Landya McCafferty, United States District Judge

Claudia Center, Pro Hac Vice, San Francisco, CA, Dale E. Ho, Pro Hac Vice, Julie A. Ebenstein, Pro Hac Vice, New York, NY, Paul J. Twomey, Twomey Law Office, Chichester, NH, Gilles R. Bissonnette, American Civil Liberties Union of New Hampshire, Concord, NH, for Mary Saucedo, et al.

Laura E. B. Lombardi, NH Attorney General's Office (Civil), Matthew T. Broadhead, NH Department of Justice, Concord, NH, for William Gardner, Secretary of State of the State of New Hampshire, in his official capacity, et al.

ORDER

Landya McCafferty, United States District Judge

This case concerns New Hampshire's signature-match requirement for absentee ballots. The act of signing one's name is often viewed as a rote task, a mechanical exercise yielding a fixed signature. A person's signature, however, may vary for a variety of reasons, both intentional and unintentional. Unintentional factors include age, physical and mental condition, disability, medication, stress, accidents, and inherent differences in a person's neuromuscular coordination and stance. Variations are more prevalent in people who are elderly, disabled, or who speak English as a second language. For the most part, signature variations are of little consequence in a person's life.

But in the context of absentee voting, these variations become profoundly consequential. The signature-match requirement in RSA 659:50, III requires every local election moderator to compare the signature on a voter's absentee-ballot application to the signature on an affidavit that the voter sends with the absentee ballot. If the signature on the affidavit does not appear "to be executed by the same person who signed the application," the moderator must reject the voter's ballot. RSA 659:50, III. The purpose of the requirement is to ensure that the same person executes both the absentee-ballot application and the affidavit. In recent elections, however, the signature-match requirement has disenfranchised hundreds of absentee voters.

As will become evident, this signature-matching process is fundamentally flawed. Not only is the disenfranchised voter given no right to participate in this process, but the voter is not even given notice that her ballot has been rejected due to a signature mismatch. Moreover, moderators receive no training in handwriting analysis or signature comparison; no statute, regulation, or guidance from the State provides functional standards to distinguish the natural variations of one writer from other variations that suggest two different writers; and the moderator's assessment is final, without any review or appeal.

Plaintiffs Mary Saucedo, Maureen P. Heard, and Thomas Fitzpatrick are among the 275 absentee voters whose ballots were rejected in the 2016 General Election as a result of RSA 659:50, III. They bring suit against defendants William M. Gardner (New Hampshire's Secretary of State), and the New Hampshire Secretary of State's Office, alleging constitutional claims under 42 U.S.C. § 1983, and a claim under the Americans with Disabilities Act ("ADA"). Before the court are the parties' cross-motions for summary judgment, as well as plaintiffs' motion to strike. For the following reasons, plaintiffs' motion for summary judgment is granted in part, defendants' cross-motion for summary judgment is granted in part, and plaintiffs' motion to strike is denied.

STANDARD OF REVIEW

A movant is entitled to summary judgment if it "shows that there is no genuine dispute as to any material fact and [that it] is entitled to judgment as a matter of law." Fed. R. Civ. P. 56(a). In reviewing the record, the court construes all facts and reasonable inferences in the light most favorable to the nonmovant. Kelley v. Corr. Med. Servs., Inc., 707 F.3d 108, 115 (1st Cir. 2013).

"On issues where the movant does not have the burden of proof at trial, the movant can succeed on summary judgment by showing ‘that there is an absence of evidence to support the nonmoving party's case.’ " OneBeacon Am. Ins. Co. v. Commercial Union Assur. Co. of Canada, 684 F.3d 237, 241 (1st Cir. 2012) (quoting Celotex Corp. v. Catrett, 477 U.S. 317, 325, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986) ). If the moving party provides evidence to show that the nonmoving party cannot prove a claim, the burden shifts to the nonmoving party to show that there is at least a genuine dispute as to a factual issue that precludes summary judgment. Woodward v. Emulex Corp., 714 F.3d 632, 637 (1st Cir. 2013).

