely to misapply the challenge process. The SBOE has provided clear guidance to CBOE's on the procedures to be followed. (See Def. Ex. 547.) CBOEs have been trained on the meaning of "clear and convincing evidence" and "in the light most favorable to the voter," (Def. Ex. 551 at 55), and county attorneys will be present at hearings to assist the CBOE in understanding and properly applying the legal standards, (Def. Ex. 547 at 4). All Evidentiary Challenge forms must be forwarded to the SBOE. (Id. at 3.) The SBOE plans to use its legal team to provide oversight of challenges and ensure CBOEs are following proper procedure. (Doc. 414 at 215–16.) The SBOE also has the ability to exercise its supervisory authority over CBOEs to correct any misapplication that may occur. (Doc. 414 at 138, 216–17.)

Plaintiffs have also failed to show that giving reasonable impediment declarants a provisional ballot is likely to impose a material burden on the right to vote. This issue was addressed by the panel in South Carolina, which observed:

[T]he word "provisional" is a bit of a misnomer in this instance. [Provisional ballots cast due to a reasonable impediment] must be counted and will be counted, at least so long as the voter does not lie when he or she fills out and signs the reasonable impediment affidavit. Counting the reasonable impediment ballots will not differ in substance from the counting of absentee ballots. When the provisional ballot process operates in this way, casting a provisional ballot instead of a regular ballot does not burden the right to vote.

898 F.Supp.2d at 41. The same is true here, as State law provides for counting these ballots, despite their "provisional" label.

Plaintiffs have also failed to show that the State's education and training efforts have been insufficient. The State has engaged in a substantial multi-media voter education program. See supra Part I.D.1.b. Of central significance, after the fall elections in November 2015, the SBOE sent every individual who received a prior mailing describing the need for voter ID (315,755 voters)—except those who reported they already possessed acceptable photo ID and those for whom prior mailings were returned as undeliverable—an additional mailing describing the reasonable impediment exception and other exceptions to the photo-ID requirement. (Def. Ex. 535 at 11.) CBOE training on the reasonable impediment exception and associated procedures has been detailed and will continue as the November 2016 presidential election approaches. See supra Part I.D.1.b.

Finally, Plaintiffs' reliance on evidence that voter ID laws depress turnout is misplaced. Not only did Kansas and Tennessee dispute the results of a study reporting a two percent decrease after they adopted their ID laws, (Doc. 407 at 75–77), other reliable evidence demonstrated that Georgia's ID law depressed turnout by 0.4%, (Doc. 410 at 220). Most importantly, however, none of these States had a reasonable impediment exception; South Carolina is the only State to do so, and no data were presented as to its effect on turnout.

In sum, this court reaches the same conclusion as the court in South Carolina: North Carolina's voter ID law with the reasonable impediment exception does not impose "a material burden" on the right to vote of any group "for purposes of the Voting Rights Act." South Carolina, 898 F.Supp.2d at 41. To put it in § 2 terms, North Carolina's voter ID law with the reasonable impediment exception does not deprive African Americans and Hispanics of an equal opportunity to participate in the political process, as compared to other groups.

b. Early Voting

Session Law 2013–381 reduced the number of early-voting days but kept the same number of early-voting hours. The General Assembly intended this change to make early-voting locations more numerous and evenly distributed, and the evidence shows the law achieved these legitimate ends. In 2014, with the ten-day voting schedule in place, there were more voting sites and more high convenience night and weekend hours.¹⁹⁰ There is reason to believe that these new hours will benefit African Americans. (Doc. 332 at 183–84 (Pastor Gregory Moses: stating that most of his congregation does not "have the flexibility to be able to leave their jobs and go and vote").) While one Sunday was removed,¹⁹¹ Plaintiffs failed to show that souls-to-the-polls organizations will not benefit at least equally from the new night and weekend hours, especially given testimony that showed that churches offer transportation services during most of early voting, rather than just on Sundays. SeeFlorida, 885 F.Supp.2d at 337 (noting that "the record evidence suggests that GOTV groups could adjust to a redistribution of the total 96 hours over a different number of days, including weekend days and a 'souls-to-the-polls' Sunday"); Brown, 895 F.Supp.2d at 1253–54. In addition, Plaintiffs' predictions for long lines in the 2016 presidential election were speculative, flawed, and unpersuasive. Therefore, what SL 2013–381 took away in early-voting days, it gave back in equal hours that appear to be even more convenient. This is demonstrated by the 2014 election data, where African American turnout actually increased after the transition from seventeen days of early voting to the new schedule.

190 In Florida's VRA § 5 lawsuit seeking pre-approval of the reduction in early-voting days, the court found that "the negative effect of reducing the number of days from 12 to 8 would likely be offset by the ameliorative effects of adding non-working weekday hours, a Sunday, and additional weekend hours." Florida, 885 F.Supp.2d at 336–37. The court even found that "[e]xpanding convenient non-working weekday and weekend voting hours should therefore help third-party [GOTV] efforts to provide transportation to the polls for such voters." Id.The Sixth Circuit similarly noted the importance of evening hours for lower-income workers with less flexibility. Husted, 768 F.3d at 555. Plaintiffs' fact witness evidence supports this conclusion. Several voters testified to problems they encountered voting during business hours when they worked. (E.g., Doc. 330 at 165 (Terrilin Cunningham: works 8:00 a.m. to 11:00 a.m. weekdays; noon to 11:00 p.m. Monday-Wednesday; and Mary Kay sales Thursday evening, Friday evening, and Saturday morning); Pl. Ex. 688 at 13–14 (Lynnette Garth: bus driver, speaking as to her daughter); Doc. 334 at 151, 158–59 (Michael Owens: relied on girlfriend for transportation who had to be at work during the day; now has auto and works 8:00 a.m. to 5:00 p.m.); Doc. 329 at 55 (Gwendolyn Farrington: works 12–hour days).)

191 The available evidence showed that most counties did not utilize both Sundays even when they were available. (SeeDoc 126–4 at 45–90; Def. Ex. 13); McCrory, 997 F.Supp.2d at 373.

Further, Plaintiffs have failed to link the claimed effect of the removal of the first seven days of early voting to social and historical conditions. While Plaintiffs established that African Americans at times disproportionately used the removed seven days,¹⁹² the evidence showed that they were not marginal voters. Instead, they were the more sophisticated voters who vote regardless of the practices in place, which is likely part of the reason Plaintiffs were unable to show that African Americans were habituated to the first seven days of early voting.¹⁹³ For example, regardless of race, those who voted in the first seven days under the seventeen day schedule in 2012 were more likely to vote in 2014 than those who voted in the last ten days of early voting in 2012. In fact, under Plaintiffs' own theory, those with socioeconomic limitations, including less education, are less likely to become interested in an election until close to Election Day. There is no reason to think that these voters are more likely to vote during the removed seven days of early voting, and the evidence bears this out.

192 Plaintiffs provided no evidence that Hispanics disproportionately used the removed days of early voting. (See, e.g., Doc. 346 discussing Hispanics only with regard to SDR, OOP, and pre-registration.)

193 AccordFlorida, 885 F.Supp.2d at 337 (finding that GOTV efforts could adjust to the redistribution of hours over fewer days).

Accordingly, for reasons similar to those articulated in Florida, 885 F.Supp.2d at 350–51, it does not appear that Plaintiffs can even show that SL 2013–381's changes to the early-voting schedule are retrogressive (i.e., that they make early voting more difficult for any group, including African Americans, than it was under the seventeen-day schedule). Certainly, considered under the totality of the circumstances, Plaintiffs have failed to show that SL 2013–381's changes to the early-voting schedule interact with social and historical conditions to make it more difficult for African Americans and Hispanics to participate under the revised early-voting schedule than for other groups. SeeBrown, 895 F.Supp.2d at 1241–42, 1255.

c. SDR

As noted above, North Carolina's twenty-five day cut-off for registration existed both before and after SL 2013–381. When SDR existed, the only way an individual could register after the cutoff and vote in the upcoming election was through SDR. After SL 2013–381, all registrations must be submitted by the cut-off.

Plaintiffs established that African Americans disproportionately used SDR.¹⁹⁴ They claim use reflects need for multiple reasons. First, they claim SDR provides a critical "fail-safe" for poor African American voters by providing an exception to the registration cut-off (the "fail-safe rationale"). In addition, because registration and voting can no longer be done at the same time, GOTV organizations have had to move their registration efforts up to comply with the cut-off. (Doc. 332 at 193) (Pastor Gregory Moss: stating that SDR created a "force multiplier").) Second, they claim SDR assisted low literacy individuals because election officials would "insure that applications for registration through SDR were properly completed" (the "assistance rationale"). (Doc. 346 at 35.) The evidence presented at trial, however, undermined Plaintiffs' claims.

194 Plaintiffs also showed that Hispanics disproportionately used SDR when it was in place. Seesupranote 111. As noted above, Plaintiffs were not able to make a showing of discrimination against Hispanics. In addition, Plaintiffs do not claim that SDR's removal imposes a heavier burden on Hispanics than African Americans.

Other than disproportionate use, the best facts for Plaintiffs were that African Americans are more likely to move between counties, and thus more likely to need to re-register, and that African Americans are more likely to end up in the incomplete registration queue.¹⁹⁵See supra Part I.D.3.

195 Plaintiffs did not provide incomplete queue data for Hispanics. (SeePl. Ex. 633 at 5 (tbl. 1).)

It is easy to see a connection between certain reasons for ending up in the incomplete registration queue and literacy.¹⁹⁶ But at the end of the day, these statistics are more a function of North Carolina's registration requirement—which has not been challenged—than a reflection of a need for SDR. The "fail-safe" rationale offered by Plaintiffs is not persuasive evidence that the cut-off date is unreasonable; the registrants in the incomplete queue are not there because they failed to submit their application by the twenty-five day cut-off. The incomplete queue does support Plaintiffs' assistance rationale, but there is no indication that the registrants were there because they lacked other available opportunities to complete their registration properly. Rather, the evidence revealed that North Carolina provides multiple ways for voters to comply with the registration requirement prior to the cut-off (which is five days more beneficial to voters than the NVRA's thirty-day requirement), most of which permit assistance.

196 The incomplete queue data provided by Plaintiffs is broken up by the reason the application was placed in the queue and by race. (Pl. Ex. 633 at 5 (tbl. 1).) Some reasons for ending up in the queue appear to have a potential connection to literacy, such as "missing signature," "date of birth missing," "incomplete address," "missing name," "incomplete data," "citizenship box not checked," "card missing/unreadable," and "need registration application." (Id.) It is, of course, possible that some individuals may just not be citizens or not know their birthday. Others, however, are less clearly connected, including "not in county," "other reason," "geocode conflict," and "underage." (Id.) In addition, one category provides no reason, even though twenty-eight people were evidently rejected for that "reason." (Id.)

The remaining ways for voters to register with assistance are sufficient to provide African Americans with an equal opportunity to participate in the political process. SeeHusted, 768 F.3d at 556 (stating that the "relative ability of African American voters to vote through the remaining options" is "clearly relevant" to whether they have "less opportunity" to vote than other groups). Registration forms can be downloaded from various federal, State, and local government websites. The forms can also be retrieved in person at the SBOE, any CBOE, public libraries, public high schools, public colleges, and military recruitment offices. SeeN.C. Gen. Stat. §§ 163–82.21, –82.22, –82.23; (Doc. 126–1 at 4). One may have a number of persons assist in completing the form, and the applicant can then submit the registration by mail, in person, or have another person submit the form for them.¹⁹⁷SeeN.C. Gen. Stat. § 163–82.6(a) ; (Doc. 341 at 17; Doc. 331 at 41–42). In addition, one may receive assistance at a voter registration drive, such as those by souls-to-the-polls and other GOTV organizations. (Doc. 332 at 197 ( [T]his last election cycle [2014] ... we had about 150 people who went out and touched doors to help get people registered."); Pl. Ex. 793 at 27–28.) Moreover, although Plaintiffs claim that socioeconomically disadvantaged individuals are less likely to seek the services of the DMV, the fact remains that when citizens appear at any of the following State agencies for services, they must be offered the chance to register to vote: the DMV (including when seeking a no-fee voter ID if the applicant is not registered), public assistance agencies (including county departments of social services and departments of public health); disability services agencies, sometimes at the registrant's home (including vocational rehabilitation offices, departments of services for the blind, departments of services for the deaf and hard of hearing, and departments of mental health services); and the Employment Security Commission. SeeN.C. Gen. Stat. §§ 163–82.19, –82.20; (Doc. 126–1 at 5). Session Law 2013–381 expanded this list by adding "[s]enior centers or facilities operated by the county" and "[p]arks and recreation services operated by the county" as public offices where a county may offer voter registration, so long as the SBOE, CBOE, and the county board of commissioners provide approval. 2013 N.C. Sess. Law 381, § 5.1. Finally, information on how to register is printed in the Judicial Voter Guide mailed before every election to every household in North Carolina having an address maintained by the United States Postal Service. N.C. Gen. Stat. § 163–278.99E(a) ; (Doc. 126–1 at 6).¹⁹⁸

197 The statute authorizes submission also by facsimile or scanned transmission. N.C. Gen. Stat. § 163–82.6(a). The SBOE website and registration form require delivery of a copy with the registrant's original signature. This discrepancy was not addressed at trial.

198 The statute requires that it be mailed not more than twenty-eight days and not less than seven days before one-stop (early) voting begins for each primary and general election. N.C. Gen. Stat. § 163–278.99E(a). Thus, it does not provide sufficient notice to register for the immediately upcoming election; its repetition provides additional notice and a reminder otherwise.

The ease of registration in North Carolina is perhaps why African American registration rates are so robust. Plaintiffs wish to attribute this to SDR, but the facts do not bear this out. For example, by the court's calculation, even when SDR registrations are not included African American registration rates nearly approximated white registration rates in 2008 and exceeded them in 2010 and 2012. (Pl. Ex. 40 at 35 (Ex. 14); Pl. Ex 684 (tbl. 15).)

Registration without SDR¹⁹⁹

199 The numbers in the accompanying table simply reflect the removal, by race, of the same-day registrants in the 2008, 2010, and 2012 primary and general elections, (Pl. Ex. 40 at 35 (Ex. 14) (Gronke)), from the registered voters in those years, (Pl. Ex. 684 (tbl. 15)). SDR registrations during both primaries and general elections were considered. To obtain the registration rate without SDR, that number was further divided by the VAP of each race. (Id.) The same examination of Dr. Stewart's registration data during early voting by race during 2008, 2010, and 2012 yields similar results. (Pl. Ex. 42 (Ex. 31).) Dr. Gronke reports considerably more SDR registrants than Dr. Stewart, which may be attributable to the fact that Dr. Gronke studied "new and changed registrations," (Pl. Ex. 40 at 35), while Dr. Stewart studied only new registrants, (Pl. Ex. 42 at 43). The court used Dr. Gronke's data because it was the most beneficial to Plaintiffs. For example, Dr. Stewart found that there were 38,697 SDR registrations by African Americans during early voting in the 2008 general election, (Pl. Ex. 42 (Ex. 31)), whereas Dr. Gronke reported that there were 90,603, (Pl. Ex. 40 at 35 (Ex. 14)).

African African White White % American American Registration Registration Differ Registration Registration without SDR Rate without ential without SDR Rate without SDR SDR 2008 1,240,979 86.93% 4,429,276 87.41% -0.48% 2010 1,315,531 87.55% 4,490,771 85.69% 1.86% 2012 1,387,089 88.55% 4,579,363 85.04% 3.51%

African American turnout in general elections also exceeded white turnout in 2008 and 2012 even when all SDR registrants are considered voters but excluded from the turnout figures. (Pl. Ex. 40 at 35 (Ex. 14); Pl. Ex. 242 at 161 (App'x U).)

Voting without SDR during general elections²⁰⁰

200 The numbers in the accompanying table were derived simply by subtracting the number of SDR registrants, by race, in the 2008, 2010, and 2012 general elections, (Pl. Ex. 40 at 35 (Ex. 14) (Gronke)), from the number of voters, by race, in those years, (Pl. Ex. 242 at 161 (App'x U)). This number of course assumes, to Plaintiffs' benefit, that every SDR registrant voted. To get the voting rate without SDR, that number was further divided by the VAP of each race. (Id.) The same discrepancy described above between Dr. Gronke's and Dr. Stewart's data applied to this calculation.

African African White White % American American Votes Turnout Differential Votes Turnout without without without without SDR SDR SDR SDR 2008 886,751 62.11% 3,048,148 60.15% 1.96% 2010 524,322 34.90% 2,038,747 38.90% -4.01% 2012 954,605 60.94% 3,129,170 58.11% 2.83%

This does not support Plaintiffs' claims that SDR was necessary for African Americans to have equal opportunity to participate in the electoral process, but rather is strong evidence that the rise in African American participation is attributable to other factors—including President Obama's candidacy and North Carolina's rising role as a battleground State.

Further, the proportion of African American registrants during the 2014 early-voting period was virtually identical to the proportion of African American registered voters as of 2014. (Def. Ex. 309 at 76.) This is perhaps why Plaintiffs were not able to show that African Americans are habituated to SDR—the SDR data suggest they are much more adaptable than Plaintiffs claim.

Accordingly, while the inter-county mover and incomplete queue evidence favor Plaintiffs, they fail to carry Plaintiffs' burden. The evidence seriously undermines Plaintiffs' argument that SDR is responsible for African Americans' lead over all races in registration since 2008.

Plaintiffs further attempted to demonstrate a burden from the removal of SDR and link it to social and historical conditions through the use of several fact witnesses. Plaintiffs' witnesses were variously African American, white, Hispanic, and "youth." Yet their testimony was mixed. For example, Carnell Brown (African American) who is illiterate, attempted to vote in a county in which he did not reside, rather than his county of residence. (Pl. 680 at 14–15.) SDR would not have permitted him to register or vote at an early-voting site in a county in which he did not reside.

Most of the other voters testifying at trial had moved into a new county in North Carolina without updating their voter registrations. But the race of these voters played no role in their failure to vote.

For example, two of the witnesses are enrolled in higher education programs. William Kittrell, a college student and aspiring English teacher, was unable to vote in 2014 because when he moved from Vance County to Guilford County for college, he never updated his voter registration. (Doc. 330 at 37–42.) He had been living in Guilford County for a year before the election, but had never availed himself of the opportunity to update his registration during that time. (Id. at 41–42.) Quisha Mallette is a student at the University of North Carolina School of Law. (Pl. Ex. 792 at 7.) She has lived in various parts of the United States throughout her life. (Id. at 8–10.) Before starting law school, she worked for AmeriCorps' Literacy Council, managing a youth academic support program. (Id. at 10.) She may have voted absentee by mail in the past. (Id. at 11.) She testified that, when she moved from Wake County to Orange County, she had attempted to update her voter registration through the United States Postal Service website, but she had not successfully done so. (Id. at 16–18.)

Plaintiffs also presented the testimony of two African American service personnel. Dale Hicks, a former sergeant in the United States Marine Corps, lived in New York City before joining the Marines as a data technician. (Doc. 329 at 69–70.) Since leaving the Marines and moving to Raleigh, North Carolina, he has been working as an IT professional. (Id. at 70.) He tried to vote early in 2014 but had to cast a provisional ballot because he had not updated his registration since moving from Onslow County. (Id. at 71–72.) He thereafter registered to vote in Wake County, where he resides. (Id. at 75.) Alexander Ealy, a current sergeant in the United States Army, grew up in Tennessee and moved to New York when he turned twenty-one. (Pl. Ex. 713 at 6.) When he registered to vote in North Carolina, there was an issue verifying his address because he did not denote his separate mailing and physical addresses, as the form requires when applicable. (Id. at 23–24, 32–37.) The United States Army and the United States Postal Service appear to rely on these addresses being correct in order for servicemen to receive their mail correctly.

Nadia Cohen is an Hispanic high school student from Cary, North Carolina; at the time of trial she was enrolled to start college at the University of North Carolina at Chapel Hill in fall 2015. (Doc. 331 at 161–63.) She turned eighteen in time to qualify to vote in the 2014 general election, but she failed to register in time. (Id. at 163.) In fact, when she testified at trial, she still had not registered to vote. (Id. at 169.) When Plaintiffs' counsel asked her why not, she answered:

Well, first, I figured I would register when I get to Chapel Hill since I am moving there, but, honestly, I am a little discouraged. It's not something that I'm, like, running to go do. If I wasn't raised to believe that voting was so important, I probably wouldn't. It's just—it just made things a lot more difficult for me than it has to be.

(Id.) Ms. Cohen learned that she had failed to register in time for the upcoming election from her father, who had looked up the election rules online. (Id. at 166–67.) When asked why she had not registered earlier, on her own, she responded,

I didn't know there was a registration deadline. I didn't know I could do—I couldn't do same-day registration. And

it's not that I don't, like, pay attention to the news or anything. It is just my two main sources of information, which are my parents and my school, either didn't know or didn't tell me, or at least not with enough time.

(Id. at 168.) When asked if she had done any research into registration deadlines at all before the 2014 general election, she testified:

No. It's not something that particularly interests me. I just assumed that it would be as it had been for my older brother and my older sister and my parents, you know, a convenient location, you know, I wouldn't have to go out of my way. My parents registered when we moved to North Carolina and they got their North Carolina driver's license. My brother registered in school. No one had to go out of their way to register, and I thought that, you know, it would be the same for me.

(Id. at 168–69.) Of course, because Ms. Cohen would have been eighteen by the time of the 2014 November general election, she could have registered and voted during the primary earlier that year. N.C. Gen. Stat. § 163–59. But her father never told her that. (Doc. 331 at 172.) Ultimately, Ms. Cohen did not vote anytime during 2014 because, in her words, "honestly, voting is not my top priority throughout the year. I'm busy with high school and work and other after-school activities. It is not—it is not something that I feel like I need to always be invested in order to know." (Id. at 173.)

These witnesses illustrate the type of testimony received at trial. Mr. Brown tried to vote in the wrong county, an issue that no prior or current election mechanism in North Carolina could have solved. The college and law school students had all the opportunity needed to update their registrations but failed to do so. In any case, they are not uneducated or socioeconomically disadvantaged. The military servicemen are not originally from North Carolina and could not have been affected by any of the State's historical discrimination; moreover, it was clear from their testimony that they had significant opportunity to update their registrations. Finally, as a capable individual accepted at one of the top public universities in the country, Ms. Cohen makes clear that, for some, given the myriad of options available in the modern age, failure to register and vote is more a reflection of motivation than ability. (See Doc. 331 at 173.) Historical discrimination is an unpersuasive basis for claiming that any of these people needed or wanted to use SDR. When proving discrimination, plaintiffs need not rely on statistical evidence alone. They may present witnesses who testify about "their personal experiences" in order to bring "the cold numbers convincingly to life." Int'l Bhd. of Teamsters v. United States, 431 U.S. 324, 339, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). In this case, however, the anecdotal evidence tended to support Defendants' statistical data analysis, rather than Plaintiffs'.

Finally, the trial evidence showed that the State's interest in removing SDR was legitimate and certainly not tenuous. The State legitimately aimed to employ an election system that provided meaningful opportunity for election officials to verify residency through the statutory process, and the repeal of SDR served this purpose. SDR's proximity to Election Day, inside the twenty-five day registration cut-off, simply did not provide a sufficient number of days for the mail verification process to work, and thus effectively nullified North Carolina's pre-existing statutory process for verifying a voter's residence. As a result, in 2012 at least 2,361 voters had their ballots counted although they later failed mail verification and were declared ineligible voters. That makes this case quite different from Husted, 768 F.3d 524, where the timing of the period for SDR provided Ohio election officials with at least thirty days to verify even those who registered and voted through SDR and, consequently, the State could show no difference between how traditional and SDR registrants would be verified on Election Day. Id. at 546–548. In addition, and again in contrast to Husted, the trial evidence in this case shows that SDR placed additional burdens on CBOE staff that made it "generally ... not possible to process the number of [SDR-registrants] within" the forty-eight hour period required by statute, despite "hir[ing] additional staff" and "staff work[ing] as efficiently as they could." (Pl. Ex. 56 at 5); Cf.Obama for America v. Husted, 697 F.3d 423, 433–34 (6th Cir.2012) (finding "no evidence that local boards of elections [had] struggled to cope with early voting in the past"). Just as Husted was decided in light of Ohio's local conditions, this case too must be decided based on North Carolina's specific circumstances, including its statutory scheme for verification of voter eligibility.