BACKGROUND

The following facts are undisputed, unless otherwise noted. As stated above, plaintiffs are voters who attempted to vote by absentee ballot in the 2016 General Election. Plaintiff Saucedo voted by absentee ballot due to a disability (blindness), and plaintiffs Fitzpatrick and Heard voted by absentee ballot because they were out of the state on Election Day. They brought suit in May 2017, after learning that their absentee ballots had been rejected. All of their ballots were rejected on the basis of the signature-match requirement in RSA 659:50, III. Defendant Gardner, as the Secretary of State, is the "Chief Election Officer" under state law. RSA 652:23. Among other things, defendants produce absentee-voting forms and documents, and provide election information and materials to local officials and the public. See RSA 652:22 ; RSA 652:23 ; RSA 657:4 ; RSA 657:7.

The court begins by describing the general procedure by which absentee ballots are processed and counted in New Hampshire, before discussing how that procedure played out in the 2016 General Election. Then, the court summarizes the evidence the parties have proffered in support of their competing motions for summary judgment.

I. Absentee Voting in New Hampshire

New Hampshire authorizes absentee voting for certain categories of voters-namely, those who cannot appear at the polls because they are: (1) absent from the municipality on Election Day; (2) observing a religious commitment; (3) unable to vote in person due to physical disability; or (4) unable to appear because of an employment obligation. RSA 657:1.

The first step in the absentee-voting process is for a voter to apply for the absentee ballot. The Secretary of State creates application forms and distributes them to municipalities. RSA 657:4, I; RSA 657:5. A voter may request a form from a town or city clerk, or from the Secretary of State. RSA 657:6. Alternatively, a voter may receive a ballot from the town or city clerk simply by providing a written statement containing all of the necessary information. RSA 657:6.

In the absentee-ballot application, the voter must identify the reason that she is qualified to vote by absentee ballot, and must provide basic biographical information-including name, address, phone number, and email address, though the phone number and email address are optional sections. What is most relevant here is that the application requires the voter to sign her name. Prior to and in the 2016 General Election, there was no notice on the application that the application signature would be compared with another signature; instead, below the signature line was the following statement: "Voter must sign to receive an absentee ballot." Doc. no. 49-9 at 2.

However, as a result of amendments to the absentee-ballot statutory scheme in 2017, the application now contains the following statement below the signature line: "The applicant must sign this form to receive an absentee ballot. The signature on this form must match the signature on the affidavit envelope in which the absentee ballot is returned, or the ballot may be rejected." RSA 657:4, I. In addition, there is a new section, which provides notice that "[a]ny person who assists a voter with a disability in executing this form shall make a statement acknowledging the assistance on the application form to assist the moderator when comparing signatures on election day." Id. Below the notice, there are lines for the assistant to print and sign her name.

Upon receipt of a properly executed application, the clerk provides the voter with: (1) an absentee ballot; (2) an affidavit envelope; and (3) a return envelope. RSA 657:15, I; RSA 657:7, I-III. The voter marks the ballot and places the ballot in the affidavit envelope. RSA 657:17. On the face of the affidavit envelope is an affidavit that the voter must execute. The affidavit requires the voter to again certify that she is voting by absentee ballot for a qualifying reason, and requires that the voter print and sign her name. As a result of the 2017 amendments to the statute, below the signature line is the following notice:

The signature on this affidavit must match the signature on the application for an absentee ballot or the ballot may be rejected. A person assisting a blind voter or voter with a disability who needs assistance executing this affidavit shall make and sign a statement on this envelope acknowledging the assistance in order to assist the moderator when comparing signatures on election day.

RSA 657:7, II. Below, there is a space for an assistant to print and sign her name. See doc. no. 54-7 at 1.

After executing the affidavit, the voter places the affidavit envelope in the return envelope, and submits the package to the town or city clerk. RSA 657:17. The clerk attaches the voter's application to the received absentee-ballot package, but does not open or otherwise process the package prior to Election Day. RSA 657:18.

On Election Day, the clerk delivers the absentee-ballot packages to the local moderator. RSA 657:23. The moderator is a local, elected position with a two-year term. RSA 40:1. Among other things, the moderator oversees the "conduct of voting" and the implementation of New Hampshire's election statutes in her municipality. RSA 659:9. Moderators "are not employees of the Department of State" and are only accountable to local voters. Doc. no. 54 at 2, 31.

Generally, moderators begin processing absentee ballots at 1:00 p.m. on Election Day, though they may open the return envelopes prior to that time. See RSA 659:49, I; RSA 659:49-b. Processing is done in public view. The moderator begins processing absentee ballots "by clearly announcing that he or she is about to open the envelopes which were delivered to him or her." RSA 659:50. The moderator then removes each affidavit envelope from the return envelope and compares the signature on the affidavit with the signature on the voter's application.

An absentee ballot is accepted if: (1) the name of the voter is on the voter checklist; (2) the affidavit "appears to be properly executed"; (3) the "signature on the affidavit appears to be executed by the same person who signed the application, unless the voter received assistance because the voter is blind or has a disability"; and (4) the "signatures appear to be the signatures of a duly qualified voter who has not voted at the election." RSA 659:50, I-IV. If all of these requirements are met, and the ballot is not challenged by another voter, see RSA 659:51, I, the moderator opens the affidavit envelope and takes out the absentee ballot to be counted. The moderator may begin counting accepted absentee ballots after polls close. RSA 659:49, I. If one of the requirements is not met, the moderator rejects the ballot, marks the affidavit envelope with the reason for rejection, and does not open the affidavit envelope. RSA 659:53.

There is no procedure by which a voter can contest a moderator's decision that two signatures do not match, nor are there any additional layers of review of that decision. In other words, the moderator's decision is final. Moreover, no formal notice of rejection is sent to the voter after Election Day. Rather, after the election, a voter may determine whether and why her absentee ballot was rejected via a website maintained by the Secretary of State. See RSA 657:26. Defendants remove this information from the website ninety days after the election, however.

II. The Signature-Match Requirement

As noted above, each local moderator is tasked with comparing the signature on the affidavit with the signature on the application to determine whether "[t]he signature on the affidavit appears to be executed by the same person who signed the application." RSA 659:50, III. As a result of the 2017 amendments to the statute, voters receiving assistance with the execution of their voting materials due to blindness or disability are exempt from the requirement. Id.

On its face, RSA 659:50, III gives no guidance on the questions that inevitably arise in applying the requirement, including what stylistic variations suggest that two signatures were made by different individuals, and what threshold number of variations is required to conclude that the signature on the affidavit does not "appear to be" executed by the same person who signed the application. No other state statute or regulation elaborates the standard set forth in RSA 659:50, III, and there does not appear to be any authoritative case law on the subject.

The record discloses only two sources that provide additional guidance to moderators: the Secretary of State's Office and, in the 2016 General Election, the New Hampshire Attorney General's Office. The Secretary of State's Office publishes an Election Procedure Manual, which offers the following:

The test for whether the application and affidavit appear to be signed by the same person is whether this is more likely than not. Absentee ballots should be rejected because the signatures do not match only if the differences in the signatures are significant.

...

Moderators should exercise careful judgment when rejecting an absentee ballot because the signature of the voter on the affidavit does not appear to be signed by the same person who signed the absentee ballot application. The test is whether it is more likely than not that the same person signed both forms. It is a natural and common occurrence that a person's signature will change over time and will have differences even when the person writes out his or her signature several times, one immediately after another. A moderator deciding to reject an

absentee ballot because the signatures do not match should be prepared to explain to the Attorney General's Office or a Superior Court judge what specific characteristics on the two signatures were the basis of the decision that they were more likely than not signed by different people. While signature verification is an important safeguard against voting fraud, as with all safeguards, the analysis starts with a presumption of validity and the decision to disenfranchise a voter must be made only when there is sufficient evidence to justify that act.

Doc. no. 49-19 at 14, 16. The Secretary of State's Office also conducts optional trainings for moderators prior to elections, at which it reiterates, but does not further elaborate on, the guidance set forth in the manual. Doc. no. 49-3 at 77, 79 (deposition of David Scanlan). The Secretary of State does not regularly monitor rates of rejection due to signature mismatch to ensure moderators' compliance with the statute, see id. at 178, 180-83, and has never engaged in a review of any statistical anomalies related to the requirement, id. at 185.