For these reasons, this court cannot conclude that the removal of SDR leaves North Carolina with an election system that provides an unequal opportunity for African Americans to register and vote compared to other groups. There was no showing that SDR increases turnout generally or for African Americans; only disproportionate use. Instead, other, larger influences were more substantially at play. Plaintiffs also did not show that African American turnout in 2014 would have been any higher had SDR been in place. As such, this court is left with the 2014 data that show that, without SDR, African Americans participated at a higher rate in 2014 than they did in the last comparable 2010 race with SDR in place. On this record, Plaintiffs have failed to prove that by returning North Carolina's registration system to that of 2007—which had never been claimed to discriminate against African Americans or any minority group, and which is more generous than that of many States—the General Assembly imposed a discriminatory burden on African Americans or any other group.

d. OOP Voting

For approximately 120 years, North Carolina has assigned each registered voter a precinct in which to cast a ballot. OOP voting changed that, however, permitting voters to cast a provisional ballot at any precinct in the voter's county of residence. Even when OOP voting was available, a voter could cast a full ballot only in his correct precinct on Election Day. If an OOP provisional ballot was used, only votes for races that would have appeared on the ballot in the voter's correct precinct were counted. This had the practical effect of disenfranchising OOP voters on some down-ballot races. After SL 2013–381, voters who do not fall within the unreported mover exception are only permitted to vote at their assigned precinct on Election Day.

Adoption of OOP was politically motivated to the extent the Democratically-controlled legislature made OOP retroactive to legislatively overrule the North Carolina Supreme Court's decision in James and to secure electoral victory for Democratic candidates in the 2004 election. In doing so, proponents inserted into the amendment a statement of retroactive legislative intent to aid African Americans. Yet, there was never a claim that North Carolina was violating § 2 of the VRA or any other law prior to OOP's enactment. In fact, the majority of States do not offer it.

Plaintiffs demonstrated that African Americans disproportionately used OOP voting when it was in place.²⁰¹ In addition, African Americans cast a disproportionate share of the provisional ballots cast in the wrong precinct in 2014. Given that African Americans have less access to reliable transportation, Plaintiffs claim they are disproportionately burdened by being required to appear at their assigned precinct. For this reason, Plaintiffs claim disproportionate use reflects need.

201 Plaintiffs also showed that Hispanics disproportionately used OOP when it was in place. (Pl. Ex. 245 at 24.) But, as noted above, Plaintiffs did not make a showing of official discrimination against Hispanics. In addition, Hispanics are more likely to have access to a vehicle than African Americans, (Pl. Ex. 45 at 13–14), and Plaintiffs do not claim that OOP's removal imposes a heavier burden on Hispanics than African Americans.

Plaintiffs, however, failed to show that voters' assigned precincts are not on average the closest precinct to their residence or work. Instead, Plaintiffs presented evidence on the average distance between the precinct at which OOP voters attempted to cast a provisional ballot and those voters' assigned precinct. (Pl. Ex. 241 at 43.) This study showed that in Mecklenburg and Wake Counties in 2014, this distance was approximately six miles. (Id.) Defendants rebutted this study with evidence that a larger percentage of African Americans cast a provisional ballot within five miles of their assigned precinct than whites. (Def. Ex. 212A at 18–19 & tbl. 23.) In any event, as recognized in James, a benefit of the precinct-based system is that it generally "puts polling places in closer proximity to voter residences," 359 N.C. at 271, 607 S.E.2d at 644–45, and Plaintiffs have failed to show that this is not the case. That is, there was no evidence of the distance from voters' home or work to their assigned precinct versus the one in which they voted.

To be sure, OOP voting was not significant to the parity in political participation achieved by African Americans since 2008. In 2008, African American turnout exceeded white turnout by 5.5%; in 2010, white turnout exceeded African American turnout by 3.4%; and in 2012, African American turnout exceeded white turnout by 6.8%. (Pl. Ex. 242 at 161 (App'x U).) By the court's estimate, if OOP ballots had not been counted in those years, African American turnout would have exceeded white turnout by 5.4% in 2008; white turnout would have exceeded African American turnout by 3.6% in 2010; and African American turnout would have exceeded white turnout by 6.8% in 2012.²⁰² In other words, not having OOP would only have changed the turnout differential by .1% in 2008 .2% in 2010, and less than .1% in 2012. (See Pl. Ex. 42 at 98 (tbl. 14) & n.126; Pl. Ex. 242 at 161 (App'x U).) In fact, OOP provisional ballots have never constituted more than a fraction of a percentage point of votes cast for either African Americans or whites. (See id.)

202 To obtain these estimates, the court subtracted the number of OOP provisional ballots cast by race and year, (Pl. Ex. 42 at 98 (tbl. 14)), from the number of voters by race and year, (Pl. Ex. 242 at 161 (App'x U)). The court then divided by the VAP of each race in each applicable year (Id.) Because Plaintiffs' data on the number of OOP provisional ballots excluded 35.4% of the records in the provisional ballot file (the race of the voter was not indicated), the court followed Plaintiffs' instruction and multiplied each number in Table 14 by 1.55. (Pl. Ex. 42 at 98 n.126 ("[T]he proper correction to apply is to multiply each number by 1/.646, or 1.55.").) This also assumes, to Plaintiffs' benefit, that OOP voters would not have been able to make it to their correct precinct and vote. The evidence demonstrated that most OOP voters are much more capable than the court has assumed in its calculation. (SeeDef. Ex. 343.)

In League, the Fourth Circuit stated that "what matters for purposes of Section 2 is not how many minority voters are being denied equal electoral opportunities but simply that 'any' minority voter is being denied equal electoral opportunities." 769 F.3d at 244. In that regard, a voting law cannot be countenanced merely because it affects relatively few people. But the court notes the figures here as to OOP only because courts have found the number of voters affected relevant to each prong of the § 2 analysis. For example, the number of voters affected is some evidence of the magnitude of the burden imposed. SeeFrank, 768 F.3d at 748–49 (finding that new voting requirement was not an obstacle to "a significant number of persons who would otherwise cast ballots"); Daniel P. Tokaji, Applying Section 2 to the New Vote Denial, 50 Harv. C.R.–C.L. L. Rev. 439, 487 (2015) ("A voting restriction affecting more people imposes a greater burden (all other things being equal) than one which affects fewer people."); cf.Florida, 885 F.Supp.2d at 340 ("[W]hile the number of affected voters will have an impact on the burden analysis ... a voting change with a retrogressive effect does not warrant preclearance merely because it affects a small number of voters."). In addition, the number of people affected is relevant to whether any burden imposed can be linked to social and historical conditions. For example, the fact that a much larger number of African Americans endure socioeconomic disparities than the few who utilized OOP suggests that something other than socioeconomic disparities is causing those voters to utilize OOP.²⁰³ In other words, the number of voters affected is relevant to whether anyone is being denied an equality of opportunity on account of a law's interaction with social and historical conditions. Cf.Frank, 768 F.3d at 748–49.

203 For example, if a group with the same or similar characteristic attempts to clear an alleged hurdle, and if less than a percent of the group fails to do so, the following inferences are supported: (1) the hurdle is not very difficult to clear for individuals with the shared characteristic, and (2) the shared characteristic does not readily explain why certain individuals cleared the hurdle and others did not.

Although just one factor under the totality of the circumstances, the data suggest that African Americans have an equal opportunity to participate in the electoral process without OOP. In addition, the relatively small number of individuals who used OOP have many remaining convenient alternatives: voting during any of the ten days of early voting where they need not vote at their assigned precinct (they can vote at any early-voting center, the number of which SL 2013–381 increased), voting at their assigned precinct on Election Day, or casting an absentee ballot by mail during the forty-five to sixty days available (depending on the election). In addition, a primary benefit of the precinct-based system, as recognized in James, is that it generally places voting locations closer to voters—a fact that Plaintiffs failed to demonstrate was not true.

Beyond statistical data, Plaintiffs presented the testimony of several voters who voted in the wrong precinct in 2014. As with SDR, these witnesses did not provide support for Plaintiffs' claims.

Gwendolyn Farrington, a middle-aged African American woman, voted a provisional ballot in the wrong precinct. She did not vote in the early-voting period, citing fatigue from her jobs, where she works six days a week. (Doc. 329 at 59.) Yet, she chose not to vote on her day off, a Sunday, because she wanted to spend it resting. (Id. 58–59.) She chose not to vote by mail because she prefers to vote in person. (Id. at 55.) On Election Day 2014, instead of going to her assigned polling location, which she knew, she went to the polling location closest to her work because she did not think she could have made it to her assigned polling location because she chose to provide transportation to her adult children instead. (Id. at 59–60, 66–67.) In addition, her assigned precinct was not assigned based on her current address, but was linked to her parents' address where she had lived previously and which she had listed on her registration some four years earlier. (Id. at 66.) Even though she had been living at her 2014 address for three years, as of Election Day 2014, she still had not updated her voter registration to reflect that address. (Id.) It is difficult to say that any burden Ms. Farrington felt is attributable to historical discrimination.

Terrilin Cunningham moved to North Carolina in 2012; she was born and raised in Missouri, and then had lived in both Baltimore and Pittsburgh before moving to North Carolina. (Doc. 330 at 164.) When Ms. Cunningham lived in Missouri and Pennsylvania, she had to vote in her assigned precinct on Election Day, as those States do not offer OOP or early voting. (Id. at 182–83.) In 2012, she voted early on a Sunday in North Carolina; she was surprised that North Carolina offered Sunday voting because she had never heard of such a thing before. (Id. at 169–71.) In 2014, Ms. Cunningham had intended to vote early again, but she missed her chance because of work and various unexpected doctors' appointments.²⁰⁴ (Id. at 172–73.) Believing that she had to vote in her county but that the precinct system would not matter for Election Day (just as it did not matter for early voting in her 2012 experience), she tried to vote in the wrong precinct on Election Day 2014. (Id. at 176.) She ended up casting a provisional ballot that was not counted. It was clear from her testimony that Ms. Cunningham is sufficiently capable of determining her correct precinct, as her previous States of residence had required. It was further plain that there is no evidence that Ms. Cunningham's failure to vote successfully had any connection with any effect of historical discrimination by the State of North Carolina or anyone else.

204 Plaintiffs seem to suggest that Ms. Cunningham could have voted early if the early-voting period had been longer, (Doc. 346 at 83), but there is no evidence that a longer period would have helped.

Lue Alice Abercrombie, a disabled African American woman, moved to North Carolina from Connecticut twenty years ago.²⁰⁵ She had never voted before the November 2014 election because she did not know where to go. (Pl. Ex. 712 at 10.) Ms. Abercrombie registered to vote at the NC DMV when she got her driver's license. (Id. at 17.) On Election Day 2014, she went to vote at the same precinct at which her fiancée was going to vote, though it was not her assigned precinct and she had never voted there before. (Id. at 11.) She learned that she was at the wrong precinct and that her correct precinct would be closed by the time she could get there. (Id. at 12.) She ultimately cast a provisional ballot that was not counted. Ms. Abercrombie has now learned where her correct precinct is and says she can get there for future elections. (Id. at 18.)

205 Consequently, she was not a discernable victim of North Carolina's historical discrimination.

Michael Owens, an African American man in his fifties, was unable to vote on Election Day 2014. (Doc. 334 at 148, 160, 153–56.) For a temporary period including Election Day, he was without a vehicle. (Id. at 151–52.) Mr. Owens' former employer had closed down and, because he lost employment, his vehicle was repossessed in May 2014 when he ran out of savings. (Id. at 149–51.) Later, Mr. Owens found a new job in Lumberton (in Robeson County). (Id. at 150–51.) He lived and worked in Lumberton during the workweek, but returned to Shannon (also in Robeson County) on the weekends to be with his girlfriend and child. (Id. at 151–52.) On Election Day 2014, when he tried to vote at a precinct in Lumberton, he learned that he was in the wrong precinct; he had no way of making it to his correct precinct in Shannon that day. (Id. at 153–56.) Since Mr. Owens was temporarily living outside of his precinct, he could have avoided voting problems in the 2014 general election by voting early or voting by absentee ballot. Mr. Owens was familiar with early voting, since he had voted early in the past. (Id. at 159.) Later in November 2014, Mr. Owens was able to acquire a car. (Id. at 158.) He now lives in Shannon and commutes to Lumberton for work. (Id. at 158–59.) He now knows that, if he votes on Election Day in the future, he will need to vote in his correct precinct. (Id. at 159–60.)

Plaintiffs only presented two witness who claimed their correct precinct was further from their residence than the precinct in which they cast provisional ballots in 2014, Timothy Washington and his wife, Yvonne Washington. The couple's correct precinct was 1.3 miles from their home, while the incorrect precinct at which they cast a provisional ballot in 2014 was 0.6 miles from their home. (Pl. Ex. 797 at 43–44.) Both have disabilities that make it difficult for them to walk and rely on others for transportation. (Pl. Ex. 679 at 39–40.) Ms. Washington is a cancer survivor with asthma and Mr. Washington has suffered from hip disabilities since birth that require him to use crutches. (Pl. Ex. 797 at 8, 21.) Ms. Washington testified that she will never be able to walk to her correct precinct. (Pl. Ex. 679 at 26.) While OOP would make their burden less, the truth is that even walking to the nearest polling location on Election Day in 2014 (still 0.6 miles) was itself a great difficulty for the couple. (Pl. Ex. 797 at 21–22.) In fact, they were not able to make it to the closest precinct without stopping to rest. (Id.) Their situation most persuasively demonstrates a need for a voting mechanism than can be done from home, such as absentee mail voting. If, however, they find mail voting to be inaccessible or unattractive, Ms. Washington testified that, although she did not seek it out in 2014, her church would likely be able to provide the couple with a ride to their correct polling place. (Pl. Ex. 679 at 29–30.) In short, walking to any precinct or polling location does not appear to be a long-term solution for the couple. In addition, the link between their difficulties and social and historical conditions was weak. Mr. Washington's hip disability arose from when his mother "fell down the steps when he was born, and she was carrying him." (Id. at 8.) Moreover, Ms. Washington only moved to North Carolina five years ago from Virginia, where she voted five different times, and always in her correct precinct, since Virginia does not permit OOP voting. (Id. at 33, 36–37.)

For these reasons, having considered the totality of the circumstances and in light of the State's demonstrated substantial interest in its precinct-based system as articulated by the Supreme Court in James, this court cannot conclude that North Carolina's current electoral system without OOP presents African Americans with an unequal opportunity to vote as compared to other voters.

e. Pre-registration

From 2009 to 2013, sixteen-year-olds were permitted to preregister. Even after SL 2013–381, seventeen-year-olds who will be eighteen by the time of the general election are still able to register starting sixty days before the related primary. N.C. Gen. Stat. § 163–59. Plaintiffs argue that the elimination of pre-registration disparately impacts African American and Hispanic youth and imposes a severe burden on all youth. (Doc. 346 at 95–98.)

Plaintiffs demonstrated that pre-registration increases youth turnout. In addition, they demonstrated that African Americans disproportionately used pre-registration.²⁰⁶ Plaintiffs, however, did not link this disproportionate use to social and historical conditions. In fact, the literature addressing pre-registration hypothesizes that "young people who are especially interested in politics might be both more likely to pre-register and more likely to vote." (Pl. Ex. 235 at 17.)

206 Plaintiffs presented some evidence that Hispanic youth disproportionately used pre-registration, (Pl. Ex. 245 at 23), but this court's conclusions as to the race-neutral effect of pre-registration applies equally to all minority groups.

In addition, Plaintiffs' own evidence demonstrated that the benefits of pre-registration are race-neutral. In fact, the studies of Plaintiffs' own expert, Dr. Hillygus, found pre-registration to be "equally effective for various demographic groups, including white versus minorities." (Id. at 24.) It may be tempting to reason that if pre-registration increases turnout and African Americans disproportionately used it, then African Americans disproportionately benefited from it. However, Dr. Hillygus's own study rebuts this inference. Dr. Hillygus did not just study whether those who pre-register are more likely to vote.²⁰⁷ Instead, her study included both those who pre-registered and those who could have pre-registered in the context of the 2008 presidential election. (Id. at 22 ("[S]ome of those eligible to preregister waited until they were older to register traditionally ....").) This is why she speaks of her studied group as "those who had the opportunity to preregister in 2008." (Id.(emphasis added).) In this sense, it appears that her race-neutral conclusion exists independent of whether one race pre-registered more. This is because, according to her, there are certain benefits that flow from just being exposed to pre-registration during an election period. (See id. at 22–23 (fig. 7) ("Exposing Youth to Preregistration Programs Increases Turnout."); Doc. 331 at 197–98.) Accordingly, even though African Americans disproportionately used pre-registration, Dr. Hillygus's study suggests that the effect on youth turnout will nevertheless be race-neutral. (See Pl. Ex. 235 at 24; Doc. 331 at 199–200, 205 ("There [are racial] differences in the use that we saw in North Carolina, but not in the mobilization effect.").) In fact, although Dr. Hillygus made many predictions about the effect of pre-registration in North Carolina, she never claimed that it disproportionately benefits African Americans. (See Pl. Ex. 235 at 16; Doc. 331 at 199–200, 205.)

207 Dr. Hillygus's concern with studying just those who pre-register was, as noted above, that pre-registrants may have characteristics that make them both more likely to pre-register and more likely to vote. (Pl. Ex. 235 at 17 ("Put differently, those who preregistered and voted might well have done so even if pre-registration had not been in place.").)

Finally, the evidence shows that pre-registration did not clearly benefit either Democrats or Republicans. (Pl. Ex. 235 at 14.) In some years, Republicans had more pre-registrants than Democrats, and in other years the reverse was true. (Id.) In all years, however, the number of unaffiliated pre-registrants was greatest. (Id.)

Accordingly, having considered the totality of the circumstances, Plaintiffs have failed to establish that the removal of pre-registration imposes a discriminatory burden on African Americans or any other racial group.

f. Cumulative Effect

Finally, the court considers the cumulative impact of the challenged provisions,²⁰⁸ to the extent the evidence in the case permits such consideration.²⁰⁹League, 769 F.3d at 242. Unlike their isolated analyses of each provision, Plaintiffs in the section of their proposed findings of fact devoted to this topic have provided little guidance and no empirical analysis of the cumulative impact of the challenged provisions. (See Doc. 346 at 102–03; Doc. 419 at 55.) The court is thus left largely with its own assessment.

208 Plaintiffs also challenge the removal of CBOE discretion to extend voting for an extra hour on Election Day and the poll observers and challengers provisions on § 2 grounds. However, they failed to present evidence of any discriminatory burden as to either. Plaintiffs provided no evidence that African Americans were disproportionately likely to vote in the discretionary last hour of voting. In any case, the evidence did show that the SBOE still has discretion to extend voting where there is a loss of voting time. Nor did Plaintiffs provide any evidence that observers or challengers abused their statutory authority so as to disadvantage African Americans in 2014, or would do so in the future.

209 Where a new law involves repealof prior provisions, the § 2 impact inquiry as to the new law by definition involves consideration of the cumulative impact.

Plaintiffs argue that voters who "manage to register" with the elimination of SDR "may now find themselves excluded because of the photo ID requirement." (Doc. 419 at 55.) This ignores the reasonable impediment exception, as it will be applied. For all the reasons noted above, the voter-ID requirement, with a two-year roll out, free ID, and the reasonable impediment exception, does not impose a material burden on minorities' right to vote. There is no evidence that the current photo-ID requirement will depress turnout or interact with other challenged provisions. Indeed, the United States abandoned this claim.

In addition, the Supreme Court's stay of the re-instatement of SDR and OOP voting presents this court with a unique set of data not ordinarily available in such cases. Because the 2014 elections were conducted under SL 2013–381, the 2014 data permit the court to see the actual—not merely estimated—cumulative effect of the removal of all of the challenged provisions under current election law. As detailed above, those data show that minority North Carolinians not only did not backslide under the new law, but rather continued to increase their participation. African Americans continue to hold a commanding lead over whites in registration rates, and their turnout rates continue to increase over comparable elections. This is powerful evidence of the cumulative effect of the law, seePurcell, 549 U.S. at 6, 127 S.Ct. 5 (Stevens, J., concurring) (preferring "historical facts rather than speculation"), and is persuasive proof that African Americans do not suffer an inequality of opportunity under SL-2013–381.

The early-voting schedule provides in substance an equivalent to, if not an actual improvement over, the former schedule. North Carolina voters still have eleven days to vote in person: ten days of early voting, at more convenient hours, and Election Day. They may also vote by absentee ballot from forty-five to up to sixty days before an election (depending upon the election).

Even without SDR, African Americans have substantial opportunities to register to vote, as demonstrated by their growing lead over whites in registration rates.²¹⁰ To the extent GOTV efforts drove SDR results, such organizations have all year to register voters and merely need to adjust their resources over the balance of the year. The 2014 results are evidence that they have done so.As for OOP voters, the evidence did not show that the elimination of OOP provisional voting interacts significantly with other challenged mechanisms of voting. SDR and OOP serve different groups of people. If an individual misses the registration cutoff, OOP will not help them, as one must be registered to cast an OOP ballot. Similarly, given that OOP voters are already registered and thus complied with the twenty-five day cut-off, SDR provides no benefit to them. In short, SDR's removal will not create OOP voters, and OOP's removal will not create SDR voters.

210 Voters who move inter-county within twenty-five days of Election Day were never able to use SDR to vote in their new county because they would not have satisfied the State's eligibility requirement that they be a resident of the county for thirty days prior to the election. N.C. Gen. Stat. § 163–55(a).

Further, the evidence showed that the vast majority of the limited number of individuals who voted OOP did so not because they found it difficult to use any of the many other voting options available to them, but either because it was easier(an extra convenience) or because they were inattentive to their assigned precinct (which appears on their voter registration card). Under current law, former OOP voters may cast ballots in their correct precinct on Election Day, outside their assigned precinct at any early-voting site during any of the ten days of early voting, from home using a mail-in absentee ballot, and at their old or new precinct on Election Day if they are an unreported mover within the county.

Finally, those who will be eighteen-years-old by Election Day may still register when seventeen-years-old. North Carolina even permits seventeen-year-olds to register sixty days in advance of, and vote in, the primary for that general election. The registration may be accomplished in any of the myriad of ways noted above, and forms remain available at all public high schools, public libraries, CBOEs, and on-line, in addition to many other places.

As a result, the court finds that the cumulative effect of the provisions of SL-2013–381 and SL 2015–103 do not impose a discriminatory burden on African Americans or any other group.

4. Discriminatory Result: Conclusion

Having applied the two-element test from the Fourth Circuit and considered the totality of the circumstances through an "intensely local" analysis, the court finds that there has been no violation of § 2 of the VRA. This court has been careful not to require Plaintiffs to show that African Americans "cannot register or vote under any circumstance" or that voting mechanisms are "practically unavailable" in order to establish a § 2 violation. League, 769 F.3d at 243. Instead, it has considered the challenged provisions separately and cumulatively, taking a functional view of the political process. But based on all these considerations, Plaintiffs have not established that, under the electoral system established by SL 2013–381 and SL 2015–103, African Americans or Hispanics "have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice." 52 U.S.C. § 10301(b).