Prior to the 2016 General Election, the Attorney General's Office issued a memorandum to local election officials, which contained the following guidance:

In determining whether signatures match, the moderator should decide whether it is more likely than not that the same person signed both forms. The more likely than not standard does not require a perfect match.... Moderators should be aware that a person's signature often varies depending on the circumstances, and it is often hard to tell whether two signatures were written by the same person. Because a mistake will deprive a citizen of his/her constitutional right to vote, moderators should take great care before ruling a ballot invalid because of signature differences.

Doc. no. 49-20 at 5.

In essence, the guidance provided to moderators constitutes a burden of proof (more likely than not), and a requirement that signature comparison be done based on objective criteria (whatever those criteria may be). But moderators receive no training in handwriting analysis, and they are not screened for conditions, such as poor eyesight, that may impede their ability to discern subtle variations in signatures. The assumption seems to be that the substantive task of signature comparison is one of common sense.

Defendants have also provided affidavits from a number of local election officials to show "how cities and towns actually implement RSA 659:50." Doc. no. 54 at 8. Defendants state that the practices of these officials are consistent with the practices of election officials statewide. See doc. no. 58 at 7.

There are three practices that are worth highlighting. First, moderators normally have a team of volunteers that help them compare signatures and determine whether signatures match for purposes RSA 659:50, III. Nevertheless, while volunteers may help a moderator reach a decision, the final decision "rests with the moderator." Doc. no. 54 at 10.

Second, in deciding whether to accept an absentee ballot, moderators "consider all of the evidence available to them, including their personal knowledge of the voter or the personal knowledge of another election official." Id. at 11. For example, one moderator states that, because he "know[s] the addresses of the assisted living facilities" in his town, he takes "that into account as a factor when conducting a signature match," and allows "for more variability in signatures in these cases." Doc. no. 54-15 at 2. Another moderator stated that, in one instance, he rejected a ballot based on an assistant moderator's personal knowledge of the voter. The assistant moderator had previously seen the voter's signature on medical documents, and the signature on the affidavit was inconsistent with the signature on the medical documents.

At the hearing, defendants clarified how moderators consider extrinsic evidence. They explain that state law contemplates a two-step process, citing RSA 659:53 and RSA 659:54 in support. At the first step, a moderator compares the signature on the affidavit envelope to the signature on the application to determine whether the documents were executed by the same person, as required by RSA 659:50, III. This determination is made solely based on the signatures themselves and without reference to extrinsic evidence. If the moderator determines that the signatures do not pass muster under RSA 659:50, III based on an examination of the signatures alone, there is a second step: the moderator proceeds to determine whether there is any extrinsic evidence available to the moderator or the assistants that would allow the moderator to conclude that the same person did, in fact, execute both documents. At the hearing, defense counsel gave the example of an absentee voter who has his affidavit envelope notarized.

Third, the affidavits show that moderators conceive of the relevant standard differently. One moderator stated that she will not reject an absentee ballot "unless the signatures on the request form and affidavit envelope are drastically different." Doc. no. 54-13 at 2 (emphasis added). By contrast, another averred that he does "not reject sets of signatures on the basis that they don't look the same, but only if there are no characteristics which suggest that they could both have been signed by the same person. In some instances, we are satisfied if one or two letters ... share a characteristic style." Doc. no. 54-14 at 2 (emphasis added).

III. Statistics from the 2016 General Election

In the 2016 General Election, .35% of all absentee ballots submitted were rejected due to a signature mismatch (275 rejections out of 78,430 absentee ballots). See doc. no. 49-22 at 9. This extremely low rate of rejection due to a signature mismatch is consistent with the rates seen in the 2012 and 2014 General Elections.However, there were some variances between polling places in the 2016 General Election. 74% of New Hampshire's polling places had no rejections due to a signature mismatch (236 of 318). Of the 26% that did, rates varied, sometimes significantly, both between municipalities and within them. For example, in Portsmouth Ward 3, 5.21% of all absentee ballots were rejected due to a signature mismatch; in Portsmouth Ward 2, the rate was .43%; in Bedford, .88%; in Hudson, 1.68%; in Manchester Ward 4, 2.17%; in Manchester Ward 6, .23%. See generally doc. no. 49-23. The parties dispute the significance of these disparities. Plaintiffs argue that these figures demonstrate the lack of uniform application of the signature-match requirement, while defendants assert that, absent further statistical analysis, any differences could be attributed to other variables.

Defendants provide some statistics of their own. Defendants compared, for the 2016 General Election, the municipalities whose election officials attended the training session held by the Secretary of State's Office and municipalities whose officials did not. They found that six of the eight towns with the highest rates of rejection due to signature mismatch had failed to attend the training. Defendants also found that the towns whose officials did not attend had a higher rate of rejection due to signature mismatch-more than double per town. On this basis, defendants contend that this demonstrates that there is no fundamental flaw in the statute, and that, at most, more training is required. Plaintiffs challenge defendants' analysis of the numbers as "based on a meaningless calculation." Doc. no. 62 at 5. They argue, among other things, that analyzing rejections on a per town basis, without consideration of total absentee ballots cast and rejected in each town, masks disparities in rejection rates among towns.

IV. Dr. Mohammed

Finally, plaintiffs retained an expert, Linton Mohammed, Ph. D, to support their claims. Defendants do not dispute Dr. Mohammed's opinions. Dr. Mohammed is a forensic document examiner, specializing in handwriting and signature identification. He opines that a person's signature may vary for a variety of reasons, both intentional and unintentional. Unintentional factors include age, physical and mental condition, disability, stress, accidental occurrences, inherent variances in neuromuscular coordination, and stance. Variations are more prevalent in writers who are elderly, disabled, ill, or who speak English as a second language. Dr. Mohammed explains that, in order to account for these variations and make an accurate determination, one needs extensive training, adequate magnification and lighting equipment, sufficient time, and excellent eyesight. Furthermore, a forensic document examiner will normally require at least ten exemplar signatures to compare to a questioned signature.

Dr. Mohammed opines that in applying RSA 659:50, III, moderators will likely make erroneous determinations. Lay moderators do not have the training, time, equipment, or number of exemplars necessary to make a proper determination. In addition, Dr. Mohammed states that laypeople erroneously tend to focus on the "eye-catching" features of single letters, rather than the holistic features of the signature, like alignment and slant. Dr. Mohammed opines that holistic features are the more significant characteristics in signature comparison.

Defendants also rely upon Dr. Mohammed's opinion. They cite his statement that "a signature is developed as a form of identification" to argue that the signature-match requirement serves to identify a voter and prevent voter fraud. Doc. no. 54-11 at 24. Plaintiffs challenge this inference, contending that Dr. Mohammed's point was merely to articulate the difference between a signature and handwriting generally, not to suggest that the signature-match requirement identifies the voter.

Defendants also highlight Dr. Mohammed's testimony regarding the different styles of signatures. Dr. Mohammed divides signatures into one of three categories: text-based, mixed-style, and stylized. A text-based signature is one where the letters can be fully read; a mixed-style signature is one where some, but not all, of the characters can be read; and a stylized signature is "basically a pattern and there's ... no characters [one] can read within that signature." Id. at 42-43. Dr. Mohammed determined that 94 of the rejected voters from the 2016 General Election used different styles in their affidavits and applications. Defendants contend that this finding shows that these voters were correctly disenfranchised because they failed to comply with the notice on the absentee-ballot instructions-that a ballot "will be counted only if it appears that the same person signed both documents." Doc. no. 54-8 at 2. Plaintiffs respond that moderators are not trained to evaluate signature styles and that, in any case, state law does not require that a voter "use the same signature style" when voting by absentee ballot. Doc. no. 62 at 9.

DISCUSSION

Plaintiffs bring four claims challenging RSA 659:50, III. The first three are grounded in the U.S. Constitution. Plaintiffs contend that the statute facially violates their procedural due process rights (Count I), their fundamental right to vote (Count II), and their right to have their votes treated uniformly, under the principles enunciated in Bush v. Gore, 531 U.S. 98, 121 S.Ct. 525, 148 L.Ed.2d 388 (2000) (Count III). On behalf of Ms. Saucedo, plaintiffs also bring an ADA claim.