Rather than establish a § 2 violation, the evidence at this court's January trial established that North Carolina's voter-ID law with a reasonable impediment exception will not impose a material burden on minority voters or deprive them of an equal opportunity to participate in the political process. In addition, the evidence from this court's July trial established that the many convenient registration and voting mechanisms that remain under SL 2013–381 provide African Americans an equal opportunity to participate in the political process. As stated above, there are simply very many easy ways for North Carolinians to register and vote under SL 2013–381. In truth, North Carolina's registration and voting options were quite generous prior to SL 2013–381—Plaintiffs even referred to them at trial as "progressive." But they nevertheless remain more than adequate to provide an equality of opportunity to all groups of interested citizens after SL 2013–381.

That said, a State is not entitled to "make changes for the purpose of curtailing black voting" simply because "blacks register and vote more frequently than whites." Frank, 768 F.3d at 753–54. Here, however, the 2014 data have provided strong evidence that African Americans have an equal opportunity under the new law. See id. At the preliminary injunction stage, this court and the Fourth Circuit could only project how African Americans would fare under SL 2013–381 based on their prior use of the removed mechanisms. The 2014 data permit this court to stop speculating and observe real life results. African Americans fared better in terms of registration and turnout rates in 2014, after the challenged mechanisms were eliminated, than in 2010, when they were in place. In fact, not only did African American turnout increase at a greater rate than other groups' turnout in 2014 with SL 2013–381 in place, but that general election saw the smallest white-African American turnout disparity in any midterm election from 2002 to 2014. (Def. Ex. 309 at 62; Pl. Ex. 229 at 7.) This is not dispositive, but it does seriously undermine Plaintiffs' contention that African Americans lack an equal opportunity to participate in the political process without the removed mechanisms. While the educational and socioeconomic disparities suffered by African Americans might suggest that the removed mechanisms would disproportionately benefit African Americans, this court cannot find an inequality of opportunity simply because educational and socioeconomic disparities suggests one might exist—there must actually be an inequality of opportunity. LULAC, 999 F.2d at 867 ("[T]he Senate Report, while not insisting upon a causal nexus between socioeconomic status and depressed participation, clearly did not dispense with proof that participation in the political process is in fact depressed among minority citizens."). The evidence shows that, at least as to North Carolina's current election system, African Americans are equally as capable as all other voters of adjusting to SL 2013–381.

Moreover, the data from when the removed mechanisms were in place further bolster the reliability of the 2014 results. As shown above, Plaintiffs were unable to sustain their claim that the removed mechanisms were responsible for the recent parity (much less lead) in political participation achieved by African Americans. Instead, there was strong evidence that other factors were more substantially at play, such as President Obama's candidacy and North Carolina's emergence as a battleground State. In short, the data from when the removed mechanisms were in place suggest that African Americans did not need the mechanisms to have an equal opportunity, and the 2014 data confirm as much.

Further, this court accepts that many variables feed into turnout and registration data. However, despite having the ability to do so, Plaintiffs did not perform the analysis necessary to separate out the causal effect of various factors. Plaintiffs did not do this analysis for the years leading up to SL 2013–381, nor did they provide this court with analysis of what the 2014 data would have been with the removed mechanisms in place. This decision may have been strategic, as the available evidence suggests that the removed mechanisms were not responsible for the parity in political participation achieved by African Americans. Nevertheless, Plaintiffs bear the burden of proof in this case, and Defendants cannot be blamed for this absence.Finally, this court has considered the evidence with an understanding that presidential elections are different from midterms. But, even having reviewed all of the evidence with that caveat in mind, the court is not persuaded that Plaintiffs' 2016 predictions will come to pass, either.

For these reasons, Plaintiffs have failed to show that SL 2013–381 provides African Americans with "less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice," 52 U.S.C. § 10301(b), and thus their § 2 claim must fail.

5. Discriminatory Intent

Plaintiffs argue that they can prove a § 2 claim through either the Gingles factors, traditional discriminatory intent law, or both. (Doc. 346 at 130–31.) The court has already determined that the Gingles factors as a whole do not favor the Plaintiffs. The court's conclusion regarding those factors would be similar here in the discriminatory intent context, even when considered in combination with the traditional intent principles. Therefore, the court turns to the traditional law of discriminatory intent, developed primarily in the equal-protection context.

Discriminatory purpose "implies more than intent as volition or intent as awareness of consequences. It implies that the decisionmaker, in this case a state legislature, selected or reaffirmed a particular course of action at least in part 'because of,' not merely 'in spite of,' its adverse effects upon an identifiable group." Pers. Adm'r of Massachusetts v. Feeney, 442 U.S. 256, 279, 99 S.Ct. 2282, 60 L.Ed.2d 870 (1979) (citations and footnote omitted); see alsoVeasey, 796 F.3d at 498–99 ("The appropriate inquiry is not whether legislators were aware of [a law's] racially discriminatory effect, but whether the law was passed because of that disparate impact. Importantly, although discriminatory effect is a relevant consideration, knowledge of a potential impact is not the same as intending such an impact." (citations and footnote omitted)). If "racial discrimination is shown to have been a 'substantial' or 'motivating' factor behind enactment of the law, the burden shifts to the law's defenders to demonstrate that the law would have been enacted without this factor." Hunter v. Underwood, 471 U.S. 222, 228, 105 S.Ct. 1916, 85 L.Ed.2d 222 (1985). Thus,

there is an element of causation that is a necessary part of plaintiff's showing, especially when plaintiff is trying to uncover the motivation of a multi-member decisionmaking body, such as a zoning board. While a plaintiff is not required to prove that "the challenged action rested solely on racially discriminatory purposes," proof that racial discrimination was a motivating factor would not end the matter. Such proof merely shifts the burden to the decisionmaking body to demonstrate that "the same decision would have resulted even had the impermissible purpose not been considered."

Sylvia Dev. Corp v. Calvert Cty., 48 F.3d 810, 819 n. 2 (4th Cir.1995) (citations omitted).

According to the Supreme Court, "Determining whether invidious discriminatory purpose was a motivating factor demands a sensitive inquiry into such circumstantial and direct evidence of intent as may be available." Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 266, 97 S.Ct. 555, 50 L.Ed.2d 450 (1977). In making such an inquiry, courts look to a non-exhaustive list of factors from Arlington Heights.

The first factor to be considered is whether the "impact of the official action ... 'bears more heavily on one race than another.' " Arlington Heights, 429 U.S. at 266, 97 S.Ct. 555 (quoting Washington v. Davis, 426 U.S. 229, 242, 96 S.Ct. 2040, 48 L.Ed.2d 597 (1976) ).

With regard to the removed voting mechanisms, Plaintiffs point to their strongest fact: African Americans disproportionately used them. But, for the reasons explained above, this does not mean that the impact of current law bears more heavily on them. Whether a change bears more heavily depends on the options remaining. Florida, 885 F.Supp.2d at 383 (three judge panel finding that changes to inter-county mover law "will not 'bear more heavily' on any minority group" because "the evidence indicates that the changes will not have materially adverse effects on the ability of minority voters to cast a ballot and effectively exercise the electoral franchise"); Brown, 895 F.Supp.2d at 1246 (finding that "the evidence before the court does not demonstrate that the changes will deny minorities equal access to the polls"). As stated above, the evidence demonstrated that North Carolina's remaining mechanisms continue to provide African Americans with an equal opportunity to participate in the political process.

Nevertheless, Plaintiffs claim the legislature was aware of certain disparities in the use of the removed mechanisms and that this evidences discriminatory intent. Specifically, Plaintiffs claim that the General Assembly was aware that African Americans disproportionately used early voting, SDR, and OOP and lacked a qualifying photo ID.²¹¹ There is no evidence that the legislature had demographic data on the use of pre-registration. (Doc. 346 at 59 (only claiming that legislators knew that African Americans disproportionately used early voting, SDR, and provisional ballots).) There was, however, evidence that certain members of the legislature requested and received demographic data for "provisional and one-stop voters." (Pl. Ex. 72 at 3; Pl. Ex. 385; Pl. Ex. 459; Pl. Ex. 436; Pl. Ex. 437.)

211 In their trial briefs, Plaintiffs limited their "knowledge" claim to early voting and SDR. (Pl. Ex. 285 at 47–48; Doc. 286 at 47–50.) However, in their proposed findings of fact and conclusions of law, Plaintiffs claim that "[l]egislators knew that African Americans disproportionately relied on early voting, SDR, and provisionalballots." (Doc. 346 at 59 (emphasis added).)

Plaintiffs' exhibits demonstrate that in January 2012, legislative research staffer, Erika Churchill, requested from the SBOE "a breakdown of the 2008 voter turnout, by race (white and black) and type of vote (early and election day)." (Pl. Ex. 437 at 3.) The SBOE appears to have provided the requested data to Ms. Churchill at the end of January. (Id. at 1–2.) In February 2012, she requested the same data for 2010. (Id. at 1.) She also requested "similar information ... on provisional ballots [for] 2008 and 2010." (Id.) Plaintiffs' exhibits do not clearly indicate whether the SBOE ever provided the data from Ms. Churchill's February requests. (See Pl. Ex. 436; Pl. Ex. 437; Pl. Ex. 459.) Instead, the last email in the email chain provided by Plaintiffs is from Ms. Churchill to the SBOE clarifying her request on provisional ballot data. (Pl. Ex. 436.) Further, even though Plaintiffs' exhibits indicate that the SBOE provided the information from the January request, the exhibits cited in Plaintiffs proposed findings of fact and conclusions of law do not provide this data. (See Doc. 346 at 59.) Therefore, among the thousands of pages of material the parties have submitted to the court, it is difficult to evaluate the character of any data that may have been received and transmitted to legislators as a result of Ms. Churchill's requests.

Plaintiffs' exhibits also indicate that on March 5, 2013, Representative Harry Warren and various other sponsors of HB 589 requested a cross-matching of registered voters who have "neither a NC Driver's License nor a NC Identification Card" and the "number of one-stop voters and provisional voters." (Pl. Ex. 72 at 3–4.) Representative Warren requested that both inquiries be broken down by all possible demographics that SBOE captures, including "party affiliation, ethnicity, age, gender, etc." (Id.) He further requested information on the capability of CBOEs to capture a digital photo of registrants and/or to produce any type of identification cards. (Id.) Later that day, SBOE Director Gary Bartlett provided a link where "[t]he data requested c[ould] be found ... [including] tabs that show[ed] summary counts based on different demographics." (Id. at 2–3.) This link is no longer active. Veronica Degraffenreid later sent Representative Warren some "categorical statistics" that Director Bartlett had requested she provide to Representative Warren. (Id. at 2.) Representative Warren further requested that the categorical data be broken down by county, which Ms. Degraffenreid provided. (Id.)

While this court accepts that Ms. Churchill and Representative Warren requested demographic data on ID possession, "one-stop voters," and "provisional voters," these requests are not necessarily as suspect as Plaintiffs claim. First, at the time of Representative Warren's request on March 5, 2013, legislators would have been preparing for the first public hearing on voter ID on March 12, 2013. (See Pl. Ex. 127.) As noted herein, opponents frequently challenge voter-ID bills on the basis of racial disparities in ID possession. Any responsible legislator would need to know the disparities in order to account for such challenges. In fact, during the preliminary injunction stage of this case, the United States would not tell this court whether it would have been better or worse for the State not to have requested demographic data. (Doc. 166 at 219–20.) Second, given that North Carolina was subject to preclearance under § 5 when the demographic data requests were made, legislators would have needed to know the racial impact of the voting changes in order to evaluate whether they were even feasible. In other words, when § 5 applied to North Carolina, evaluating racial impact was a prerequisite to evaluating the likelihood that any voting change would be precleared by the Attorney General. Accordingly, while Plaintiffs seek the inference that legislators requested demographic information because they sought to discriminate against African Americans, alternative explanations are considerably more persuasive.

Next, Plaintiffs presented evidence that Director Strach emailed some data to Representative Lewis, one of the bill's House sponsors, on July 25, the day of the House concurrence vote. (Pl. Ex. 198.) This data primarily consisted of the verification rates for SDR in the 2010 and 2012 election and information about the types of IDs presented by same-day registrants. (Id. at 3–20.) It also included a spreadsheet that contained race data for individual same-day registrants and whether those registrants were verified. (See id. at 14, 16.) The report did not provide aggregate percentages for SDR use by race. (See id.) In addition, given that the report was not provided until the day of the House concurrence vote, it is not possible that any disparities that could be inferred from the individual voter data provided by Ms. Strach were used in drafting HB 589.

Next, Senator Stein provided evidence of disproportionate use during Senate debate of HB 589. Specifically, Senator Stein stated in debate that "[m]inorities take advantage ... of same day registration ... more than the general population." (Pl. Ex. 550 at 34–35.) He also shared graphs indicating that 34% of the nearly 100,000 individuals who used SDR in 2012 were African American.²¹² (See Pl. Ex. 18, Ex. A at 6.) Senator Stein provided similar evidence on early voting and stated in debate that minorities disproportionately used the removed seven days of early voting. (Pl. Ex. 550 at 34; Doc. 335 at 185.) Senator Stein did not provide any disparate use evidence for OOP or pre-registration. (Pl. Ex. 550 at 34–35.) Given that HB 589 had already been drafted, the evidence that Senator Stein presented in debate is more probative of the fact that the legislature enacted HB 589 despite the disparities outlined, rather than because of them.

212 In the debate, Senator Stein stated the 100,000 voter figure, (Pl. Ex. 550 at 28), but he did not openly state the 34% figure, (seePl. Ex. 202; Pl. Ex. 549; Pl. 550). That figure was within one of Senator Stein's charts, which he evidently made available to other Senators through their dashboards. (Doc. 335 at 185–86.)

Finally, Plaintiffs argue that the legislature must have been aware of OOP's disproportionate use given that the legislature that enacted OOP made the finding that "of those registered voters who happened to vote provisional ballots outside their resident precincts on the day of the November 2004 General Election, a disproportionately high percentage were African American." 2005 N.C. Sess. Law 2, § 1. While it can be assumed that the General Assembly was aware of its prior findings, it does not follow that any future decision to reverse course evidences racial motivation, especially given the substantial interests served by a precinct-based system endorsed by the Supreme Court in James.

Plaintiffs also argue that the foreseeable effect of the photo-ID requirement of SL 2013–381 is further evidence of a discriminatory intent that infects the other challenged provisions.²¹³ Specifically, an April 2013 SBOE report found that, of those who could not be matched to having a DMV-issued photo ID based on the DMV's data, 33.8% were African American (compared to their approximately 22% share of the population) and 54.2% were white.²¹⁴(Pl. Ex. 534 at 9.) Members of the legislature appear to have made multiple requests for demographic data on ID possession, including the March 5, 2013, request by Representative Warren.

213 Plaintiffs point to the legislature's decision to entrust the provision of no-fee ID to the DMV, about which they say there was "widespread knowledge of dysfunction" at the time of SL 2013–381's enactment. (Doc. 419 at 70.) As noted above, sixteen counties do not have a brick and mortar DMV licensing location. These counties are serviced, if at all, by mobile units. During the debate of SL 2013–381, Representative Paul Tine stated on the House floor that two of the counties he serves, Washington and Hyde, receive DMV mobile unit service one or two days a month. (Pl. Ex. 138 at 68.) He further added that the availability of DMV services was "actually lower" at that time, due to mobile units being broken down. (Id.at 68–69.) At trial, Commissioner Kelly acknowledged that when he was hired in October 2013, the DMV's computer system was outdated, DMV was unpopular with customers, wait times were too long, continuing education of examiners was insufficient, and the mobile unit fleet was either outdated or in disrepair. (Doc. 410 at 154–155, 195–198.) There was no testimony before the legislature that the DMV was dysfunctional in issuing IDs or that this would contribute to the burden imposed upon voters. Representative Tine's critique was limited to the accessibility of DMV locations in certain communities. (SeePl. Ex. 138 at 68–69.) But even if legislators knew of problems at the DMV, either through personal experience or common perception, the General Assembly's decision to place no-fee ID in the hands of the DMV is not substantial evidence of discriminatory intent. First, if legislators were looking for an agency with experience in issuing photo IDs, the DMV was an obvious choice. Second, the two-year soft roll out substantially undermines Plaintiffs' claim that the General Assembly intended to disadvantage voters by forcing them to interact with the DMV.

214 Plaintiffs also rely on Dr. Stewart's first no-match list as additional evidence of the foreseeable impact of SL 2013–381. There, Dr. Stewart was not able to match 397,971 (6.1%) registered voters to an acceptable ID, 147,111 (10.1%) African Americans and 212,656 (4.6%) whites. (Pl. Ex. 242 at 38 (tbl. 7).) As noted above, Defendants dispute the validity of this analysis. It is without dispute that this analysis did not include "refined" matching criteria that Dr. Stewart now considers necessary. (Pl. Ex. 891 at 19 (tbl. 11).) In any case, this court finds the April 2013 SBOE report to be the most probative matching analysis for intent purposes (despite questioning the reliability of all of the available no-match analyses). None of Dr. Stewart's no-match analyses existed when SL 2013–381 was enacted.

As indicated above, there are legitimate non-discriminatory reasons to have made these requests. In addition, the type of disparity indicated by Plaintiffs' data has existed in most cases where a state's ID scheme has been challenged.²¹⁵See, e.g., Veasey, 796 F.3d at 505–06 (finding racial disparities in ID possession but vacating and remanding the district court's finding of discriminatory intent); South Carolina, 898 F.Supp.2d at 40 (finding racial disparities in ID possessions but concluding that South Carolina's ID requirement was not enacted for a discriminatory purpose); cf.Frank, 768 F.3d at 752 (finding racial disparities in ID possession but concluding that no § 2 results violation existed). In fact, Dr. Stewart indicated at this court's January trial that ID possession disparities exist nationwide and that he has yet to find a combination of acceptable photo IDs that will make these disparities go away. (Doc. 408 at 159–60.) To accept this as not only necessary, but sufficient, proof of discriminatory intent would likely invalidate voter-ID laws in any State where they are enacted, regardless of the assortment of IDs selected. (See id.) But Crawford rules that out, and the question is whether the particular assortment of permissible IDs evidences discriminatory intent, given the specific facts on the ground in that State. Crawford, 553 U.S. at 192–97, 128 S.Ct. 1610.

215 Plaintiffs also argue that African Americans are more likely to have suspended or revoked licenses and that this somehow evidences discriminatory intent. (Doc. 419 at 15.) Plaintiffs claim that the legislature was aware of this disparity due to its consideration of another bill presented by Orange and Chatham County Assistant District Attorney, Jeff Nieman. (Id.) Mr. Nieman's bill sought to "remove the automatic revocation for a conviction for driving while license revoked." (Doc. 409 at 124.) One of his key arguments was that those with low socioeconomic status were more likely to end up with suspended licenses. (Id.at 126–27.) But even if legislators accepted this as true and inferred that African Americans were more likely to have suspended licenses, they could not have anticipated an impact under SL 2013–381 without also knowing that North Carolina law prohibits knowingly possessing or displaying a suspended license. N.C. Gen. Stat. § 20–30(1). There is no evidence that this was before the legislature. (SeeDoc. 419 at 15–16.) On its face, SL 2013–381 does not prohibit voting with a revoked or suspended license. N.C. Gen. Stat. § 163–166.13.

Moreover, during the March 2013 hearings on the initial version of the voter-ID requirement, the legislature received the testimony of two witnesses who supported the need for a photo-ID requirement. One was Hans von Spakovsky, a senior legal fellow at the Heritage Foundation who served in the U.S. Department of Justice for four years enforcing voting rights laws, served two years on the Federal Election Commission, and was a consulting expert to the Carter-Baker Report. (Def. Ex. 506 at 40.) While conceding there was no "massive" voter fraud, he urged that the "potential for abuse exists," noted instances of prior voter fraud, cited polls "across ethnic and racial lines" showing support for voter photo ID, and testified that claims of burdens imposed on voters were substantially overstated. (See id. at 39, and attached Statement at 2–5.) He claimed that early voter-ID bills had received bipartisan support in Kansas and Rhode Island, (id. at 39–40), and disagreed that photo ID reduces turnout among any racial group, citing statistics and examples, (id. at 42–43, Statement at 4). For example, he claimed that turnout increased in Georgia in 2012 after it implemented its voter-ID law, despite decreasing in notable States without a photo-ID requirement, including New York and California. (Id. at 42.) He also cited to Indiana, where he claimed Democratic turnout increased in the wake of the State's photo-ID requirement and President Obama became the first Democrat to carry the State since 1964. (Id. at 43.) He likewise argued that these turnout numbers were not anomalies attributable to President Obama, as he claimed that African American turnout in the 2010 midterm exceeded that of the 2008 election. (Id.) Mr. von Spakovsky also attacked the idea that the SBOE's no-match list, which then contained over 613,000 voters, was anywhere close to accurate. (Id. at 43–50.) (Of course, Dr. Stewart's revised no-match list proves as much.) He claimed that evidence from other States indicated that North Carolina's voter rolls were likely inflated and that this inflation was a probable cause of the no-match result because a large share of those on the no-match list were inactive voters. (Id. at 43–47.) He reported that the no-match list in Indiana had been rejected because the court found it unreliable, (id. at 44–45), and that despite large no-match lists in Georgia and Kansas, relatively few voters had actually sought photo-ID in those States, (id. at 49–50). In his mind, this suggested that those States' no-match lists were seriously inflated. (See id.) As a final example, Mr. von Spakovsky claimed that he himself would have appeared on the "no-match list," as the name on his registration was "Hans A. Von Spakovsky," while his DMV issued ID listed his name as "Hans Anatol von Spakovsky." (Id. at 49.) He concluded his testimony by claiming that other western democracies require voters to present photo ID, including Canada and Mexico. (Id. at 51.)

Francis De Luca, president of the Civitas Institute in North Carolina, and who holds a master's degree in political science, provided similar testimony during the March 2013 hearings. (Def. Ex. 507 at 14.) Mr. De Luca claimed that changes in American society had made voting without a photo-ID requirement insecure. (Id. at 15.) He said that the "ability of poll workers to serve as an effective check on voter impersonation" has been "rendered obsolete" by increases in early voting and "the increased urbanization and mobility of the population." (Id.) North Carolina, he further noted, is no longer a rural state, and voters are considerably less likely to vote in their home precincts than in the past. (Id.) He also observed that poll workers do not have access to voters' signatures and, in any case, are not handwriting experts. (Id. at 17.)

Mr. De Luca further provided polling data indicating that majorities of voters wanted an ID requirement. (Id. at 18–19.) He cited polls showing: "2 percent or less of registered voters ... lack a government-issued photo ID"; two-thirds of registered voters support a photo-ID requirement; 70% "of all adults, not just of registered voters," support a photo-ID requirement; three out of four voters responded that "requiring a photo ID would increase the likelihood of them voting"; and, "when asked to describe in their own words what would make elections more secure, 47 percent of voters volunteered that voter ID would make elections more secure." (Id.) Finally, Mr. De Luca claimed that voter turnout was higher in Georgia in 2012 than in North Carolina, even though Georgia had a photo-ID requirement in place and North Carolina did not. (Id. at 21.)Regardless of whether or not the claims made by Mr. von Spakovsky and Mr. De Luca are true,²¹⁶ the legislature could reasonably have believed them to be true. This is in large part because Mr. von Spakovsky's and Mr. De Luca's ideas are consistent with other authoritative sources. A legislator could have looked at the precipitous drop in the number of voters on the no-match list from January 2013 (612,995) to April 2013 (318,643) and credited Mr. von Spakovsky's claims that the no-match process grossly overestimates the number of North Carolinians who lack photo ID. The clear trend was that each time the no-match list was revised, the number of no-matches fell substantially. (Pl. Ex. 891 at 4 (tbl. 2).) This trend has continued through the course of this litigation. (Id. at 19 (tbl. 11).) Other courts' examination of no-match analyses could likewise have led legislators to doubt the accuracy of the process. See, e.g., Billups, 554 F.3d at 1354 (11th Cir.2009) ; Billups, 504 F.Supp.2d at 1360–63, 1378 & n. 6 (N.D.Ga.2007) (noting that the presiding Judge "appeared on one of the no-match lists"). Moreover, Mr. De Luca's claim that the modernization and urbanization of the American electorate has made poll workers an obsolete guard against voter impersonation is consistent with the Carter-Baker Report. See Carter-Baker Report, at 18. Finally, that the benefits of a photo-ID requirement outweigh any burden on voters is part and parcel of Crawford, where the Supreme Court recognized that there are many legitimate (i.e., nondiscriminatory) reasons for a legislature to enact an ID requirement. Crawford, 553 U.S. at 192–97, 128 S.Ct. 1610.