Because the parties express some disagreement over how the court should analyze plaintiffs' claims, given that they are facial challenges to RSA 659:50, III, the court will provide some clarification before addressing the merits.

"The Supreme Court has articulated two formulations of the standard for assessing facial challenges to statutes." Libertarian Party of N.H. v. Gardner, 843 F.3d 20, 24 (1st Cir. 2016). "In United States v. Salerno, 481 U.S. 739, 107 S.Ct. 2095, 95 L.Ed.2d 697 (1987), the Court held that a facial challenge can only succeed where the plaintiff ‘establishes that no set of circumstances exists under which the Act would be valid.’ " Id. Alternatively, a plaintiff bringing a facial challenge to a statute must establish that it lacks any "plainly legitimate sweep." Hightower v. City of Boston, 693 F.3d 61, 77 (1st Cir. 2012). The First Circuit recently relied on the latter formulation in a ballot-access case. See Gardner, 843 F.3d at 24. But see Pharm. Research & Mfrs. of America v. Concannon, 249 F.3d 66, 77 (1st Cir. 2001) (citing Salerno test favorably).

These standards may obscure the relevant inquiry, however, as they could be taken to suggest that a court's task is to "conjure up" hypothetical situations "in which application of the statute might be valid." United States v. Sup. Ct. of N.M., 839 F.3d 888, 917 (10th Cir. 2016). But, as courts have noted, the Supreme Court "has often considered facial challenges simply by applying the relevant constitutional test to the challenged statute, without trying to dream up whether or not there exists some hypothetical situation in which application of the statute might be valid." Bruni v. City of Pittsburgh, 824 F.3d 353, 363 (3d Cir. 2016) (collecting cases); see also Doe v. City of Albuquerque, 667 F.3d 1111, 1124 (10th Cir. 2012).

Therefore, in practice, "[a] facial challenge is best understood as a challenge to the terms of the statute, not hypothetical applications, and is resolved simply by applying the relevant constitutional test to the challenged statute." Sup. Ct. of N.M., 839 F.3d at 917 (internal quotation marks and citation omitted); see also Wash. State Grange v. Wash. State Republican Party, 552 U.S. 442, 449-50, 128 S.Ct. 1184, 170 L.Ed.2d 151 (2008) ("In determining whether a law is facially invalid, we must be careful not to go beyond the statute's facial requirements and speculate about ‘hypothetical’ or ‘imaginary’ cases."); Ezell v. City of Chicago, 651 F.3d 684, 698 (7th Cir. 2011).

I. Procedural Due Process (Count I)

Plaintiffs first argue that RSA 659:50, III violates the requirements of procedural due process because it lacks any pre-deprivation process: voters receive neither prior notice of, nor an opportunity to cure, a rejection due to a signature mismatch. Defendants respond that the "extremely slight risk of an erroneous deprivation," in conjunction with the "significant burden on the State" to create additional procedures, support the conclusion that RSA 659:50, III does not violate procedural due process. Doc. no. 54 at 22.

"To establish a procedural due process violation, [a] plaintiff must identify a protected liberty or property interest and allege that the defendants, acting under color of state law, deprived [him] of that interest without constitutionally adequate process." González-Droz v. González-Colón, 660 F.3d 1, 13 (1st Cir. 2011) (internal quotation marks omitted). "No rigid taxonomy exists for evaluating the adequacy of state procedures in a given case; rather, due process is flexible and calls for such procedural protections as the particular situation demands." Id. (internal quotation marks omitted). Still, "[t]he basic guarantee of procedural due process is that, before a significant deprivation of liberty or property takes place at the state's hands, the affected individual must be forewarned and afforded an opportunity to be heard at a meaningful time and in a meaningful manner." Id. (internal quotation marks omitted).