216 They were not offered for their truth or falsity, but only for notice and intent.

Plaintiffs rely on the opinions and studies of Dr. Lichtman, offered as an expert in American political history, electoral analysis, and historical and quantitative methodology. Dr. Lichtman was offered to show that SL 2013–381 was passed with intent to discriminate against African Americans. This court does not credit Dr. Lichtman's opinions for several reasons. First, although Plaintiffs argued otherwise, Dr. Lichtman's ultimate opinions on legislative intent, like those of Plaintiffs' other two experts on legislative intent, Drs. Steven Lawson²¹⁷ and Morgan Kousser,²¹⁸ constituted nothing more than his attempt to decide the ultimate issue for the court, rather than assisting the trier of fact in understanding the evidence or any fact at issue. SeeFed. R. Evid. 702(a). Basically, all of these experts gathered evidence, principally from newspaper and magazine articles, that they believed fit under each Arlington Heights factor. Then, they opined on how the Arlington Heights analysis, (or their variant of it) ought to be performed, but contended they were doing so to determine intent "as historians."

217 Professor Emeritus of History at Rutgers University. (Doc. 334 at 18.)

218 William R. Kenan, Jr. Professor of History and Social Science at the California Institute of Technology. (Doc. 330 at 42–43.)

The court doubts seriously that this is the proper role for expert testimony. "[O]pinion testimony that states a legal standard or draws a legal conclusion by applying law to the facts is generally inadmissible." United States v. McIver, 470 F.3d 550, 562 (4th Cir.2006). The Fourth Circuit identifies "improper legal conclusions by determining whether the terms used by the witness have a separate, distinct and specialized meaning in the law different from that present in the vernacular." Id.(citations and internal quotation marks omitted). Such improper conclusions include whether a defendant's actions constituted "extortion" or "deadly force," whether there existed a "fiduciary relationship," and whether a product is "unreasonably dangerous." Id. It also includes intent, whether legislative or otherwise, being a conclusion that is "particularly within the province of the trier of fact." Allen Organ Co. v. Kimball Int'l, Inc., 839 F.2d 1556, 1567 (Fed.Cir.1988) ; accordSuperior Prod. P'ship v. Gordon Auto Body Parts Co., 784 F.3d 311, 323–24 (6th Cir.2015) ; United States v. Rahman, 189 F.3d 88, 136 (2d Cir.1999) ; DePaepe v. Gen. Motors Corp., 141 F.3d 715, 720 (7th Cir.1998) ; Woods v. Lecureux, 110 F.3d 1215, 1220 (6th Cir.1997) ; Salas v. Carpenter, 980 F.2d 299, 305 (5th Cir.1992) ; United States v. Benson, 941 F.2d 598, 604 (7th Cir.1991) ; Price v. Austin Indep. Sch. Dist., 945 F.2d 1307, 1317–18 (5th Cir.1991) ; SMD Software, Inc. v. EMove, Inc., 945 F.Supp.2d 628, 643 (E.D.N.C.2013) ; Safeway, Inc. v. Sugarloaf P'ship, LLC, 423 F.Supp.2d 531, 538–39 (D.Md.2006) ; Smith v. Wyeth–Ayerst Labs. Co., 278 F.Supp.2d 684, 699–700 (W.D.N.C.2003) ; U.S. Gypsum Co. v. Lafarge N. Am. Inc., 670 F.Supp.2d 768, 775 (N.D.Ill.2009).

Second, and independently, the court disregards Dr. Lichtman's opinions because his approach was single-minded and purposefully excluded evidence that contradicted his conclusions. By way of example, Dr. Lichtman did not include in his report, nor did he find relevant during trial, the fact that North Carolina had a two-and-a-half-year roll out of the photo-ID requirement, which was plainly designed to make it as simple as possible for citizens to comply with the requirement. In fact, to the best of Dr. Kousser's knowledge, North Carolina had "the longest rollout period of any state that has enacted a photo-ID" requirement, of which there were many in 2012. (Doc. 330 at 92 (Dr. Kousser: "I looked at other states, and they were required quickly. They didn't have an intervening election before they were required to go into effect."); Doc. 333 at 192 (Dr. Lichtman: "I didn't do a state-by-state comparison [of roll out periods]."); see also Def. Ex. 2 at 4–5.) In this respect, SL 2013–381 is substantially consistent with the bipartisan recommendation by former President Carter and former Secretary of State Baker that States adopt a photo-ID requirement for voting and that it be implemented with notice over time to ease the transition. Carter Baker Report, at 18–19; seeCrawford, 553 U.S. at 193–94, 128 S.Ct. 1610.²¹⁹

219 In fact, Justice Breyer based his objection to the Indiana voter-ID law in Crawfordin part on the fact that Indiana failed to follow this recommendation of the Carter-Baker Report, whose findings he characterized as "highly relevant to both legislative and judicial determinations of the reasonableness of a photo ID requirement." Crawford, 553 U.S. at 238–41, 128 S.Ct. 1610 (Breyer, J., dissenting). He also objected to what he saw as Indiana's failure to abide by the Carter-Baker Report's other condition—that IDs "be easily available and issued free of charge." Id.at 238–39, 128 S.Ct. 1610. As noted herein, SL 2013–381 alleviates the cost of obtaining an ID for those who need to obtain one. CompareN.C. Gen. Stat. § 130A–93.1(c) (waiving the usual ten dollar fee for obtaining a birth certificate or marriage license if a voter declares she needs such a document in order to vote), withCrawford, 553 U.S. at 239, 128 S.Ct. 1610 (noting that those needing a birth certificate in Indiana would still have to pay the State's usual twelve dollar fee, and the indigency exception required voters to travel to the county clerk's office or CBOE after each election to sign an affidavit).

Although quick to compare North Carolina's law to that of other jurisdictions, Dr. Lichtman did so selectively, comparing only claimed negative aspects and omitting any positive aspect, such as the photo-ID soft roll out and voter education campaign. Similarly, he ignored the exemption in the photo-ID requirement for curbside voters, of which African Americans make up a disproportionately large share, because, in his words, this is "such a small group." (Doc. 333 at 184.) In 2014, there were 37,396 curbside voters, of which 17,415 were African American. (Doc. 332 at 110–11.)²²⁰

220 By contrast, there were only 1,387 OOP ballots excluded from Election Day in 2014. (Pl. Ex. 689.)

Third, Dr. Lichtman refused to consider that legislators can have legitimate policy differences over these election mechanisms unrelated to race, given historical evidence of problems with their use and implementation, such as with SDR and OOP. This is surprising given that Dr. Lichtman is himself a politician and a former Democratic candidate for the United States Senate in Maryland. (Doc. 333 at 173.)

Fourth, at trial Dr. Lichtman practiced a propensity to respond to questions not with responsive answers, but with non-responsive arguments supporting his opinions. He also demonstrated a willingness to obfuscate when detail became important. (E.g., Doc. 342 at 166–70, 172–75 (Dr. Lichtman: attempting to avoid acknowledging that he lacked personal knowledge of SEIMS and the data extraction performed by a third party, Mr. David Ely, who "does the data work for [him]").) In short, he presented as an advocate, leaving the court at a loss for where facts left off and advocacy began. The court has substantial questions about his credibility and has difficulty relying on much of his testimony.

The most substantial piece of evidence put forth by Dr. Lichtman was that certain forms of ID that were retained by SL 2013–381—DMV IDs, expired IDs for those over age seventy, U.S. passports, and veteran and military IDS—"provide relatively greater access to IDs for whites," while the forms of ID not retained by SL 2013–381—student IDs, government employee IDs, public assistance IDs, and expired IDs—"provide relatively greater access to IDs for African Americans." (Pl. Ex. 231 at 99 (tbl. 32); Pl. Ex. 716, AL 6; Doc. 419 at 19–20.) There is no evidence, however, that the legislature had data showing that government employee IDs or expired IDs were disproportionately held by African Americans. It would have been reasonable for the legislature to view expired IDs as less secure, and to strike a balance by allowing them for those over seventy but not allowing them for younger individuals. After all, it is reasonable to assume that those over seventy are more likely than younger individuals to suffer from negative health that would make a DMV visit, or continued driving, more difficult. Second, although Representative Lewis sought data on the number of student IDs created and the percentage of those who were African American, the data that he was provided actually suggested that African Americans were less likely to hold college IDs.²²¹ Further, the legislature offered at least plausible reasons for excluding student IDs: (1) there was inconsistency in the way college IDs were done, (Pl. Ex. 202 at 68–69), and (2) permitting student IDs would be redundant because some schools require a photo ID to get a student ID, (Pl. Ex. 138 at 13).²²² Nevertheless, the removal of public assistance IDs is somewhat suspect. While there is no evidence of public assistance ID possession rates in this case, and no evidence that such data were before the legislature, a reasonable legislator aware of the socioeconomic disparities endured by African Americans could have surmised that African Americans would be more likely to possess this form of ID. However, as a whole, the IDs retained and removed by SL 2013–381 are not nearly as suspect as Plaintiffs claim.

221 Representative Lewis's original request was for system-wide numbers. (Pl. Ex. 334 at 2.) But given that the University of North Carolina did not have a way to pull the numbers for all seventeen campuses, Representative Lewis was provided an estimate. (Id.at 1.) Given that "you have to have [a student ID] for everything—library, food, etc.," the university system reasoned that an adequate approach would be to provide numbers on the enrollment of African American students. (Id.) Accordingly, Representative Lewis was informed that African Americans were 8.9% of students at UNC-Chapel Hill and 21.1% of students at the UNC System as a whole. (Id.) If Dr. Lichtman is correct that African Americans are more likely to be enrolled in public colleges and universities in the State than whites (and thus, by inference, more likely to possess college IDs from those institutions), (seePl. Ex. 716, AL 6; Pl. Ex. 231 at 50–51), then this result is counterintuitive given the educational disparities put forth by Plaintiffs in this case.

222 SeeinfraPart II.D (Twenty-Sixth Amendment analysis).

For all of these reasons, the impact from implementation of SL 2013–381 does not significantly favor a finding of discriminatory purpose.

The next factor to be considered is whether there is "evidence of a 'consistent pattern' of actions by the decisionmaking body disparately impacting members of a particular class of persons." Sylvia, 48 F.3d at 819. As set out above, North Carolina's past is mired by official discrimination against African Americans involving the General Assembly. However, there is little evidence of official discrimination since the 1980s.²²³See supra Part II.A.2.b.

223 Although not addressed by the parties, the court is aware that North Carolina has been accused of considering race as part of the redistricting process after the 1980s. Seesupranote 138. Many of these cases involved accusations that the State had improperly enhanced minority voting power. Seesupranote 138 (discussing Bartlett v. Strickland, 556 U.S. 1, 8, 129 S.Ct. 1231, 173 L.Ed.2d 173 (2009) ; Shaw v. Hunt, 517 U.S. 899, 906, 116 S.Ct. 1894, 135 L.Ed.2d 207 (1996) ). Courts examining North Carolina's current redistricting efforts have reached differing conclusions as to the role that race played in the creation of those districts. CompareHarris v. McCrory, 159 F.Supp.3d 600, 602–04, 625–26, No. 1:13cv949, 2016 WL 482052, at *2, *21 (M.D.N.C. Feb. 5, 2016) (finding that certain congressional districts were drawn primarily on the basis of race but expressly declining to make a finding as to whether the State acted in good faith); id.at 629–30, 2016 WL 482052 at *23 (Osteen, J., concurring in part and dissenting in part) (noting that the plaintiffs "[c]onceded at trial they did not seek to prove any ill-intent"), withDickson v. Rucho, 368 N.C. 481, 781 S.E.2d 404, 441 (2015) (finding no federal constitutional violation for several State congressional districts, including the districts at issue in Harris). But even if recent redistricting efforts indisputably proved that that the legislature was willing to consider race when it was not required to do so, that would not change the result here in light of the legitimate interests served by SL 2013–381 and the other considerations discussed herein. Moreover, this issue is of limited probative value to any discriminatory impact claim because there is no evidence that recent redistricting efforts had any effect on the socioeconomic conditions of minorities.

The next factor to be considered is the "historical background of the decision, which may take into account any history of discrimination by the decisionmaking body or the jurisdiction it represents." Sylvia, 48 F.3d at 819. This encompasses "the specific sequence of events leading up to the particular decision being challenged, including any significant departures from normal procedures." Id. Plaintiffs urge the court to find discriminatory intent based on the timing and character of amendments to SL 2013–381 following the Supreme Court's Shelby County decision.

Originally, HB 589 was focused on adding a photo-ID requirement to voting. It was debated over a few weeks in the House, where it originated. There is no contention that the bill received anything but full consideration in the House. Representative Rick Glazier, who strongly opposed the bill, testified that he felt that "[f]or a large bill," HB 589 received up to this point "the best process possible" in the House, one he characterized as "excellent." (Doc. 165 at 56–57.)

HB 589 then passed to the Senate on April 25, 2013, where it remained in the Senate Rules Committee. At the time, the Supreme Court had heard argument on a challenge to the constitutionality of the formula the VRA used to determine the "covered" jurisdictions subject to preclearance. Many of North Carolina's counties were covered jurisdictions.

On June 25, the Supreme Court issued its decision in Shelby County, declaring the preclearance coverage formula to be unconstitutional because, among other things, it was based on historical conditions in the 1960s and failed to take into account States' current conditions. The next day, Senator Apodaca, Republican Chairman of the Rules Committee, publicly stated, "I think we'll have an omnibus bill coming out" or words to the effect that the Senate would move ahead with the "full bill." (Pl. Exs. 81, 714.) It was not until July 23, however, that an expanded bill, including the election changes challenged in this case, was released. As detailed in the factual findings above, the bill was debated over portions of the next three days and ultimately passed both houses on July 25.

Plaintiffs find racial motivation in the way the General Assembly enlarged the bill after Shelby County. In the words of their expert, Dr. Lichtman:

The H.B. 589 was substantially changed after the Shelby decision .... The timing is of crucial significance here. Because the Shelby decision relates only to race, and if, in fact, these subtractions and additions had nothing to do with race and were clearly nondiscriminatory, then the timing would not have to wait until the issuance of the Shelby decision to make these fundamental changes in H.B. 589 and, of course, without going through the same extended, exacting process.

(Doc. 333 at 106.) But this is not the only, or even the most persuasive, explanation for the enlargement's proximity to Shelby County. Because Shelby County removed North Carolina from preclearance, it greatly altered the burden of proof calculus for North Carolina legislators considering changes to voting laws. Despite Dr. Lichtman's characterization of the case, prior thereto legislators still would have had to demonstrate that the changes were not retrogressive—something that, while perhaps easier now with the 2014 data, assuredly would have been hotly contested by opponents. It would not have been unreasonable for the North Carolina Senate to conclude that passing the "full bill" before Shelby County was decided was simply not worth the administrative and financial cost of seeking permission from the United States. In addition, opponents of SL 2013–381 made clear at the time that North Carolina still faced being sued under § 2 of the VRA, and it would not be unreasonable for the legislature to have believed that some voting changes may survive a § 2 challenge but not one under § 5, especially given the shifted burden of proof and the uncertainty of § 2's application to the removal of only recently-adopted voting mechanisms. On this front, it is highly relevant that proponents of the bill had received in their deliberations an analysis of other legal challenges to voter photo-ID laws—indicating that such laws passed muster with the courts. (See Def. Ex. 507 at 35–36 (De Luca Report).) Further, while Plaintiffs portray the law's expansion from sixteen to fifty-seven pages as significant, as noted above the provisions challenged in the present lawsuit and the voter-ID requirement comprise approximately fifteen of HB 589's fifty-seven pages, nine of which related to voter ID and virtually all of which had been debated before Shelby County. (See Pl. Ex. 107.) Moreover, all changes to the original version of the bill were highlighted in the draft bill. (See id.) It is also not accurate to say that HB 589 contained novel ideas that opponents were unprepared to address, as virtually all of the changes were the subject of prior and pending bills.

Next, Plaintiffs contend that in enacting HB 589, the General Assembly departed from the normal procedural sequence for legislation. (Doc. 346 at 136.) But their experts and fact witnesses (including Democratic members of the General Assembly) all concede that the General Assembly acted within all the procedural rules when it enacted HB 589; not a single opponent raised a point of order. (See, e.g., Doc. 330 at 94–96 (Dr. Kousser); Doc. 333 at 215–18 (Dr. Lichtman); Doc. 334 at 62–63 (Dr. Lawson); Doc. 335 at 193–94 (Sen. Stein); Doc. 336 at 37–38 (Rep. Michaux).) According to Senator Stein, in some respects, even more process was accorded the opponents of HB 589 than was required by the Senate's rules. (Doc. 335 at 209–10 (noting that Senate Rules Committee chairman allowed ten members of the public to speak on HB 589, even though Senate rules do not require public comment).) In every respect, the legislation was considered and passed in accordance with the procedural protections of the formal legislative rules.

Plaintiffs further contend that, even so, the proponents nevertheless departed from general custom and practice. They argue that the bill was "rushed" through both chambers "in only three days, without sufficient time or opportunity to assess on the record the likely impact of the bill." (Doc. 346 at 136.) Given that formal rules were followed, this claim in essence is that the process for the bill was unusual for a bill of this magnitude. (See Doc. 165 at 67 (Rep. Glazier: "I was shocked by it, not by, in some respects, some of the provisions, but by the—and, again, my comments on the floor that night made it clear—by the process"), 69 ("[t]he process this bill got was nothing more than what we give to a golf cart bill"); Pl. Ex. 18 at 3 (Sen. Stein: describing the Senate proceedings as "irregular for a bill of this magnitude").) The trial evidence revealed that these claims, when evaluated in the light of contemporary legislative process, are weak.

First, if these claims were true, it would have been odd for the Senate Minority Leader, Martin Nesbitt, a Democrat, to have declared at the end of the debate, as he did: "Thank you, Mr. President; Mr. President and members of the Senate, we've had a good and thorough debate on this bill over two days. We've had a sense of history. I think we've reviewed the bill in great detail. I think everyone in the room knows what we're doing now." (Pl. Ex. 550 at 90–91.) In fact, every opponent of HB 589 was given an opportunity to voice any opposition openly on the floor of each chamber to the point that their criticisms became repetitive.

Further, as noted previously, many of the amendments to HB 589 were drawn from several bills already introduced and pending in the General Assembly. See supra Part I.B.2. Although not disclosed in a combined bill, some of the changes reflected continuations of debates that had been ongoing for years and were far from secret. (Pl. Ex. 789 at 30–31 (former member of the House, Carolyn Justice) ("[T]hese things that were eventually added in the Senate were not new surprising ideas. They were pieces of other bills.").) This is demonstrated by the fact that once the Senate introduced the newly-amended version of HB 589, opponents had already prepared their opposition. The very next day, Senator Stein opposed the bill presenting data, charts, and statistics prepared by advocacy groups. (See Pl. Ex. 18 at 17–24.) This suggests strongly that they had been monitoring the bills in the hopper and were prepared to oppose them.

The evidence at trial further showed that the process known as "gut-and-amend" used to transform the voter-ID bill into the bill that became SL 2013–381 is not uncommon in the General Assembly. (Doc. 164 at 133 (Senator Blue, an opponent of the bill: acknowledging that gut-and-amend happens "quite a bit" and "too often" in the General Assembly).) Such a process occurs because the General Assembly must meet a "cut-off" date—known as the "cross-over date"—by which a piece of legislation must be approved by one chamber lest it die for the remainder of the session. (Id. at 131–33.) Plaintiffs' legislator-witnesses admitted that it is not uncommon for a bill to return to its originating house with significant material not originally part of the bill. (Id. at 133; Doc. 165 at 85–88 (Rep. Glazier).)

Plaintiffs focus on the three days of consideration the bill received in July 2013 as being abrupt. This, of course, ignores the extensive debate and consideration the initial voter-ID bill received in the spring, which opponents acknowledged was "excellent," (Doc. 165 at 56–57), as well as Senator Nesbitt's contemporaneous statements in July. As noted, the vast majority of the bill, as amended, restated prior law or dealt with issues not under contention now. The three days of debate in July focused almost exclusively on the few amendments at issue here. Moreover, trial evidence revealed other bills of similar import that moved on faster tracks. For example, in 2003, the Democratically-controlled General Assembly passed a redistricting bill in a special session that drew complaints on the floor about the way it was considered and the quick process it received. (Doc. 336 at 45–52 (Rep. Michaux); Def. Ex. 217 at 8–10 (former Rep. Carolyn Justice).) It passed nevertheless. HB 589 received considerably more debate and public comment.

Plaintiffs also argue that only one member of the House spoke in favor of HB 589. (Doc. 346 at 70.) While it is true that only Representative Lewis spoke in the House before the vote to concur in the Senate's changes, several Republican senators spoke in favor of the bill both during the Senate Rules Committee meeting and during the two Senate floor sessions. (See generally Doc. 134–4.) Additionally, the initial bill was debated over several committee sessions and a floor session in March and April 2013. (See generally Pl. Exs. 127–134.) It is not necessarily nefarious that no Republican in the House other than Representative Lewis rose to speak in favor of the bill when it was late in the evening, the caucus knew it had the votes to pass the bill, and the end of the legislative session was approaching.²²⁴

224 Indeed, an opponent of the bill candidly testified at the hearing that, had he been the lawyer for the Republicans, he would have similarly advised the strategy to avoid further discussion. (Doc. 165 at 70.)

Next, in considering the process a bill receives, it is relevant whether any "ameliorative" amendments or amendments proposed by the bill's opponents were accepted. SeeVeasey, 796 F.3d at 511 ; cf.Brown, 895 F.Supp.2d at 1246–49. Most significantly, Senator Stein, a Democrat and opponent of HB 589, proposed the same-hours requirement for early voting that was overwhelmingly approved. As noted above, this amendment ensured that voters retained the same number of aggregate hours as under the previous early-voting schedule. There is no indication that the proponents accepted this amendment for any reason other than because they agreed with the interests the amendment advanced. The Senate also accepted an amendment dealing with electioneering from Senator Dan Blue (Democrat), (Pl. Ex. 550 at 82–83), and two amendments from Senator Clark (an African American Democrat) to facilitate the use of absentee voting during the early-voting period, (Pl. Ex. 202 at 23–29).

The final Arlington Heights factor examines "contemporary statements by decisionmakers on the record or in minutes of their meetings." Sylvia, 48 F.3d at 819. The most probative statements come from the proponents, not opponents, of the bill. See, e.g., Butts v. City of New York, 779 F.2d 141, 147 (2d Cir.1985) ; Veasey, 796 F.3d at 502 (collecting cases). Statements from no more "than a handful" of the bill's proponents are also of limited probative value. Hunter v. Underwood, 471 U.S. 222, 228, 105 S.Ct. 1916, 85 L.Ed.2d 222 (1985) (quoting United States v. O'Brien, 391 U.S. 367, 383–84, 88 S.Ct. 1673, 20 L.Ed.2d 672 (1968) ). Moreover, statements by proponents should generally be contemporaneous with the legislature's consideration of the bill rather than post facto statements by the bill's proponents. Barber v. Thomas, 560 U.S. 474, 485–86, 130 S.Ct. 2499, 177 L.Ed.2d 1 (2010) ("And whatever interpretive force one attaches to legislative history, the Court normally gives little weight to statements, such as those of individual legislators, made after the bill in question has become law."); Veasey, 796 F.3d at 502 (collecting cases).

Plaintiffs argue that HB 589's proponents' statements on the floor are pretextual for racial animus and addressed problems that were "merely imaginable." (Doc. 346 at 137.) As the court's analysis of the ninth Gingles factor has shown, see supra Part II.A.2.j, the contemporaneous justifications for the changes to the early-voting schedule, and the removal of SDR, OOP,²²⁵ and pre-registration²²⁶ were non-tenuous. In addition, the opponents' objections have been subsequently proved inaccurate by later election data. Defendants also offered non-tenuous contemporaneous justifications for the voter-ID requirement, namely establishing integrity and confidence in the electoral process. (Pl. Ex. 202 at 3); Crawford, 553 U.S. at 204, 128 S.Ct. 1610.

225 The contemporaneous justification offered for the repeal of OOP voting was described as "basically mov[ing] the law back to the way it was prior to 2005." (SeePl. Ex. 202 at 12.) The court interprets this as a reference to the precinct-based system affirmed and justified by the Supreme Court in 2005 in James. Given the Jamesdecision's ample discussion of the benefits of a precinct-based system, and given that no legislator in the House or Senate openly opposed the removal of OOP voting during the legislative debate, (see, e.g., Doc. 335 at 208), the brevity of this explanation is understandable.

226 Although the contemporaneous justification offered for the removal of pre-registration was weaker than for early voting, SDR, or OOP, as noted above, there is no evidence that the legislature had before it data on the use of pre-registration by race.

Finally, the court has considered the cumulative evidence Plaintiffs have presented on intent under the totality of the circumstances. It is, of course, possible that an action may appear legitimate when considered in isolation but discriminatory when considered with other related decisions. Indeed, here, Plaintiffs argue that HB 589, when considered as a whole, evidences discriminatory intent to suppress voting.

Taken as a whole, SL 2013–381 enacted many changes to North Carolina's election laws that are not challenged here, suggesting strongly that those were legitimate, non-discriminatory revisions. The State's proffered justifications for the combined mechanisms under review here are consistent with the larger purpose of achieving integrity, uniformity, and efficiency in the political process. Collectively, the changes were designed to make early-voting locations more numerous and evenly distributed and voting hours more uniform; reduce the number of individuals who forego traditional registration (where they can be subjected to statutory mail verification) and instead register and vote too close to Election Day such that their vote is counted despite later failing mail verification; re-establish the benefits of a precinct-based system recognized in James; reduce voter confusion among pre-registrants; and, as recognized in Crawford, promote the integrity and reliability of the electoral process while increasing public confidence in North Carolina's electoral system through implementation of a voter-ID requirement. These are legitimate and consistent interests.

Plaintiffs charge that—because the common theme is that SL 2013–381 removed multiple voting procedures that African Americans disproportionately used—the legislature must have meant to disenfranchise African Americans. Because the court finds that early-voting availability remains materially the same under the new law, this charge is leveled principally at SDR and OOP. In light of the whole trial record, the court cannot infer a racial motive given the consistency and legitimacy of the State's proffered justifications, the relatively recent implementation of these mechanisms, and the evidence that voters were knowingly exploiting SDR and OOP in increasing fashion so as to undermine the benefits of traditional registration (statutory verification) and precinct management that proponents claim to protect.

In sum, there was evidence that the legislature had data on disparate use of early voting, SDR, and OOP by African Americans, although some of the data were not provided until after HB 589 was drafted and introduced; there is no evidence that the legislature had demographic data on the use of pre-registration. The legislature had data that African Americans disproportionately lacked DMV-issued IDs. But, as noted above, there were legitimate non-discriminatory reasons to request demographic data for each of the voting changes, especially prior to Shelby County when § 5 was in force. Finally, Plaintiffs presented evidence that the forms of ID not retained by HB 589 were more available to African Americans, but given the evidence that has been shown to have been before the legislature, the only form of ID for which the legislature plausibly could have inferred disproportionate use was public assistance IDs.

By contrast, the legislature offered non-tenuous justifications for the actions taken. In fact, with regard to voter ID, early voting, SDR, and OOP, this court has found the justifications offered to serve substantial State interests. In addition, in implementing the voter-ID requirement, the legislature provided for a two-year implementation to give notice and made voter IDs available for free, consistent with recommendations and the spirit of the Carter-Baker Report. In practice, this meant that the legislature attempted to soften any burden imposed by the ID requirement by giving voters over two years to acquire a free ID. Further, the bill was passed under ordinary legislative rules. All opponents were permitted to speak, and several amendments were accepted, including a significant one from Senator Stein that ameliorated the asserted burden imposed by the reduction in early-voting days. What remains under the law provides all voters with an equal and ample opportunity to participate in the political process. But even if this court had concluded that HB 589 results in an inequality of opportunity, the countervailing evidence nevertheless compels the conclusion that HB 589 was enacted in spite of, not because of, this impact.

Having considered the entire record as a whole, this court is not persuaded that racial discrimination was a motivating factor of HB 589. Accordingly, Plaintiffs have failed to establish that the legislature acted with discriminatory intent.

6. Additional Problems with the § 2 Results Claim

The above analysis is sufficient for the court to find that Defendants are is entitled to judgment on Plaintiffs' § 2 results claims. However, there are two additional problems with Plaintiffs' results claims.

At trial, the court pressed Plaintiffs as to, assuming it were to grant injunctive relief, how it would measure the quantum of relief necessary to provide them "equal opportunity," keeping in mind that the State has the right to propose a proportional remedy. Plaintiffs were unable to articulate a response, other than to urge return of the repealed mechanisms, or suggest that the parties may reach a settlement. (Doc. 343 at 26–35.) As the court inquired, if there is no way to know when voting rights are in parity, how does one determine whether there is a violation? (Id. at 35.)

In an apparent response to the court's concern, Plaintiffs set out the following theory of § 2 in their post-trial briefing:

A Section 2 case looks forward—to "what the right to vote ought to be." [Reno v. Bossier Par r ish [Parish] Sch. Bd., 528 U.S. 320, 334 [120 S.Ct. 866, 145 L.Ed.2d 845] (2000) ]. In many Section 2 cases, the standard, practice, or procedure that affords all citizens an equal opportunity to participate will be one that has never before been implemented by the jurisdiction because the jurisdiction has historically never provided minority citizens with the equal opportunity to participate and elect representatives of their choice. Put somewhat differently, Section 2 does not require that plaintiffs identify a preexisting practice within the jurisdiction to serve as a "baseline" against which the challenged practice should be measured. That being said, the Supreme Court in Bossier Parish made clear that where a Section 2 claim alleges that a change to an existing practice violates Section 2, a court can look to the prior practice as part of the results test inquiry. In this case, the fact that North Carolina successfully used SDR, OOP, 17 days of early voting, and pre-registration over several election cycles makes these practices highly appropriate to use an illustrative example because, unlike a hypothetical, the feasibility of, and results under, these practices are known and documented.

(Doc. 346 at 122.) Consistent with this position, Plaintiffs contend that SDR permitted African Americans and Hispanics to register on a "more equal" basis with whites. (Doc. 346 at 124.) Plaintiffs never state that the voting mechanisms in place prior to SL 2013–381 provided an equal opportunity to African Americans and other disadvantaged groups. Instead, under Plaintiffs' theory of the case, even though North Carolina was a covered jurisdiction for fifty years under the VRA subject to DOJ supervision, the State has never offered an equal opportunity for minorities to participate in the electoral process.

Of course, this theory conflicts with the data since 2000, even without Plaintiffs' preferred voting procedures. Moreover, it presents an elusive basis for fashioning an appropriate remedy. While this court does not suggest that a plaintiff is entitled to no relief simply because under his theory of the case the remedy would not be complete, the fact remains that in this case Plaintiffs' failure to offer a principled, non-arbitrary way to grant relief is part-and- parcel of their inability to show inequality of opportunity.

Vote dilution cases offer an illustration of the type of problem that granting relief in this case would involve. In Holder v. Hall, the Supreme Court rejected a § 2 claim challenging the size of a governing authority for lacking any "reasonable alternative benchmarks." 512 U.S. 874, 885, 114 S.Ct. 2581, 129 L.Ed.2d 687 (1994). There, African American plaintiffs brought a § 2 claim to the use of a single commissioner to govern the county. Id. at 876–77, 114 S.Ct. 2581. Such had been the practice in Bleckley County since 1912. Id. In 1985, the Georgia legislature authorized, but did not mandate, the county to adopt a multi-member county commission consisting of five commissioners elected from single-member districts. Id. at 877, 114 S.Ct. 2581. In a referendum, the electorate rejected the change. Id. Plaintiffs claimed that, under the single-member district plan, African Americans could constitute a majority in one district. Id. at 878–79, 114 S.Ct. 2581.

The Supreme Court rejected this challenge and discussed the differences between § 2 and § 5 of the VRA throughout the opinion. In the plurality opinion, Justice Kennedy explained that § 2 requires plaintiffs to prove "a reasonable alternative practice as a benchmark against which to measure the existing voting practice." Id. at 880, 114 S.Ct. 2581. "In certain cases," Justice Kennedy noted,

the benchmark for comparison in a § 2 dilution suit is obvious. The effect of an anti-single-shot voting rule, for instance, can be evaluated by comparing the system with that rule to the system without that rule. But where there is no objective and workable standard for choosing a reasonable benchmark by which to evaluate a challenged voting practice, it follows that the voting practice cannot be challenged as dilutive under § 2.

Id. at 880–81, 114 S.Ct. 2581. Turning to the particular claim at issue, Justice Kennedy explained that there "is no principled reason why one size [governing authority] should be picked over another as the benchmark for comparison." Id. at 881, 114 S.Ct. 2581. The challengers gave several reasons for choosing the five-commissioner benchmark, but the Court rejected them all, noting that there must be a "convincing reason" for choosing a particular benchmark. Id. at 882, 114 S.Ct. 2581 ("One gets the sense that respondents and the United States have chosen a benchmark for the sake of having a benchmark. But it is one thing to say that a benchmark can be found, quite another to give a convincing reason for finding it in the first place.").

The Court rejected the notion that retrogression could be imported from the § 5 context into the § 2 context to establish a benchmark, explaining the differing purposes of the two sections. Id. at 883–84, 114 S.Ct. 2581. Justice Kennedy then observed,

Retrogression is not the inquiry in § 2 dilution cases. 42 U.S.C. § 1973(a) (whether voting practice "results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color"); S. Rep. No. 97–417, p. 68, n. 224 (1982) U.S.C.C.A.N. 1982, p. 177 ("Plaintiffs could not establish a Section 2 violation merely by showing that a challenged reapportionment or annexation, for example, involved a retrogressive effect on the political strength of a minority group"). Unlike in § 5 cases, therefore, a benchmark does not exist by definition in § 2 dilution cases. And as explained above, with some voting practices, there in fact may be no appropriate benchmark to determine if an existing voting practice is dilutive under § 2. For that reason, a voting practice that is subject to the preclearance requirements of § 5

is not necessarily subject to a dilution challenge under § 2.

Id. at 884, 114 S.Ct. 2581.

Justice O'Connor, in her concurrence, agreed with Justice Kennedy, observing that, "to determine whether voters possess the potential to elect representatives of choice in the absence of the challenged structure, courts must choose an objectively reasonable alternative practice as a benchmark for the dilution comparison." Id. at 887, 114 S.Ct. 2581 (O'Connor, J., concurring in part and concurring in the judgment). She rejected the challengers' notion that a benchmark could be established in Holder because the "wide range of possibilities [of a governing body's size] makes the choice inherently standardless." Id. at 889, 114 S.Ct. 2581. She also rejected the notion that the search for a benchmark should even be limited to the experiences of Georgia, wondering whether there should not be a twenty- or forty-member commission as could be found in some other States. Id. Finding no way to determine a benchmark, she concluded that the challengers "do not explain how common an alternative practice must be before it can be a reliable alternative benchmark for the dilution comparison, nor do they explain where the search for alternative benchmarks should begin and end." Id. at 890, 114 S.Ct. 2581.²²⁷

227 Justice Thomas wrote a separate concurrence, joined by Justice Scalia, rejecting the notion that vote dilution claims are cognizable under § 2, though he also seemed to agree with the reasons given by the plurality. Id.at 891, 945, 114 S.Ct. 2581 (Thomas, J., concurring in the judgment). The opinions of Justices Kennedy and O'Connor are controlling since they are based on narrower grounds.

Subsequently, in Reno v. Bossier Parish School Board, a majority of the Court further explained the differing benchmarks at issue in cases under § 2 and § 5:

The term "abridge," however—whose core meaning is "shorten"—necessarily entails a comparison. It makes no sense to suggest that a voting practice "abridges" the right to vote without some baseline with which to compare the practice. In § 5 preclearance proceedings—which uniquely deal only and specifically with changes in voting procedures—the baseline is the status quo that is proposed to be changed: If the change "abridges the right to vote" relative to the status quo, preclearance is denied, and the status quo (however discriminatory it may be) remains in effect. In § 2 or Fifteenth Amendment proceedings, by contrast, which involve not only changes but (much more commonly) the status quo itself, the comparison must be made with a hypothetical alternative: If the status quo"results in [an] abridgement of the right to vote" or "abridge[s] [the right to vote]" relative to what the right to vote ought to be, the status quo itself must be changed. Our reading of "abridging" as referring only to retrogression in § 5, but to discrimination more generally in § 2 and the Fifteenth Amendment, is faithful to the differing contexts in which the term is used.

528 U.S. 320, 333–34, 120 S.Ct. 866, 145 L.Ed.2d 845 (2000) (alternation in original; citations and footnote omitted). Thus, in § 2 cases, Plaintiffs must prove what the right to vote "ought to be."²²⁸Gingles itself requires plaintiffs to prove, as an initial matter, three preconditions not found in the text of the VRA:

228 In its now vacated opinion, the Sixth Circuit distinguished these cases as being in the vote dilution context. Citing the text of § 2, the court reasoned:

vote denial claims inherently provide a clear, workable benchmark. Again, under Section 2(b), the relevant inquiry is whether minority voters "have less opportunity than othermembers of the electorateto participate in the political process and to elect representatives of their choice." 42 U.S.C. § 1973(b) (emphasis added). The benchmark is thus quite straightforward—under the challenged law or practice, how do minorities fare in their ability "to participate in the political process" as compared to other groups of voters?

Husted, 768 F.3d at 556 (emphasis in original). Of course, vote dilution and vote denial claims derive from the same statutory text. Renodoes not speak to vote dilution claims but by its own terms addresses § 2 claims generally, and Holder v. Hallinterpreted the same statutory text.

First, the minority group must be able to demonstrate that it is sufficiently large and geographically compact to constitute a majority in a single-member district. ... Second, the minority group must be able to show that it is politically cohesive. ... Third, the minority must be able to demonstrate that the white majority votes sufficiently as a bloc to enable it ... usually to defeat the minority's preferred candidate.

478 U.S. at 50–51, 106 S.Ct. 2752. The preconditions require plaintiffs to show, from the outset, whether they have a colorable vote-dilution injury that can be remedied by § 2. Growe v. Emison, 507 U.S. 25, 39–41, 113 S.Ct. 1075, 122 L.Ed.2d 388 (1993) ("Our precedent requires that, to establish a vote-dilution claim with respect to a multimember districting plan (and hence to justify a super-majority districting remedy), a plaintiff must prove three threshold conditions .... Unless these points are established, there neither has been a wrong nor can be a remedy."); Nipper v. Smith, 39 F.3d 1494, 1530, 1533 (11th Cir.1994) (en banc) ("The first Gingles precondition, informed by the second, dictates that the issue of remedy is part of the plaintiff's prima facie case in section 2 vote dilution cases. ... [Holder v. Hall] also confirms that, from the inception of a section 2 case, the existence of a workable remedy within the confines of the state's system of government is critical to the success of a vote dilution claim. The absence of an available remedy is not only relevant at the remedial stage of the litigation, but also precludes, under the totality of the circumstances inquiry, a finding of liability.").

There is a similar problem in this case, particularly insofar as Plaintiffs' theory of the case is that they will be "more equal" if the removal of the voting mechanisms are enjoined. Section 2 calls for "equal opportunity," not "more equal opportunity." This court has no way to assess where "more equal"—but nevertheless allegedly discriminatory—ends and the "equal opportunity" § 2 mandates begins. This is a significant problem given that the scope of any remedy imposed by this court "must be proportional to the scope of the violation, and the order must extend no further than necessary to remedy the violation." Brown v. Plata, 563 U.S. 493, 131 S.Ct. 1910, 1940, 179 L.Ed.2d 969 (2011) ; accordCrawford, 553 U.S. at 203, 128 S.Ct. 1610 ; Freeman v. Pitts, 503 U.S. 467, 489, 112 S.Ct. 1430, 118 L.Ed.2d 108 (1992) ; Milliken v. Bradley, 418 U.S. 717, 750, 94 S.Ct. 3112, 41 L.Ed.2d 1069 (1974) ; Swann v. Charlotte–Mecklenburg Bd. of Ed., 402 U.S. 1, 16, 91 S.Ct. 1267, 28 L.Ed.2d 554 (1971) ; Veasey, 796 F.3d at 518–19 & n. 37 ; Frank, 768 F.3d at 755.

For example, Plaintiffs fail to articulate any measure for how long an early-voting period must run or how many days should be included. If Plaintiffs do not have an equal opportunity with seventeen days, might they need twenty days? If, instead, Plaintiffs have an equal opportunity with seventeen days, is it possible that they still have an equal opportunity with eleven days? What about thirteen or fifteen? As in Holder v. Hall, the number of days of early voting is ultimately arbitrary on this record. If the court were to examine the practices in other States, as Justice O'Connor did in Holder, it would likewise find no consensus at all. (See Def. Ex. 270 at 20, 23.) Plaintiffs can provide no evidence or standard for this court to select a different period. The inquiry under Plaintiffs' theory is inherently standardless.

SDR presents the same issue. As the court could not order North Carolina to provide any more days of SDR than is necessary to remedy the violation, would equality of opportunity be achieved only with SDR on every day of early voting? If so, how many? Under Plaintiffs' theory, seventeen days of SDR would still be insufficient. Indeed, Plaintiffs cite registration itself as the largest "impediment" or "barrier" to voting, (e.g., Doc. 331 at 215), so would Plaintiffs' theory not similarly require North Carolina to eliminate any registration deadline altogether, like North Dakota has done? Or, because the evidence showed that EDR was associated with increased turnout, and Plaintiffs' evidence showed that provisional ballots cast for "no record of registration" on Election Day in 2014 were disproportionately African American, (Pl. Ex. 689), what is the principled basis to preclude requiring North Carolina from adopting EDR under Plaintiffs' theory of § 2? On the present facts, Plaintiffs' theories of recovery are limited only by their present request for relief, not by any principled measurement of equality of opportunity.

OOP voting and pre-registration present similar difficulties under Plaintiffs' theory. Before SL 2013–381, voters could only vote in an unassigned precinct on Election Day if they at least showed up in the correct county. But if the elimination of OOP voting imposed a discriminatory burden, there is no basis for thinking that the county restriction did not also impose a discriminatory burden. Regarding pre-registration, how many extra months of pre-registration are required to provide equal opportunity?

All of these qualitative issues with the relief sought by Plaintiffs also give rise to a serious temporal problem. Absent an identifiable standard for measurement of relief proportional to the violation, how long should an injunction remain in place? Under the old law, African American registration rates already exceeded that of whites, and turnout exceeded that of whites in presidential elections. Under SL 2013–381, African American registration continues to exceed that of all others, and turnout exceeds that of comparable previous elections under the old law. Would the injunction stay in place until all racial socioeconomic disparities disappear? If the Plaintiffs' theory of § 2 is correct, and their evidence in this case suffices for a § 2 violation, then this court would need to issue an injunction for an indefinite period without any articulable basis for measuring its successful attainment. But this court has no authority to do so. Parents Involved in Cmty. Schs. v. Seattle Sch. Dist. No. 1, 551 U.S. 701, 760, 127 S.Ct. 2738, 168 L.Ed.2d 508 (2007) (Thomas, J., concurring); Freeman, 503 U.S. at 495, 112 S.Ct. 1430 ; Bd. of Educ. of Oklahoma City Pub. Schs. v. Dowell, 498 U.S. 237, 249, 111 S.Ct. 630, 112 L.Ed.2d 715 (1991) ; City of Richmond v. J.A. Croson Co., 488 U.S. 469, 498, 109 S.Ct. 706, 102 L.Ed.2d 854 (1989) ; Wygant v. Jackson Bd. of Educ., 476 U.S. 267, 276, 106 S.Ct. 1842, 90 L.Ed.2d 260 (1986) ; Pasadena City Bd. of Ed. v. Spangler, 427 U.S. 424, 436–37, 96 S.Ct. 2697, 49 L.Ed.2d 599 (1976).

During closing arguments, the court reiterated its concerns over all of these questions should Plaintiffs' theory of the case prevail. Counsel for DOJ expressed that this case presents a "unique situation," given that minority turnout and registration rates exceed those of whites. (Doc. 343 at 121.) When the court asked how it would know when to dissolve any injunction it issued, and after much pressing, (id. at 121–29), the best direction counsel advised was: "Well, you're the ultimate[ ] fact finder, so you'll know it when you see it." (Id. at 125.) To date, Plaintiffs have not provided any better guidance, and the court cannot imagine a workable standard under Plaintiffs' theory, either.

In the end, however characterized, Plaintiffs' inability to articulate a principled way for this court to properly redress their alleged injury is highly problematic, not just for this case, but for the fabric of the law. See, e.g., Bartlett v. Strickland, 556 U.S. 1, 17–18, 129 S.Ct. 1231, 173 L.Ed.2d 173 (2009) (noting the "need for workable standards" that produce a "rule [that] draws clear lines for courts and legislatures alike"); Holder, 512 U.S. at 885, 114 S.Ct. 2581 ("With respect to challenges to the size of a governing authority, respondents fail to explain where the search for reasonable alternative benchmarks should begin and end, and they provide no acceptable principles for deciding future cases."). For, while the consideration here is intensely local, it surely must make sense in the scheme of the VRA, which applies nationally.

In grappling with the difficult questions this case presents, the court pressed Plaintiffs on its concern that their theory threatens a "one-way ratchet" voting mechanism that, once enacted, could not practically be removed. Did Congress intend this result? Cf.Nixon v. Mo. Mun. League, 541 U.S. 125, 137, 124 S.Ct. 1555, 158 L.Ed.2d 291 (2004) (eschewing federal creation of a "one-way ratchet"). Federal courts do not normally consider that Congress would pass a statute carrying such burdensome disincentives. SeeTex. Dep't of Hous. & Cmty. Affairs v. Inclusive Cmtys. Project, Inc., ––– U.S. ––––, 135 S.Ct. 2507, 2524, 192 L.Ed.2d 514 (2015) ("If the specter of disparate-impact litigation causes private developers to no longer construct or renovate housing units for low-income individuals, then the FHA would have undermined its own purpose as well as the free-market system. And as to governmental entities, they must not be prevented from achieving legitimate objectives, such as ensuring compliance with health and safety codes."); cf.Palmer v. Thompson, 403 U.S. 217, 227–28, 91 S.Ct. 1940, 29 L.Ed.2d 438 (1971) (Burger, C.J., concurring) ("To hold, as petitioners would have us do, that every public facility or service, once opened, constitutionally 'locks in' the public sponsor so that it may not be dropped ... would plainly discourage the expansion and enlargement of needed services in the long run."). Federal courts have appreciated this problem specifically in cases under the VRA:

The Court also notes that an acceptance of Plaintiffs' argument that a Section 2 violation occurs merely because some counties have more early polling sites would have far-reaching implications. Consider the fact that many states do not engage in any form of early voting. Following Plaintiffs' theory to its next logical step, it would seem that if a state with a higher percentage of registered African–American voters than Florida did not implement an early voting program a Section 2 violation would occur because African–American voters in that state would have less of an opportunity to vote than voters in Florida. It would also follow that a Section 2 violation could occur in Florida if a state with a lower percentage of African–American voters employed an early voting system, as commented on above, that lasts three weeks instead of the two week system currently used in Florida. This simply cannot be the standard for establishing a Section 2 violation.

Hood, 351 F.Supp.2d at 1335–36 ; accordBrown, 895 F.Supp.2d at 1254. Plaintiffs have not provided a satisfactory response to this concern, either. Absent some meaningful measurement of equality, an arbitrary injunction in this case would send a cold wind through legislatures in the multitude of States—like New York, Michigan, and Washington—that presently have none of these conveniences but may wish to consider them.

In the end, these important questions serve to undermine Plaintiffs' results-based theory of recovery under § 2.

B. "Traditional" Fourteenth and Fifteenth Amendment Claims

The Fifteenth Amendment to the United States Constitution provides, "The right of citizens of the United States to vote shall not be denied or abridged by the United States or by any State on account of race, color, or previous condition of servitude." Its prohibition, for purposes of this case at least, patterns an intent-based claim under § 2 of the VRA. See, e.g., 52 U.S.C. § 10301(a) ; Chisom v. Roemer, 501 U.S. 380, 391–93, 111 S.Ct. 2354, 115 L.Ed.2d 348 (1991). Furthermore, under both the Fifteenth Amendment and the Equal Protection Clause, plaintiffs must prove that the State acted with a discriminatory purpose. Reno v. Bossier Parish School Bd., 520 U.S. 471, 481, 117 S.Ct. 1491, 137 L.Ed.2d 730 (1997) ; Voinovich v. Quilter, 507 U.S. 146, 158–59, 113 S.Ct. 1149, 122 L.Ed.2d 500 (1993).

The court has already found that SL 2013–381 was not passed with racially-discriminatory intent. For this reason, these two constitutional claims fail.

C. Anderson–Burdick Claim

Plaintiffs also bring another Fourteenth Amendment claim—which is not based on race or age, but on the right to vote generally—under the Anderson v. Celebrezze, 460 U.S. 780, 103 S.Ct. 1564, 75 L.Ed.2d 547 (1983), and Burdick v. Takushi, 504 U.S. 428, 112 S.Ct. 2059, 119 L.Ed.2d 245 (1992), line of cases.

The right to vote is fundamental, but it does not entail an absolute right to vote in any particular manner. Burdick, 504 U.S. at 433, 112 S.Ct. 2059. "The Constitution provides that States may prescribe '[t]he Times, Places and Manner of holding Elections for Senators and Representatives,' " id.(citing U.S. Const. Art. I, § 4, cl. 1 ), and it is well established that "there must be a substantial regulation of elections if they are to be fair and honest and if some sort of order, rather than chaos, is to accompany the democratic processes," id.(quoting Storer v. Brown, 415 U.S. 724, 730, 94 S.Ct. 1274, 39 L.Ed.2d 714 (1974) ). Accordingly, even though "[e]lection laws will invariably impose some burden upon individual voters," not all burdens are unconstitutional. Id.

Plaintiffs contend SL 2013–381 imposes an unconstitutional burden on the right to vote. Regulations that impose a severe burden on the right to vote must be "narrowly drawn to advance a state interest of compelling importance." Id. at 434, 112 S.Ct. 2059 (quoting Norman v. Reed, 502 U.S. 279, 289, 112 S.Ct. 698, 116 L.Ed.2d 711 (1992) ). Non-severe burdens, by contrast, are subject to Anderson–Burdick's more flexible balancing standard, under which this court must

weigh "the character and magnitude of the asserted injury to the rights protected by the First and Fourteenth Amendments that the plaintiff seeks to vindicate" against "the precise interests put forward by the State as justifications for the burden imposed by its rule," taking into consideration "the extent to which those interests make it necessary to burden the plaintiff's rights."

Id.(quoting Anderson, 460 U.S. at 789, 103 S.Ct. 1564 ). Within Anderson–Burdick's balancing standard, the "rigorousness of our inquiry into the propriety of [a regulation] depends" upon the severity of the burden imposed. Id. Accordingly, Anderson–Burdick balancing operates on a sliding scale: the greater the burden imposed, the more important a State's justification must be. Id.

Where challengers fail to show that a law imposes a "substantial burden on the right to vote, or even represent[s] a significant increase over the usual burdens of voting," Crawford, 553 U.S. at 198, 128 S.Ct. 1610 (per Stevens, J.), a State need only show that relevant and legitimate State interests are "sufficiently weighty to justify the limitation," id. at 191, 128 S.Ct. 1610 (quoting Norman, 502 U.S. at 288–89, 112 S.Ct. 698 ). For example, in Burdick, the Court found Hawaii's legitimate interest in prohibiting write-in voting to outweigh the "limited" and "slight" burden imposed by the restriction. 504 U.S. at 437, 439, 112 S.Ct. 2059.

In addition, "when a state election law provision imposes only 'reasonable, nondiscriminatory restrictions' upon the First and Fourteenth Amendment rights of voters, 'the State's important regulatory interests are generally sufficient to justify' the restrictions." Id.(quoting Anderson, 460 U.S. at 788, 103 S.Ct. 1564 ). The Supreme Court recently illustrated this principle in Crawford.

As discussed above, Crawford considered whether Indiana's voter-ID requirement violated the Equal Protection Clause of the Fourteenth Amendment. 553 U.S. at 185, 128 S.Ct. 1610. Under the challenged law, citizens "voting in person on election day, or casting a ballot in person at the office of the circuit court clerk prior to election day," were required to present a government-issued photo identification. Id. The law applied "to in-person voting at both primary and general elections," but did not "apply to absentee ballots submitted by mail" and contained "an exception for persons living and voting in a state-licensed facility such as a nursing home." Id. at 185–86, 128 S.Ct. 1610. In addition, the law permitted a voter "who has photo identification but is unable to present that identification on election day [to] file a provisional ballot that [would] be counted if [the voter] [brought] her photo identification to the circuit court clerk's office within 10 days." Id. at 186, 128 S.Ct. 1610. Thus, voters who availed themselves of this mechanism were still required to "travel to the circuit court clerk's office within 10 days" of voting to provide proper identification. Id. at 199, 128 S.Ct. 1610. There was no photo identification requirement to register to vote, and free photo ID was offered to those able to establish their residency and identity. Id. at 186, 128 S.Ct. 1610. Challengers claimed that the law was not necessary to avoid election fraud and "substantially burden[ed] the right to vote" by placing an arbitrary and unjustified burden upon qualified voters who did not possess, and could not readily obtain, the necessary identification. Id. at 187, 128 S.Ct. 1610.

Justice Stevens, writing for himself and two other members of the court, found that even though compliance with the law involved the "inconvenience of making a trip to the [Bureau of Motor Vehicles], gathering the required documents, and posing for a photograph," the challengers had failed to show that, as for any class of voters, the law imposed "a substantial burden on the right to vote, or even represent[ed] a significant increase over the usual burdens of voting." Id. at 198, 128 S.Ct. 1610. Justice Scalia, writing for himself and two other members of the Court, would have decided the case on broader grounds,²²⁹ but nevertheless agreed that "[t]he burden of acquiring, possessing, and showing a free photo identification [was] simply not severe, because it d[id] not even represent a significant increase over the usual burdens of voting." Id. at 209, 128 S.Ct. 1610 (Scalia, J., concurring) (characterizing the burden as "minimal"). Accordingly, six members of the Court found Indiana's legitimate interest in "orderly administration and accurate record keeping," including safeguarding voter confidence, to be sufficient to justify the law's burden upon the right to vote. Id. at 192–97, 128 S.Ct. 1610.

229 Justice Scalia viewed Justice Steven's opinion as assuming that the voter identification law " 'may have imposed a special burden on' some voters." Crawford, 553 U.S. at 204, 128 S.Ct. 1610 (Scalia, J., concurring) (quoting 553 U.S. at 199, 128 S.Ct. 1610 (Stevens, J., lead opinion)). Justice Scalia and two other Justices would have considered only the burden the law imposed on voters generally, as opposed to any particular class of voters. Id.at 208, 128 S.Ct. 1610. To the extent that some debate remains as to whether the applicable burden under Anderson–Burdickis to be evaluated for subclasses, this court has assumed as much because, even under that standard, Plaintiffs cannot prevail. SeeHusted, 768 F.3d at 543–45 (discussing the fractured opinion from Crawfordand whether Anderson–Burdickcontemplates subclasses).

Even though the Court stated in Burdick that "[e]lection laws will invariably impose some burden upon individual voters," 504 U.S. at 433, 112 S.Ct. 2059, neither the Supreme Court nor the Fourth Circuit has considered a State's reduction or removal of voting conveniences/"fail-safes." The traditional burden that the Supreme Court has encountered in its Anderson–Burdick jurisprudence has been affirmative(i.e., a burden that places an additional obstacle between voters and voting). SeeCrawford, 553 U.S. at 185, 128 S.Ct. 1610 (requiring voters to have photo identification); Burdick, 504 U.S. at 428, 112 S.Ct. 2059 (requiring candidates to have participated in Hawaii's open primary in order to appear on the general election ballot and prohibiting write-in voting); Anderson, 460 U.S. at 780, 103 S.Ct. 1564 (requiring candidates to meet early filing deadline). The Sixth Circuit, however, has applied Anderson–Burdick in weighing the burdens imposed by the removal of voting conveniences in two cases. SeeHusted, 768 F.3d 524 (applying Anderson–Burdick to find that plaintiffs were likely to succeed on their claim that Ohio's reduction of early-voting days violated their equal protection rights); Obama for America, 697 F.3d at 427, 436 (applying Anderson–Burdick to find that plaintiffs were likely to succeed on their claim that Ohio violated the equal protection clause of the Fourteenth Amendment by permitting certain military voters—but not other voters—to vote during the three days prior to Election Day). In the present case, the Fourth Circuit did not reach the Anderson–Burdick claims.

Assuming, therefore, that the removal of a convenience voting mechanism is the type of burden contemplated by the equal protection clause, the court finds that Plaintiffs have failed to show that SL 2013–381 and SL 2015–103 impose a "substantial burden" on any subclass of voters "or even represent a significant increase over the usual burdens of voting." Crawford, 553 U.S. at 198, 128 S.Ct. 1610. Each provision is addressed in turn.

1. Voter ID

Plaintiffs, other than the United States,²³⁰ claim that North Carolina's voter-ID law imposes a substantial burden on the right to vote. Throughout this litigation, Plaintiffs have frequently claimed that North Carolina's voter-ID requirement is one of the strictest in the country. This may have been a reasonable claim prior to SL 2015–103, but with the enactment of the reasonable impediment exception, it is simply not true today.²³¹

230 The United States makes no claim that North Carolina's current ID law violates the Fourteenth Amendment under the Anderson–Burdick line of cases. (Doc. 419 at 77.)

231 Intervenor Plaintiffs (representing "young voters") opted not to participate in this court's January 2016 trial on voter ID, resting instead on the intent-related ID evidence presented at the July 2015 trial. Accordingly, they presented no evidence—and thus waived any claim—that young voters will be unlawfully burdened by North Carolina's voter-ID requirement now that it has a reasonable impediment exception. Nevertheless, this court's independent analysis confirms that this will not be the case. Facially, out-of-State college students might seem more at risk because they are unlikely to have a North Carolina driver's license. But those who are not residents for voting purposes, seeN.C. Gen. Stat. § 163–57(11), are not eligible to vote in the State anyway and should vote in their State of residence by options available there. Moreover, those who are North Carolina residents for voting purposes retain several low-burden means of complying with the ID requirement. If they are new to the State (e.g., a student enrolling in the fall semester), they will be able to use their out-of-State driver's license for the November Presidential election because they will have registered to vote within ninety days of the election. N.C. Gen. Stat. § 163–166.13(e)(8). The same will be true for students who delay registration until the twenty-five day cut-off. Id.Once students' out-of-State licenses cease to be acceptable for voting purposes, those who do not intend to drive and lack otherwise qualifying ID (passport, etc.) are eligible for a no-fee voter ID and can obtain one for free at the DMV. Those who intend to drive must obtain a North Carolina driver's license within sixty days of becoming a resident anyway. N.C. Gen. Stat. § 20–7(a). Most importantly, the reasonable impediment exception is available for those who have any impediment to acquiring acceptable ID (e.g., students who live on campus and do not have transportation), including young voters who are not college students. Further, no one has suggested that college students lack the literacy required to vote under the reasonable impediment exception.

SL 2013–381, as amended, is clearly less burdensome than Indiana's ID requirement upheld by the Supreme Court in Crawford, Wisconsin's ID requirement upheld by the Seventh Circuit in Frank, or Georgia's ID requirement upheld by the Eleventh Circuit in Billups. SeeCrawford, 553 U.S. at 198, 128 S.Ct. 1610 ; Frank, 768 F.3d at 747 ; Billups, 554 F.3d at 1354–55. It is also less burdensome than the Texas ID requirement that the Fifth Circuit struck down on § 2 grounds in Veasey, 796 F.3d at 513 (petition for rehearing en banc granted March 9, 2016). And none of these States provided voters with a reasonable impediment exception.²³²SeeSouth Carolina, 898 F.Supp.2d at 46–48. The only State to do so is South Carolina, whose law was found by a three-judge panel not to impose a "material burden" on minorities' right to vote. Id. at 41. North Carolina's reasonable impediment exception, which effectively codifies the South Carolina law approved by the court, is therefore not unique because it is strict; it is unique because it offers socioeconomically-disadvantaged individuals without qualifying ID a significant voter accommodation that few other States with ID requirements provide. In light of the reasonable impediment exception, and for the reasons extensively stated above, this court cannot say that North Carolina's voter-ID requirement imposes a "substantial burden" on any subclass of voters "or even represent[s] a significant increase over the usual burdens of voting." Crawford, 553 U.S. at 198, 128 S.Ct. 1610.

232 Indiana's law contained an exception for "indigent" voters. Crawford, 553 U.S. at 216, 128 S.Ct. 1610 (Souter, J., dissenting). However, this exception was much narrower than North Carolina's reasonable impediment exception. To vote provisionally under Indiana's indigency exception, a voter had to "(at the circuit court clerk's office) sign an affidavit affirming that she [was] 'indigent' and 'unable to obtain proof of identification without the payment of a fee.' " Id.at 216 & n. 19, 128 S.Ct. 1610 (citing Ind. Code. Ann. § 3–11.7–5–2.5(c)(2)(A) ).

Moreover, despite accommodating those expressing genuine difficulties in acquiring photo ID, North Carolina's voter-ID requirement still serves legitimate State interests articulated in Crawford. See supra Part II.A.2.j.i. As noted above, SL 2013–381 and SL 2015–103 had the effect of raising the security level of both absentee mail voting and in-person voting. See supra Part II.A.2.j.i. This court finds any alleged diminution in achieving the State's purported interest to be more than offset by the reduction of burden achieved by the reasonable impediment exception. Accordingly, North Carolina's voter-ID requirement serves legitimate State interests that are "sufficiently weighty to justify the limitation." Crawford, 553 U.S. at 191, 198, 128 S.Ct. 1610.

2. Early Voting

Plaintiffs contend that voters have become habituated to seventeen days of early voting and that decreasing early voting to ten days imposes a "substantial" burden by exacerbating congestion, increasing wait times, and decreasing flexibility for those with inflexible schedules and responsibilities. (Doc. 346 at 141.)

The evidence in this case contradicts Plaintiffs' assertions. First, as discussed above, Plaintiffs failed to put forth credible evidence that SL 2013–381 produced congestion or increased wait times during the 2014 midterm election, or that these results will occur in future elections. This is largely because of the same-hours requirement. This does not mean that there may not be lines, as early voting use has been proven to fluctuate with natural peak use times irrespective of the schedule.

Second, there is no evidence that North Carolina's introduction of early voting or use of seventeen days of early voting caused increased political participation either overall or for any subclass. In fact, the scholarly consensus is that early voting tends to depress voter turnout.

Third, due to the interaction between SL 2013–381's same-hours requirement and the reduction in early-voting days, CBOEs are forced to either open more early-voting sites or keep existing sites open for longer hours, including expanding weekend hours. The 2014 midterm illustrated this interaction, as high-convenience night and weekend voting hours actually increased when SL 2013–381 was implemented. (See, e.g., Pl. Ex. 242 at 167—68.) The same-hours requirement can only be waived by a unanimous vote of a bipartisan CBOE and bipartisan SBOE, which in total requires agreement among five members of the majority party and three members of the minority party; a single dissent will veto a waiver. (See Doc. 340 at 204–05.) Thus, GOTV efforts will have the same availability of early-voting hours as before.

Fourth, the evidence showed that, even when seventeen days of early voting could be offered, it was the last ten days that were most heavily used and most offered. (See Def. Ex. 362 at 1–3; Doc. 338 at 134–41.) Further, not all of the seven days of early voting that were eliminated were actually used. CBOEs were not required to offer weekend hours during the first seven days, (Doc. 340 at 197–98), and no county offered early voting on the first Sunday in 2010, (see Doc 126–4 at 45–90).

Fifth, under the new ten-day early-voting schedule, African American use of early in-person voting increased, turnout for both African Americans and whites increased, and the disparity in turnout between African American and white voters decreased.²³³ (Def. Ex. 309 at 59–62, 68—69.) Though young voters' use of early voting decreased in 2014, (Pl. Ex. 236 at 22–23.), young voter turnout still increased, (Def. Ex. 309 at 78). Young voters are disproportionately more likely to put off voting until later in the election period.²³⁴ (Pl. Ex. 236 at 22.) In any case, there is no evidence that young voters are disproportionately disadvantaged in any way that would affect their ability to vote within the provided ten days of early voting.

233 Plaintiffs do not claim that Hispanics disproportionately used the removed times of early voting. (Pl. Ex. 346 at 24–27.)

234 The evidence was that young voters were disproportionately likely to vote during early voting after 1 p.m. on the final day of early voting, which was removed by SL 2013–381. (Pl. Ex. 236 at 21–22.) Plaintiffs do not claim, however, that SL 2013–381's changes to the early-voting schedule fall more heavily on young voters than African Americans.

Finally, Plaintiffs failed to show that any voters are habituated to early voting. Instead, the evidence showed that voters were able adapt to the new schedule. In fact, those who voted during the first seven days of early voting in 2012 were more likely to vote in 2014 than those who voted in the last ten days.

Plaintiffs also argue that the new schedule eliminates a Sunday of voting, which was important for GOTV efforts. This is true, but insofar as the schedule maintains one Sunday of voting and increases night and weekend availability, the court cannot say that this results in an impermissible burden on those efforts.²³⁵

235 It is highly questionable whether the State has any obligation to provide for voting on Sunday, a day historically regarded as a non-work day. Because the parties have not addressed that issue, the court need not do so, either.

For all these reasons, and those laid out elsewhere in this opinion, Plaintiffs have failed to show that the changes to the early-voting schedule impose a "substantial burden" on any subclass of voters "or even represent a [n] ... increase over the usual burdens of voting," Crawford, 553 U.S. at 198, 128 S.Ct. 1610.

As discussed above, the changes to the early-voting schedule (with the same-hours amendment) were intended to make early voting more accessible and ensure that different parts of a county are more equally served. By removing seven days of early voting but requiring that the same number of aggregate early-voting hours be offered, SL 2013–381 furthers the State's relevant and substantial interests by requiring CBOEs to either open more early-voting sites or keep existing sites open for more hours, including expanding weekend hours. If a CBOE does the former, different parts of a county are more equally served. If it does the latter, voters are provided voting hours that are more convenient for typical work schedules, especially those of lower economic means. Both appear to have occurred in the 2014 midterm, when both more locations and more night and weekend hours were offered. In other words, SL 2013–381 effectively reallocated the first seven days under the former schedule, which were the least used days, into times that are more accessible to voters.

Accordingly, this court finds the reduction in early-voting days to be tailored to achieving the State's relevant and substantial interest in producing a more efficient and fair early-voting process. In addition, given that Plaintiffs have failed to show that the changes to the early-voting schedule impose a "substantial burden" on any subclass of voters "or even represent a[n] ... increase over the usual burdens of voting," this court finds the State's precise interest to be "sufficiently weighty to justify the limitation." Crawford, 553 U.S. at 191, 198, 128 S.Ct. 1610.

3. SDR

Plaintiffs contend that the removal of SDR imposes a "severe" burden by removing a "fail-safe" necessary to avoid the disenfranchisement of thousands of transient, low-income, and young voters.²³⁶ (Doc. 346 at 141.)

236 Plaintiffs also produced several witnesses who testified that they believed they registered at the DMV leading up to the 2014 election only to learn that they were not registered when they tried to vote. (Doc. 346 at 79–80.) If the DMV failed to successfully register these persons, it would be highly troubling because the NVRA requires North Carolina's DMV to do so. In any case, SL 2013–381 did not cause the DMV's errors. Moreover, even if the DMV's registration errors affect the burden imposed by SDR's removal, Plaintiffs presented no evidence that these errors were anything but random. In fact, one of Plaintiffs' key arguments about the sufficiency of the DMV as a registration option is that minorities are less likely to seek DMV services. As such, any burden must be evaluated as to voters generally. In Wake County during the 2014 election, approximately 250 voters claimed they had attempted to register at the DMV, even though there was no record of their registration at the voting location. (Pl. Ex. 817 at 122–24.) But even if each voter in fact attempted to register at the DMV, and assuming that each voter's registration problem was attributable to DMV error, this would only account for .87% of the roughly 28,800 voter registrations transmitted from the DMV to Wake County each year. (Pl. Ex. 800 at 152–53 (noting 120 per day for 240 days).) Accordingly, this court is unable to find that the DMV's errors impose a substantial burden on voters generally or any definable subclass of voters. In any case, the burden is outweighed by the State's asserted interest in the removal of SDR.

As discussed above, North Carolina provides a myriad of ways for voters to comply with the registration requirement prior to the cut-off (which is five days more generous than the NVRA's thirty-day requirement), most of which permit assistance. See supra Part II.A.3.c. Those who avail themselves of North Carolina's many registration opportunities at least twenty-five days before Election Day have the opportunity to vote either during the ten days of early voting (at any early-voting site), on Election Day (at their correct precinct unless they are an unreported mover), or through the State's mail-in absentee ballot process. The ease of registering and voting in North Carolina without SDR is bolstered by the fact that, even after SDR was eliminated, African American and young voter²³⁷ registration and turnout rates increased. (See Pl. Ex. 684; Pl. Ex. 229 at 7; Def. Ex. 309 at 66, 78; Def. Ex. 344.) Hispanic²³⁸ turnout rates also increased at a higher rate than white voters from 2010 to 2014. (Def. Ex. 309 at 61–66 (tbls. 4–8).) Further, the fact that the vast majority of African American voters have successfully registered without SDR (even when it was available) demonstrates that the State's registration opportunities do not impose "severe" or even "substantial" burdens on any subclass of voters "or even represent ... a[n] increase over the usual burdens of voting." Crawford, 553 U.S. at 198, 128 S.Ct. 1610 ; Marston, 410 U.S. at 681, 93 S.Ct. 1211 (endorsing State's voter registration cut-off where justified as "necessary to promote the State's important interest in accurate voter lists").

237 Plaintiffs showed that young voters disproportionately used SDR when it was in place. (Pl. Ex. 236 at 7, 15–16.) Although young voters are disproportionately African American and Hispanic, Plaintiffs have presented no evidence that young voters experience socioeconomic disparities that make SDR's removal fall more heavily on them than African Americans or Hispanics generally.

238 Plaintiffs presented evidence that Hispanics were more likely to use SDR when it was in place. (Pl. Ex. 245 at 23.) As with the other provisions in this case, Plaintiffs' focus was on African Americans. Plaintiffs do not claim that SDR's removal imposes a heavier burden on Hispanics than African Americans.

By contrast, SDR's removal serves relevant and sufficiently substantial State interests by ameliorating several significant problems. First, and most importantly, SDR's close proximity to Election Day increased the risk (over traditional registrants) that individuals would have their vote counted despite later failing mail verification. See supra Part II.A.2.j.iii. In 2012, 2,361 votes were counted despite the registrant later failing mail verification, not because of insufficient staffing, but because SDR extended the registration date too close to Election Day to give the verification process an opportunity to work. Moreover, by making registration and voting contemporaneous, SDR negated the State's ability to cancel registrations and rendered N.C. Gen. Stat. § 163–82.7(f) a ity. With SDR removed, the number of late registrants who vote but thereafter fail mail verification will be reduced. Second, SDR created administrative burdens on CBOEs that were more than merely imaginable. See supra Part II.A.2.j.iii. For example, there were 252,684 SDR-registrations in 2008. (Pl. Ex. 40 at 35 (Ex. 14).) The SBOE executive director's post-2008 analysis indicates that, despite "hir[ing] additional staff" and "working as efficiently as they could" for "long hours and workweeks," CBOEs found it "not possible" to meet the statutory forty-eight hour deadline "due to the volume" of SDR-registrations. (Pl. Ex. 56 at 5.) SDR also burdened CBOEs by requiring them to research and process "intersecting registrations" "on a case by case basis to ensure that there was no possible fraud being committed." (Id. at 7); see also supra Part II.A.2.j.iii.

For these reasons, this case is significantly different from Husted. There, the State was unable to identify any real burden in having five days of ordinary registration overlap with the first five days of the thirty-five day early-voting period ("Golden Week"), or any benefit to the State from eliminating those five days overlap because the State still had thirty days to verify SDR registrants. Husted, 768 F.3d at 547 (finding that "the difficulties of verifying voter registration before counting ballots did not clearly pertain to problems with Golden Week specifically").

Given that Plaintiffs have failed to show that the removal of SDR imposes a "substantial burden" on any subclass of voters "or even represent[s] a[n] ... increase over the usual burdens of voting," this court finds the State's precise interests furthered by the removal of SDR to be "sufficiently weighty to justify the limitation." Crawford, 553 U.S. at 191, 198, 128 S.Ct. 1610.

4. OOP

Plaintiffs contend that the removal of OOP imposes a severe burden by disenfranchising those who attempt to vote outside of their precinct on Election Day. (Doc. 346 at 142.) Because OOP provisional ballots have never constituted more than a fraction of a percentage point of votes cast, and voting in the correct precinct merely requires one to become informed of his or her correct precinct and travel to it within the county, the burden (if any) to the general voting population from the removal of OOP is very slight. Insofar as North Carolina has employed a precinct-based system for more than a century, in no case is it "substantial," nor does it "represent a[n] ... increase over the usual burdens of voting." Crawford, 553 U.S. at 198, 128 S.Ct. 1610. Nevertheless, Plaintiffs contend that it imposes a more substantial burden on certain subgroups who disproportionately lack access to reliable transportation.

Plaintiffs have presented evidence that African Americans are disproportionately more likely to be socioeconomically disadvantaged and to have less stable access to transportation.²³⁹ In addition, Plaintiffs demonstrated that African Americans are disproportionately more likely to attempt to vote out of their precinct than whites. (Pl. Ex. 42 (Ex. 49).) Nevertheless, the extent of the burden claimed in getting to an assigned precinct must be evaluated in light of the fact that OOP ballots have never constituted more than a fraction of a single percentage point of ballots cast by African Americans. (Id.; Pl. Ex. 242 at 161 (App'x U).) In addition, of the small sliver of African American voters who voted in the wrong precinct, the evidence showed that the majority (60.3%) voted at a precinct within five miles of their assigned precinct. (Def. Ex. 212A at 18 & tbl. 23.)

239 Again, the court assumes, without deciding, that burdens on subclasses are cognizable. Plaintiffs also showed that Hispanics and young voters disproportionately used OOP when it was in place. Seesupranote 117 (Hispanic and young voter SDR data). But, as noted above, Hispanics are more likely to have access to a vehicle than African Americans, (Pl. Ex. 45 at 13–14), and Plaintiffs do not claim that OOP's removal imposes a heavier burden on Hispanics than African Americans. As for young voters, although they are disproportionately African American and Hispanic, Plaintiffs have presented no evidence that young voters experience socioeconomic disparities that make OOP's removal fall more heavily on them than African Americans and Hispanics generally.

Further, a benefit of a precinct-based system is, as recognized in James, that it generally places voting locations closer to voters. The fact that Plaintiffs' evidence fell far short of establishing that OOP voters' correct precinct was not in fact the closest one to their address of registration supports the conclusion that the precinct system is serving the benefits outlined in James. In fact, only two witnesses (a married couple) testified that their correct precinct was not the closest precinct to their residence. But the disabilities of these witnesses were such that even voting at the closest, albeit incorrect, precinct was nearly an unmanageable burden. More than anything, these witnesses demonstrated the need for mail voting, which North Carolina provides.²⁴⁰

240 The evidence demonstrated that North Carolina has one of the longest absentee mail ballot periods in the Nation. (Def. Ex. 270 at 35.)

As to Plaintiffs' other fact witnesses, even though OOP voters must be registered and every registered voter is mailed a registration card that provides his or her correct precinct, it was apparent that many such voters had never attempted to check their correct precinct. (See, e.g., Doc. 330 at 176–77.) Of course, voters who determine their correct precinct will never have to travel from the wrong precinct to the correct precinct. Nevertheless, the evidence does not support Plaintiffs' claims even for those who first appeared at an incorrect precinct. As discussed above, Plaintiffs presented the deposition testimony of a poll observer who, in attempting to measure the impact of SL 2013–381, collected the names of fifty-nine individuals who left a precinct because they were assigned to vote in another precinct. (Pl. Ex. 796.) However, Defendants' follow-up revealed that, of the fifty-two who could be identified as registered voters, forty-nine ultimately voted on Election Day. (Def. Ex. 343.) Accordingly, 94.2% were able to travel from the wrong precinct to the right one on Election Day. Only three of the fifty-two never successfully voted; two were white, and the other was African American who does not appear to have been eligible to vote anyway. (Id. at 132–34, 140–43.) This evidence undermines Plaintiffs' contention that requiring voters to vote in the correct precinct imposes a substantial burden.

Accordingly, while the removal of OOP requires voters to travel to their correct precinct, the incredibly small number of voters (including African Americans) affected, the inherent proximity of their correct precinct, and, though not dispositive, African American's continued rise in turnout since the removal of OOP, suggest that the burden imposed on them as a subclass is not "substantial" and does not "represent a[n] ... increase over the usual burdens of voting." Crawford, 553 U.S. at 191, 128 S.Ct. 1610.

As discussed above, OOP's removal served relevant and substantial State interests that have been well-established and thoroughly laid out by both the North Carolina Supreme Court and the Sixth Circuit. See supra Part II.A.2.j.iv. The precinct system caps the number of voters attempting to vote in one place on Election Day, allows each precinct to prepare a ballot that all voters who vote at that precinct will be eligible to complete in full, makes ballots less confusing by only including those candidates for whom a citizen may vote, makes it easier to monitor votes and prevent fraud, and generally results in polling places being closer to voter residences. James, 359 N.C. at 270–71, 607 S.E.2d at 644–45 ; Sandusky Cty. Democratic Party, 387 F.3d at 569.

Although OOP voting may have increased the proximity of the nearest Election Day voting location for some voters, it did so only by undermining many of the other advantages of a precinct-based system. First, it removed the cap on the number of voters attempting to vote in one place on Election Day, and the evidence shows that political organizations transported voters to precincts without making any determination of their correct precinct. (Pl. Ex. 9 at 5; Pl. Ex. 811 at 46.) Second, OOP voting produced significant administrative burdens. See supra Part II.A.2.j.iv. Finally, OOP partially disenfranchised voters. See supra Part II.A.2.j.iv. Although OOP voters may have thought that there was no difference between going to their correct precinct and a different precinct with OOP in place, and that all of their provisional votes would count, this was not the case. Thus, as laid out in James, OOP made voting more confusing because OOP voters were not actually eligible to vote for all of the races on the provisional ballot they were provided. There were also races omitted from the provisional ballot that OOP voters would have been eligible to vote for if they had presented at their correct precinct. OOP's removal significantly ameliorated each of these problems and restored the advantages of a precinct system outlined in James.

Given that Plaintiffs have failed to show that the removal of OOP imposes a "substantial burden" on any subclass of voters "or even represent[s] a[n] ... increase over the usual burdens of voting," this court finds the State's precise interests furthered by the removal of OOP to be "sufficiently weighty to justify the limitation." Crawford, 553 U.S. at 191, 198, 128 S.Ct. 1610.

5. Pre-registration

Plaintiffs contend that the removal of pre-registration imposes a "severe" burden on young voters by increasing confusion and forcing them to register in a more burdensome manner. (Doc. 346 at 142.) Plaintiffs also claim that, without pre-registration, youth registration and turnout will decrease, particularly among African American and Hispanic citizens. (Id.)

In support of their claim, Plaintiffs have demonstrated that pre-registration increases youth turnout, albeit in a race-neutral manner.²⁴¹See supra Part II.A.3.e. Plaintiffs' expert, Dr. Hillygus, predicts that 8,000 to 50,000 fewer young voters will vote in 2016 than would have if pre-registration remained in place.

241 Plaintiffs' contention that minority registration and turnout rates will fall for young minorities without pre-registration is not supported by the evidence. African American registration rates reached parity with white registration rates in 2008, before pre-registration was adopted, and continued to climb in 2014, after pre-registration was eliminated. (Pl. Ex. 684.) African American turnout rates also increased in 2014. (Def. Ex. 309 at 59–62, 68–69.)

First, because pre-registration's removal did not eliminate a registration opportunity for anyone who is actually eligible to vote, this court doubts whether the elimination of pre-registration imposes a cognizable burden on the right to vote. Plaintiffs have provided no authority for the proposition that the Constitution's prohibition of certain unjustified burdens on the right to vote extends to burdens imposed on those who are not eligible to vote.

But assuming cognizable burdens extend beyond eligible voters, this court cannot say that registering and voting is materially more difficult for young voters without pre-registration. Even assuming that pre-registration at a mandatory high school voter registration drive is the most convenient way for young North Carolinians to register, as set forth above, North Carolina offers a multitude of highly convenient ways to register. See supra Part II.A.3.c. Young North Carolinians who are unable to pre-register at sixteen due the removal of pre-registration but who wish to do so if they will be eighteen before the next election need look no further than their own high school to find all the materials that are required. SeeN.C. Gen. Stat § 163–82.23 ("Every public high school shall make available to its students and others who are eligible to register to vote [voter registration application forms], and shall keep a sufficient supply of the forms so that they are always available."). Even without pre-registration, North Carolinians who are seventeen-years-old but will be eighteen by the time of the general election can begin to take advantage of these ample registration opportunities as early as sixty days before the primary preceding the general election. N.C. Gen. Stat. § 163–59. In addition, young individuals could not disregard these ample alternative means of registration simply because they pre-registered. In fact, even when pre-registration was in place, pre-registrants who moved residences between counties prior to voting, such as college students, needed to use these alternative methods of registration to re-register. Simply put, the removal of pre-registration does not hinder young North Carolinians from using any of the remaining methods of registration and voting that this court has found to be sufficient for other groups; it merely requires that they wait until they will be eligible to vote in the next election.

For these reasons, to the extent that the removal of pre-registration imposes a burden upon young voters that is cognizable under Anderson–Burdick, that burden is extremely slight. In no event is the burden that young North Carolinians face to register any greater than the "usual burdens of voting." Crawford, 553 U.S. at 198, 128 S.Ct. 1610. It is merely the same burden that every eligible voter faces.

As noted above, the State's interest for pre-registration—reducing confusion—was far weaker than for the changes to early voting and the removal of SDR and OOP. In fact, for the registrant, the removal of pre-registration makes the issue of when he or she is eligible to register marginally more complicated. Nevertheless, there was evidence of voter confusion with pre-registration, namely that some pre-registrants reasonably believed that they would receive a voter registration card after pre-registering. Senator Rucho also claimed during legislative debate of SL 2013–381 that he and his son experienced confusion with pre-registration. See supra Part II.A.2.J.v.Moreover, the State has a legitimate interest in establishing a minimum age for the registration of new voters, so long as the Twenty-Sixth Amendment is not violated. If the State does not have a legitimate interest in tying registration to when individuals become eligible to vote, then it is difficult to imagine any line that could properly be drawn. If sixteen-year-olds cannot legitimately be prohibited from registering, then why can fifteen-year-olds or fourteen-year-olds? Surely the Constitution permits the State to draw some lines, especially where the burden imposed is not material and the age of registration the State chooses is tied to when individuals actually will be eligible to vote in the next general election. See, e.g., Marston, 410 U.S. at 681, 93 S.Ct. 1211 ("The Constitution is not so rigid that that determination [fifty- versus thirty-day registration deadline] and others like it may not stand.").

Therefore, since the burden on the prospective pre-registrants is extremely slight or, more likely, non-existent (since they are not eligible voters), the State's interests are sufficient to overcome this "burden."

6. CBOE Discretion

Plaintiffs contend that the removal of CBOE discretion to keep polling locations open for an extra hour on Election Day burdens "some voters," but they do not characterize the burden's nature. (Doc. 346 at 142.) Plaintiffs do not specify who these voters are likely to be, nor do they allege that they share any characteristic that would make it more difficult for them to vote during normal Election Day hours (or for that matter during the ten days of early voting or by an absentee mail-in ballot). In fact, what Plaintiffs characterize as a loss of voting time is actually a reallocation of decision making to the SBOE. Even after SL 2013–381, if there is an interruption in voting for more than fifteen minutes, the SBOE can extend hours for the length of the interruption. N.C. Gen. Stat. § 163–166.01. In addition, anyone that is in line at the time the polls close is allowed to vote regardless of how long it takes. In other words, both before SL 2013–381 and after, there is a mechanism to accommodate voters. Session Law 2013–381 simply places this mechanism with the SBOE. Plaintiffs did not present any evidence that anyone was burdened by this transfer of discretion. Following a full trial, it is insufficient for Plaintiffs to speculate as to the effect of an election change.

The State has a legitimate interest in promoting uniformity in the hours of voting offered. North Carolina's decision to place decision making-authority with the SBOE, rather than 100 CBOEs, is tailored to achieving this interest. Even if SL 2013–381's reallocation of decision-making authority imposed the slightest of burdens upon voters, which it does not, the State's interest is sufficient to overcome it.

7. Poll Observers and Challengers

Plaintiffs also challenge SL 2013–381's expansion of poll observers and challengers as an unjustified burden on the right to vote. (Doc. 384 in case 1:13cv658 at 5, 34.)

Plaintiffs do not challenge the use of poll observers or challengers generally; they instead argue that the law's expansion of them burdens voters. Any use of a poll observer to intimidate a voter is unlawful and will not be tolerated. But Plaintiffs' concerns as to future fears remain speculative, even after trial. They have presented no evidence that any observer or challenger would or did abuse his statutory power or that observers or challengers imposed any burden on voters in 2014. As such, no State justification is needed to sustain the poll observer provision. SeeMcLaughlin v. North Carolina Bd. of Elections, 65 F.3d 1215, 1221 n. 6 (4th Cir.1995) ("If a regulation imposes no burdens, it would not fail the balancing test even though it also served no discernible state interest.").

8. Cumulative Effect of Provisions

The court now considers the cumulative effect of the challenged provisions. See, e.g., Pisano v. Strach, 743 F.3d 927, 933 (4th Cir.2014) ("[W]e evaluate the combined effect of the state's ballot-access regulations."); Wood v. Meadows, 207 F.3d 708, 713 (4th Cir.2000). As discussed above in this section and for the same reasons discussed more fully in the VRA analysis, Plaintiffs failed to show that the changes brought about by SL 2013–381 impose cognizable burdens that outweigh the State's asserted interests.

The cumulative effect of the challenged provisions is no more than slight to modest. Even though SL 2013–381 added a voter-ID requirement (with reasonable impediment exception) and removed or amended more than one measure, the measures had provided exceptions to traditional and long-standing voting rules in the State. North Carolinians who wish to register and vote still have many convenient ways that provide ample opportunity to do so. North Carolinians can submit an absentee mail-in ballot between forty-five and sixty days before Election Day (depending on the election), vote during the ten days of early voting during expanded hours, or vote on Election Day. Although North Carolinians must now register by twenty-five days before the election, registration is easy, open all year, and can be accomplished by methods that remain plentiful. That this is so is reflected by the increase in registration and turnout rates seen in the 2014 primary and general elections. And, while Plaintiffs characterize North Carolina as unique in altering its election law provisions in the same bill,²⁴² and while the court's appraisal has been local, it is notable that the State still compares very favorably to most States—even formerly § 5–covered jurisdictions like New York, Alabama, and Mississippi—that today have significantly less progressive voting rules.

242 This is clearly not the case, as Florida's reduction in early voting was part of eighty sets of changes, four of which were initially challenged. SeeFlorida, 885 F.Supp.2d at 302–07 (addressing challenges to early-voting reduction of days and inter-county movers).

In sum, Plaintiffs have failed to show that any North Carolinian who wishes to vote faces anything other than the "usual burdens of voting." Crawford, 553 U.S. at 198, 128 S.Ct. 1610. The State, by contrast, has shown that SL 2013–381 furthers several legitimate State interests. Accordingly, this court finds the State's interests to outweigh the, at most, slight to modest burden that the provisions cumulatively impose upon the right to vote.

D. Twenty-Sixth Amendment Claim

Finally, Intervenor Plaintiffs, "young voters," argue that the voter-ID requirement, the transfer of discretion to extend early-voting hours, the change in the early-voting schedule, the elimination of SDR, OOP, pre-registration, and mandatory high school registration drives all violate the Twenty-Sixth Amendment to the United States Constitution by "intentionally burden[ing] the ability of young people ... to register and to vote." (Doc. 285 at 55.) Intervenors, all eighteen and older, do not bring their claims on behalf of a class of voters; they bring them on their own behalf, only. Intervenors' claims are novel and appeared murky throughout this litigation. Trial did not make them less so.

The Twenty-Sixth Amendment provides: "The right of citizens of the United States, who are 18 years of age or older, to vote, shall not be denied or abridged by the United States or any state on account of age." The amendment is modeled after the Fifteenth and Nineteenth Amendments. Pub. Law. 89–110, S. Rep. No. 92–26 at 2, 1st Sess. (1971). In this respect, the court considers the amendment to proscribe both the denial and abridgement of the right to vote.

Few cases have considered the Twenty-Sixth Amendment. Plaintiffs rely on Walgren v. Board of Selectmen of Town of Amherst, 519 F.2d 1364 (1st Cir.1975), where a class of eighteen-to twenty-year-old college students sought to invalidate a 1973 town election held during a winter recess, when some 10,000 students were away, on grounds of equal protection and the Twenty-Sixth Amendment. Id. at 1364–65. After a trial on the merits, the district court dismissed the case, finding plaintiffs' claims of disproportionate burden speculative. Id. at 1367. In a prior opinion, the circuit court had "opine [d] that 'it seems only sensible that if a condition, not insignificant, disproportionately affects the voting rights of citizens specifically protected by [the Twenty-Sixth Amendment], the burden must shift to the governmental unit to show how the statutory scheme effectuates, in the least drastic way, some compelling governmental objective.' " Id.(quoting Walgren v. Howes, 482 F.2d 95, 102 (1st Cir.1973) (emphasis added)). The district court purported to apply this test in considering the Twenty-Sixth Amendment claims. Id. Nevertheless, rather than establish that this is the correct test under the Twenty-Sixth Amendment, the circuit court avoided the issue by finding that the State was entitled to prevail even under a "rigorous" standard of review. Id. at 1367–68 (reiterating that it remained "without the assistance of any precedents ... in evaluating the impact of the Twenty-Sixth Amendment" and that it "need not now face this issue"). As such, the court affirmed on the ground that the shortened time frame within which the town selectmen had to make an admittedly novel decision about setting the election date on the heels of the passage of the Twenty-Sixth Amendment justified the town's action. Id. at 1368.

Defendants cite Gaunt v. Brown, 341 F.Supp. 1187 (S.D.Ohio 1972). In Gaunt, plaintiffs challenged a State law that prohibited seventeen-year-olds who would have been eighteen by the time of the general election from participating in the primary for that general election since they would not be eighteen by the time of the primary. Id. at 1188. A three-judge district court dismissed the claims, finding plaintiffs' argument amounted to little more than that "the State, in reforming its laws to bring it into line with the Twenty-sixth Amendment, has not gone far enough." Id. at 1192. The court stated that under the Elections Clause and Tenth Amendment, the State had a "clear" right to limit plaintiffs' right to vote in primaries, subject to constitutional limitations. Id. at 1189. The court rejected calls for a " 'compelling interest' standard ... because no State could demonstrate a 'compelling interest' in drawing the line with respect to age at one point rather than another." Id. The court observed that the Twenty-Sixth Amendment "does not grant the right to vote to 18–year-olds and was not intended to. It simply bans age qualifications above 18." Id. at 1191. Accordingly, the court found "nothing to indicate an intent [by Congress] to interfere with state action fixing at 18 the minimum age limit for participating in such [primary] nominating process." Id. So, finding that the Twenty-Sixth Amendment was "in no way intended to 'offend our federalism,' " and unable to locate any case where a State was "called on to justify its drawing the line for qualifications at 18 years of age," the court rejected plaintiffs' claim. Id. at 1192. The Supreme Court affirmed by judgment without any opinion. 409 U.S. 809, 93 S.Ct. 69, 34 L.Ed.2d 71 (1972).

Accordingly this court faces a dearth of guidance on what test applies to Twenty- Sixth Amendment claims. Nashville Student Org. Comm. v. Hargett, 155 F.Supp.3d 749, 756–57, No. 3:15cv00210, 2015 WL 9307284, at *6 (M.D.Tenn. Dec. 21, 2015) ("[T]here is no controlling caselaw ... regarding the proper interpretation of the Twenty-Sixth Amendment or the standard to be used in deciding claims for Twenty-Sixth Amendment violations based on an alleged abridgment or denial of the right to vote.") It is far from clear that the Twenty-Sixth Amendment was intended to encompass the removal of the voting conveniences challenged in this case. Nevertheless, even assuming that Plaintiffs' theory of the Twenty-Sixth Amendment is correct—and it is effectively the Fifteenth Amendment but with young voters as the relevant class—Plaintiffs have failed to establish that SL 2013–381 was intended to discriminate against young voters. This alone is fatal to Plaintiffs' claim because, even under the Fifteenth Amendment, plaintiffs must prove that the State acted with a discriminatory purpose. Reno, 520 U.S. at 481, 117 S.Ct. 1491 ; Voinovich, 507 U.S. at 158–59, 113 S.Ct. 1149.

Plaintiffs presented evidence that young North Carolina voters were more likely than older voters to use SDR and OOP when they were in place. (Pl. Ex. 236 at 7–8, 35); supra note 117. Young voters were also more likely to vote after 1 p.m. on the final day of early voting, and HB 589 removed CBOE discretion to extend early voting from 1 p.m. to 5 p.m. on the final day of early voting. (Pl. Ex. 236 at 7.) There was also evidence that young voters are less likely to have qualifying photo ID. (Id. at 8.) Finally, by definition, young individuals were the only group that could pre-register or participate when high school voter registration drives were mandatory. There is no evidence that young voters disproportionately used the discretionary hour of voting SL 2013–381 transferred from CBOEs to the SBOE.

There is also evidence that, along with every other possible demographic, Representative Warren requested a DMV cross-matching of DMV issued IDs with registered voters broken down by age. (Pl. Ex. 69 at 3 ("[W]e would need to have [the cross-matching] broken down into different categories within each county by all possible demographics that your department typically captures (party affiliation, ethnicity, age, gender, etc.).").) He also requested the same information for one-stop voters and provisional voters. (Id.) Similarly, Senator Stein and other representatives claimed that HB 589 would disproportionately burden young voters. (See, e.g., Pl. Ex. 550 at 30, 33.) But seeButts, 779 F.2d at 147 ; Veasey, 796 F.3d at 487 (collecting cases) ("Conjecture by the opponents of [a law] as to the motivations of those legislators supporting the law is not reliable evidence.").

Next, Plaintiffs point to the fact that college IDs were not permitted under HB 589's voter-ID provisions. But, as noted above, the legislature offered at least plausible reasons for excluding student IDs: (1) there was inconsistency in the way colleges issued IDs, (Pl. Ex. 202 at 68–69), and (2) permitting student IDs would be redundant because some schools require a photo ID to obtain a student ID, (Pl. Ex. 138 at 13).²⁴³ In addition, a district court recently dismissed a Twenty-Sixth Amendment claim against Tennessee's voter ID law, which also omitted student IDs. Nashville Student Org. Comm., 2015 WL 9307284, at *6–7.

243 An earlier version of HB 589 permitted ID cards "issued by the University of North Carolina or its constituent institutions" and ID cards "issued by a North Carolina community college." H.B. 589, 2013 Gen. Assemb., Reg. Sess. § 4 (N.C. 2013) (Edition 5). There are seventeen universities within the University of North Carolina system and fifty-eight public community colleges in the State. University of North Carolina, http://www.northcarolina.edu/ (last visited April 15, 2016); NC Community Colleges, http://www.nccommunitycolleges.edu/ (last visited April 15, 2016); Fed. R. Evid. 201. This would have required poll workers to verify seventy-five additional forms of photo ID of varying formats that were largely redundant of permitted IDs. (Pl. Ex. 549 at 91–92 ("What does Mayland Community College ID look like? What does a Western Carolina ID look like? What does East Carolina's ID look like?").) Failure to do this is at best weak evidence of discriminatory intent.

Even having considered the above with all the other evidence of intent presented by Plaintiffs, this court finds that the legislature enacted HB 589 in spite of, not because of, these disparities.²⁴⁴ Importantly, the legislature offered a non-tenuous reason for the removal of pre-registration and substantial reasons for the voter-ID requirement, the changes to the early-voting schedule, and the removal of SDR and OOP. Further, as noted in the Anderson–Burdick analysis, the removal of CBOE discretion to extend voting for an extra hour on Election Day promotes uniformity by placing extension decisions with one decision maker (the SBOE) rather than 100 CBOEs. See supra Part II.C.6. In addition, in implementing the voter-ID requirement, the legislature followed recommendations of the Carter-Baker Report and provided for more than two years of "soft roll out" to give notice and made voter IDs available for free. Thus the legislature attempted to soften any burden imposed by the ID requirement. For all the reasons noted earlier, the legislative process was not indicative of discriminatory intent. CBOEs may no longer be able to unilaterally extend hours until 5 p.m. on the last day of early voting, but all persons in line by closing time remain entitled to vote, and the new early-voting schedule provides additional night and weekend hours to compensate for this loss.

244 The remaining evidence on intent presented by Plaintiffs consists of the statement of an individual legislator made during debate of the prior version of the voter-ID provision, which Plaintiffs themselves characterize as a "joke," (Doc. 285 at 59 (Representative Collins)); statements of a non-legislator activist seeking to claim credit, whom the court did not find credible, (Doc. 346 at 115–16 (Jay DeLancy)), for ideas that were not necessarily his, (seePl. Ex. 789 at 30–31 (saying that the provisions of HB 589 were not novel ideas, but instead had been "swirling around ... since 2012") (former Rep. Carolyn Justice)); actions of the SBOE after the law's enactment, (Doc. 346 at 116 (Strach direction to DMV)); and early-voting placement decisions by certain CBOEs, (id.at 116–17). This court finds these non-contemporaneous actions and statements to be of limited probative value in determining the motivation behind HB 589. See, e.g., Barber, 560 U.S. at 485–86, 130 S.Ct. 2499 ; Veasey, 796 F.3d at 502–503 ; Jones v. Lubbock, 727 F.2d 364, 371 n. 3 (5th Cir.1984) (refusing to judge intent from statements made by a single member of the legislative body).

In short, what remains under the law provides all voters, including young voters, with an equal and ample opportunity to participate in the political process. This court realizes that the relevant question for intent is limited to the time of enactment, but the fact that young voter turnout increased under the new law is some further indication that undermines Plaintiffs' intent claim.

As such, and having considered the entire record as a whole, the court finds that Plaintiffs have failed to establish that the legislature intended to discriminate against young voters by enacting SL 2013–381.²⁴⁵ This finding is alone sufficient to dispose of Plaintiffs' Twenty-Sixth Amendment claim, but Plaintiffs have also failed to show that the law denies or abridges the right of any eligible eighteen-year-old to vote.

245 This finding also disposes of Intervenor Plaintiffs' Fourteenth Amendment claim. (Doc. 63 in case 1:13cv660 at 22.)

Plaintiffs have failed to show that SL 2013–381 imposes a heavier burden on young voters.²⁴⁶ Given North Carolina's ample alternative registration and voting mechanisms, this court cannot say that young voters' right to vote—or others' right to help young voters register and vote²⁴⁷ —has been denied or abridged. North Carolina, even after SL 2013–381, offers a voting opportunity that the Gaunt court found not to be required by the Twenty-Sixth Amendment: the right of a seventeen-year-old who will be eighteen by the time of the general election to register and vote in the primary election even though they will be only seventeen at the time of the primary. N.C. Gen. Stat. § 163–59. Once an individual is eligible to register—at age seventeen preceding the primary for the first general election in which they will be eligible to vote—North Carolina allows these seventeen-year-olds to register in the same fashion as all other registrants. Even without pre-registration, these opportunities provide Intervenors with sufficient time and methods to register other young voters. North Carolina also offers young voters the same ample voting opportunities as every other age group, namely the opportunity to vote during ten days of early voting, and on Election Day.²⁴⁸ Plaintiffs have failed to show that this system discriminates based on age. In fact, after SL 2013–381, the number of both "young voter" registrants and actual "young voters" increased from 2010 to 2014 (9.7% to 10.4% registrants; 3.9% to 4.2% voters). (Def. Ex. 309 at 77, 78.)

246 As noted above, Intervenor Plaintiffs opted out of participation in this court's January 2016 trial on voter ID, electing to rest on the intent-related ID evidence that was presented at this court's July 2015 trial.

247 Preventing an individual from registering others to vote has been recognized as a legally cognizable claim. Coal. for Sensible and Humane Sols. v. Wamser, 771 F.2d 395, 399 (8th Cir.1985). Testimony at trial indicated that individuals worked with high-schoolers to pre-register them to vote. (E.g., Doc. 329 at 150–51 (Plaintiff Rev. Maria Palmer); Doc. 337 at 123–31 (Helen Compton).) But the right to assist in registration can be no greater than the right of the underlying voters. Having found that so-called young voters have no right to pre-register, it follows that any claim by GOTV efforts on their behalf must fail.

248 Although the First Circuit expressed some concern in Walgrenover the district court's finding that the burdens of students "returning during recess to vote in person or of going through the application and notarial execution process of absentee voting are insignificant," 519 F.2d at 1367, that is far from the case here. As noted above, Plaintiffs have ample voting opportunities that require neither significant travel (as in returning to school) nor voting by mail. It appears that the First Circuit's concern focused on the fact that students in that case had been away from school when elections were held. Surely, they did not have an additional ten days of early voting to vote, as here.

Intervenors' Twenty-Sixth Amendment claim, therefore, fails and will be dismissed.

E. Remedy

Plaintiffs argue that their claims entitle them to various forms of relief: a permanent injunction of the challenged provisions; the appointment of federal election observers through 2019 under § 3(a) of the VRA; and subjecting North Carolina to pre-clearance through 2023. (Doc. 346 at 149.) As a result of the court's findings and conclusions, no permanent injunctive or other relief is warranted.

Following the lifting of the Supreme Court's stay in this case, the Fourth Circuit's pretrial injunction against repeal of SDR and OOP went into place and has been contemplated by voters and would-be registrants for the 2016 primary season. Under Purcell, the court should not alter election laws too close to an election. 549 U.S. at 5, 127 S.Ct. 5 ("Court orders affecting elections, especially conflicting orders, can themselves result in voter confusion and consequent incentive to remain away from the polls. As an election draws closer, that risk will increase."). North Carolina recently completed its March 15, 2016 presidential primary and other Statewide primary election. The State's next primary election for certain U.S. House districts is currently set for June 7, 2016. Because the primary is within weeks of this decision, the court finds that it may likely be disruptive to implement the decision as to the SDR and OOP provisions before that primary. Absentee voting began April 18, and the registration cut-off is May 13. Therefore, insofar as Purcell prescribes it, the court will order the extension of the injunction against repeal of SDR and OOP until June 8, 2016, at which time these two voting mechanisms will be permitted to lapse under SL 2013–381.

III. CONCLUSION

Any restriction on the fundamental right to vote, which is preservative of all others, must be sufficiently justified. Crawford, 553 U.S. at 191, 128 S.Ct. 1610. However, the Supreme Court has admonished that there is a strong presumption that the acts of a duly-elected legislative body are valid. United States v. Nat'l Dairy Prods. Corp., 372 U.S. 29, 32, 83 S.Ct. 594, 9 L.Ed.2d 561 (1963). Consequently, "any time a State is enjoined by a court from effectuating statutes enacted by representatives of its people, it suffers a form of irreparable injury." New Motor Vehicle Bd. of Cal. v. Orrin W. Fox Co., 434 U.S. 1345, 1351, 98 S.Ct. 359, 54 L.Ed.2d 439 (1977). In the midst of these competing principles arise the challenges to the changes in the voting rules in this case.

Not long ago, States observed registration cut-off dates and made voting available only on Election Day. Indeed, even today many States, including New York, one of the Nation's largest and portions of which were previously covered by § 5 of the VRA, continue to do so. Yet, by any measure, North Carolina had recently become progressive nationally by permitting absentee voting; in-person voting for up to seventeen days in addition to Election Day; additional registration after the cut-off, including up to three days before Election Day; the casting of a ballot on Election Day in any unassigned precinct in the voter's county; and "pre-registration" by sixteen-year-olds. Some of these measures were controversial when enacted, and one—out-of-precinct provisional voting—was passed on a purely partisan basis in a manner designed to gain Democrats electoral victory retroactively.

In 2013, North Carolina retrenched. Changes to the State's election laws were contained in a fifty-seven-page bill, but in truth only about fifteen pages addressed the provisions at issue today. They reduced voting days but extended hours in like fashion, on the ground that doing so restored fairness and prevented previous gaming of the system; eliminated same-day registration, on the ground that experience showed that juxtaposing registration so close to Election Day resulted in the counting of ballots by voters who later failed the State's statutory mail verification system for ensuring eligibility; required voters to appear on Election Day at their assigned voting precinct, on the ground that the precinct-based system (on the authority of the North Carolina Supreme Court) serves recognized and legitimate State interests; returned minors' registration eligibility to the age of eligibility, which still allows them to vote in primaries at age seventeen and is more generous than required by law; and joined almost half of the Nation's other States in passing a form of voter photo ID in furtherance of State interests recognized by the Supreme Court in Crawford as legitimate.

Because these provisions, although originally proposed in separate bills, were consolidated into one bill considered together and only after the Supreme Court's decision in Shelby County, opponents cried foul and infer racial intent, particularly given that some of the provisions were used disproportionately by African Americans. Proponents argue that the changes were based on legitimate and substantial concerns of election integrity and fairness and leave an electoral system that provides generous, fair, and equal opportunity for voters of all races and ages.

Plaintiffs themselves acknowledge that the removed mechanisms were "conveniences" and "fail-safes" to the ordinary rules for voting. By definition, therefore, any repeal or modification results in a marginal reduction or modification in options for those who preferred them. The question is not what this court would prefer or find best as a policy matter. That is the province of legislatures and executives, both of which represent the will of the people as elected representatives. Rather, the question in this case is whether Plaintiffs have demonstrated that the measures violate the VRA or the Constitution.

At the preliminary injunction stage, Plaintiffs argued that the mechanisms were adopted to increase minority participation and forecast that their elimination would result in reductions (indeed, sizeable ones) in African American participation in the electoral process. This court was not persuaded that Plaintiffs had made a clear showing for relief and denied the request. On that record, the Fourth Circuit affirmed in part but directed the court to enjoin the repeal of same-day registration and out-of-precinct provisional voting. However, as a result of the Supreme Court's stay of that mandate, the State proceeded under SL 2013–381, which provided data from the 2014 general election as to the actual results of the implementation of the law. Such data are precisely the sort of electoral information courts are encouraged to consider, because they permit an understanding of the effect of a law based on "historical facts rather than speculation." Purcell, 549 U.S. at 6, 127 S.Ct. 5 (Stevens, J., concurring).

After twenty-five days of trial and reviewing over 20,000 pages of record and the testimony of over 20 expert and 100 fact witnesses, and after considerable reflection, the court is in a position to evaluate the effect of SL 2013–381 based on actual historical facts, rather than speculation.

The evidence shows that African Americans have fared better in terms of registration and turnout rates in 2014, after the new law was implemented, than in 2010, when the old provisions were in place. Presently, African Americans hold a commanding lead over all other races in voter registration (e.g., 91.2% African American vs. 83.4% white). Since SL 2013–381 has been in place, African American turnout not only increased but did so at a greater rate than that of other groups (including whites). In fact, the 2014 general election saw the smallest white-African American turnout disparity in any midterm election from 2002 to 2014. These facts may surprise some, including Plaintiffs' experts (who predicted the opposite before trial), but the outcomes are less surprising when one realizes that Plaintiffs were unable to show that the removed mechanisms were responsible for the greater-than-parity political participation achieved by African Americans. Instead, there was strong evidence that other factors were more substantially at play, such as President Obama's candidacy and North Carolina's emergence as a battleground State. Contrary to Plaintiffs' predictions and somewhat surprisingly, the scholarly consensus of Plaintiffs' own experts revealed that early voting depresses turnout. Moreover, same-day registration has not been shown to increase turnout in a statistically significant manner. Only Election Day registration, which is not at issue here, does.

The 2014 data merely confirm what the remaining data suggest: that minorities enjoy equal and constitutionally-compliant opportunity to participate in the electoral process. No doubt, many variables fed into the 2014 aggregate turnout and registration data. But Plaintiffs' evidence failed to provide anything more persuasive. For example, despite having the ability to do so, Plaintiffs did not perform the type of analysis necessary to separate out the causal effect of various factors said to drive election participation results, which even Plaintiffs' experts seemed to agree would be the appropriate analysis to measure the alleged effects of the measures at issue. Nor could Plaintiffs show that the 2014 results would have been any better for minorities had the removed mechanisms remained in place. Plaintiffs bear the burden of proof in this case, and Defendants cannot be blamed for this absence.

Moreover, while the educational and socioeconomic disparities suffered by African Americans might suggest that the removed mechanisms would disproportionately benefit them, this court cannot find an inequality of opportunity simply because educational and socioeconomic disparities suggest one might exist—there must actually be an inequality of opportunity. LULAC, 999 F.2d at 867 ("[T]he Senate Report, while not insisting upon a causal nexus between socioeconomic status and depressed participation, clearly did not dispense with proof that participation in the political process is in fact depressed among minority citizens."). The evidence shows that, like all voters, African Americans are not only capable of adjusting, but have adjusted, to SL 2013–381.

North Carolina's ample remaining alternative registration and voting mechanisms are the likely explanation for this adjustment. Even after SL 2013–381 as amended, North Carolinians of all races can register in many convenient ways and can vote by mail, during ten days of early voting—now at more sites and at more convenient hours—or on Election Day at their correct precinct. It is a given that government should endeavor to make it as easy as practicable to exercise the right to vote. But the fact that voting can almost always be made easier does not render a State's failure to do so, or a State's repeal of a convenience or "fail-safe," unlawful or unconstitutional per se. With every relaxation of the rules there is often an attendant trade-off or effect on verification and election integrity. The State demonstrated as much here.

Same-day registration created administrative burdens that could not be overcome despite great effort by additional CBOE staff. Most importantly, it was incompatible with the State's statutory mail verification process for eligibility. Same-day registration's close proximity to Election Day did not just create a risk that voters would not be mail-verified before the same-day registrant's vote was counted, it ensured that a significant number of ballots (2,361 voters in 2012 alone) would be counted in each election even though the registrant later failed mail verification after any challenge could be made. This may be why most States do not implement same-day registration. In any event, other States use mail verification, and Plaintiffs have not demonstrated that North Carolina's historical reliance on it is unreasonable or reflects a racial bias or result. Plaintiffs urge that same-day registration's requirement that a registrant show a HAVA-compliant document (showing name and address) is sufficient to protect the State's interests, but this ignores the fact that the very voters about whom Defendants complain did just that yet still failed mail verification.

As for early voting, the evidence was that the old schedule had several days with low usage, many counties did not use the full allotment of days (including Sundays), and sites were placed inconsistently and based on partisan advantage. By requiring that the number of early-voting hours nevertheless remain the same, the new law creates more early-voting sites and additional night and weekend hours that are more convenient to workers and lower socio-economic voters. The 2014 data proved as much.

With regard to out-of-precinct provisional voting, the trial evidence demonstrated that voters tended to use it out of added convenience or inattentiveness to their assigned precinct. For over 100 years, no one seriously contended that the State's statutorily-provided precinct system provided inequality of opportunity to vote. Judged against the immaterial burden of requiring a voter to cast a ballot in his or her assigned precinct, and given the benefits of the precinct system recognized by the North Carolina Supreme Court in James and the Sixth Circuit, this court cannot say that returning to it (circa 2004) imposes an unlawful or unconstitutional burden not justified by the benefits.

As for pre-registration, there can be no dispute that allowing sixteen-year-olds to "pre-register" is not required for equality of electoral opportunity. North Carolina already provides ample time to register and permits even seventeen-year-olds to vote in the primary before their eligible general election.

Finally, North Carolina's voter-ID requirement, now with a reasonable impediment exception, serves legitimate State interests recognized by the Supreme Court in Crawford without imposing a material burden on any group of voters. Plaintiffs' contention that North Carolina's requirement is one of the strictest in the country ignores the reasonable impediment exception. If North Carolina is an outlier, it is because it is one of only two States in the Nation to accommodate voters who wish to vote in person but for whatever reason face an impediment to acquiring qualifying ID. The court's conclusion is fully consistent with the same finding of the three-judge court that approved a virtually identical photo-ID law with reasonable impediment in South Carolina, 898 F.Supp.2d 30.

In short, North Carolina has provided legitimate State interests for its voter-ID requirement and electoral system that provides registration all year long up to twenty-five days before an election, absentee voting for up to sixty days before an election, ten days of early voting at extended hours convenient for workers that includes one Sunday and two Saturdays, and Election Day voting. Plaintiffs oppose this system because they preferred one that they say was even more convenient—which they used disproportionately during certain elections—and point to some fraction of voters who did not vote or register. Plaintiffs' contention that such voters did not do so because of vestiges of historical official discrimination is rebutted by the facts. There is strong evidence that some other reason is at play for the failure of these persons to register and/or vote. The unprecedented gains by African Americans in registration and turnout, both during and even in 2014 after SL 2013–381, bolster this conclusion. While the consideration is clearly local and practical in nature, based on North Carolina's unique facts, it would no doubt bear relevance if North Carolina were seeking to return to an electoral system that was not in the mainstream of other States. It is not.

For all these reasons, Plaintiffs have failed to demonstrate that Defendants have violated § 2 of the VRA or the Fourteenth, Fifteenth, or Twenty-Sixth Amendments to the United States Constitution.

For the reasons stated, therefore,

IT IS ORDERED in cases 1:13CV861, 1:13CV658, and 1:13CV660 as follows:

1. The United States' Motion in Limine to Exclude Improper Expert Opinion Testimony of Brian Neesby, (Doc. 265), the United States' Motion to Strike Certain Testimony of Brian Neesby, (Doc. 326), and the NAACP Plaintiffs' Motion in Limine to Exclude Defendants' Untimely Discovery Materials," (Doc. 291), are DENIED IN PART to the extent the challenged materials are relied upon in this memorandum opinion and are otherwise DEEMED MOOT. The Duke Intervenor-Plaintiffs' Motion to Strike Brian Neesby's Testimony Regarding Mail-Verification Failure Rates Among Preregistrants and to Exclude Defendants' Exhibit BN-3, (Doc. 327), need not be resolved insofar as the court did not rely on this evidence, and the motion is therefore DEEMED MOOT.

2. Plaintiffs' Motion to Strike the Declarations of Sean P. Trende and Motion to Exclude his Testimony at Trial, (Doc. 271), is DENIED IN PART to the extent relied upon in this memorandum opinion and is otherwise DEEMED MOOT.

3. Plaintiffs' Motion to Strike the Declarations of Thomas Brooks Hofeller and Motion in Limine to Exclude his Testimony at Trial, (Doc. 273), is DEEMED MOOT in light of Plaintiffs' withdrawal of the motion at trial, (Doc. 340 at 140–41). This court's memorandum opinion relied only upon the redacted version of Dr. Hofeller's report. (Def. Ex. 212A.)

4. Plaintiffs' Motion to Strike Portions of the Declarations and Motion in Limine to Exclude Certain Testimony of Dr. Janet Thornton at Trial, (Doc. 275), was granted in part at trial and otherwise is DENIED to the extent her testimony is relied upon in this memorandum opinion.

5. Defendants' Motion to Exclude Testimony by the United States' Expert, Dr. Charles Stewart, (Doc. 269), was held in abeyance by Defendants when trial was bifurcated in July 2015, (Doc. 279), and never renewed. In any event, it is DENIED, as the arguments for exclusion have been considered by the court as to the weight of the evidence.

6. Defendants' Motion in Limine to exclude certain testimony of Dr. Allan Lichtman, (Docs. 287), is GRANTED IN PART AND DENIED IN PART. Defendants' motion to exclude Dr. Lichtman's supplemental report of June 9, 2015, is DENIED, but the arguments for exclusion have been considered as to the weight of the evidence. Defendants' motion to exclude Dr. Lichtman's matching analysis is DENIED, but the arguments for exclusion have been considered as to the weight of the evidence. Defendants' motion to exclude certain emails is GRANTED as to the truth of the matter asserted on the ground of hearsay and is otherwise DENIED. Defendants' motion to exclude certain conclusions by Dr. Lichtman as to discriminatory intent is GRANTED for the reasons addressed in this memorandum opinion. Defendants' objection to certain newspaper articles relied upon by Dr. Lichtman was sustained at trial. (Doc. 333 at 154–55.)

7. Defendants' motion to exclude rebuttal testimony of Dr. Lichtman, (Doc. 342 at 186–90; Docs. 324, 325), is GRANTED as to mail verification rates for 2014, (see Doc. 325 at 9–12), and is otherwise DENIED.

IT IS FURTHER ORDERED that in cases 1:13CV861, 1:13CV658, and 1:13CV660 all claims of Plaintiffs and Plaintiff-Intervenors are DISMISSED WITH PREJUDICE and judgment shall be entered in favor of Defendants, but that this court's injunction against the repeal of same-day registration and out-of-precinct provisional voting entered October 3, 2014, at the direction of the Fourth Circuit, (Doc. 189), (which was thereafter recalled by the Fourth Circuit until May 5, 2015) shall remain in effect until June 8, 2016, under the Purcell doctrine.

Any motion for recovery of costs and/or attorneys' fees shall be governed by the Federal Rules of Civil Procedure, this court's Local Rules, and any other applicable rule.

An appropriate Judgment will be entered separately.