Under Mathews v. Eldridge, 424 U.S. 319, 96 S.Ct. 893, 47 L.Ed.2d 18 (1976), "determining what process is due requires balancing three factors: first, the private interest that will be affected by the official action; second, the risk of an erroneous deprivation of such interest through the procedures used, and the probable value, if any, of additional or substitute procedural safeguards; and finally, the Government's interests." Collins v. Univ. of N.H., 664 F.3d 8, 17 (1st Cir. 2011) (internal quotation marks and brackets omitted). In addition, the Mathews court emphasized that "procedural due process rules are shaped by the risk of error inherent in the truthfinding process as applied to the generality of cases, not the rare exceptions." Mathews, 424 U.S. at 344, 96 S.Ct. 893. As a result, and given that this is a facial challenge, the court conducts its analysis by reference to the statute's facial requirements and the undisputed, material facts relevant to the signature-matching process generally.Two cases are particularly helpful to the court's analysis. The first is Zessar v. Helander, No. 05-C-1917, 2006 WL 642646 (N.D. Ill. Mar. 13, 2006), in which the Northern District of Illinois considered an Illinois statute imposing a signature-match requirement for absentee voters. The process for obtaining and casting an absentee ballot in Illinois was similar to that in New Hampshire. A voter received an absentee ballot by executing and sending an absentee-ballot application to the appropriate authority. Id. at *1. After marking the ballot, the voter placed the ballot inside an envelope, on which was a certification form that the voter would sign and date. Id. On Election day, the package was delivered to the appropriate precinct, and the election judge "cast" the ballot for the absentee voter. Id. at *2. A ballot could be rejected if the signature on the application did not correspond to the signature on the ballot envelope or on the voter's registration card. Unlike New Hampshire law, a voter was mailed a notice after the election if his ballot was rejected. Id. at *3. Illinois law also had a procedure for provisional voting that could extend up to fourteen days after the election. Id. at *4.

The plaintiff in Zessar, whose vote was rejected due to a signature mismatch, sued a number of local and state officials, arguing that the lack of notice and an opportunity to cure for rejected voters was unconstitutional as a matter of procedural due process. See id. at *5. The district court agreed. On the first Mathews factor, it found that, while "[t]he right to vote by absentee ballot is not, in and of itself, a fundamental right," a voter has a sufficient liberty interest once "the State permits voters to vote absentee." Id.

On the second factor, the Zessar plaintiff suggested that voters should receive immediate notice of the rejection, followed by an informal administrative hearing in front of an election authority to confirm that the absentee ballot belongs to the voter. Id. at *8. The defendants responded that such procedures would be "hugely disproportionate" to the problem, and questioned whether they would even be effective for absentee voters who are absent from their residences for an extended period. Id. The Zessar court agreed that there was not a "tremendous" risk of erroneous deprivation, given that only approximately .43% of all absentee ballots returned to election authorities were rejected for any reason. See id. Nevertheless, the court found that the probable value of additional procedures was great in light of the otherwise irremediable denial of absentee voters' right to vote.

On the third factor, defendants argued that election authorities "face a cascade of statutory obligations in the time period leading up to and following the election," which would make additional procedures "an untenable burden." Id. at *9. The Zessar court was not persuaded. The court did recognize that new procedures would pose "some additional administrative and fiscal burden on the election authorities," but even so, the procedures would be fairly circumscribed affairs, given that election authorities have already verified that absentee voters are entitled to vote prior to issuing the absentee ballots. For that reason, the court found that any burden "would [not] be so great as to overwhelm plaintiff's interest in protecting his vote." Id.

Other courts have reached similar conclusions to that of the Zessar court. See Raetzel v. Parks/Bellemont Absentee Election Bd., 762 F.Supp. 1354 (D. Ariz. 1990) (concluding that Arizona absentee-voting statute, which failed to provide absentee voters with any post-deprivation notice or opportunity to be heard when their votes were challenged and rejected, did not afford adequate procedural due process); La Follette v. Padilla, No. CPF-17-515931, 2018 WL 3953766 (Cal. Super. Ct. Mar. 5, 2018) (concluding that provision of California Election Code, which required election officials to reject a ballot if the signatures did not "compare," was facially unconstitutional because it failed to provide pre-deprivation notice or an opportunity to cure), available at doc. no. 49-4.

Lemons v. Bradbury, 538 F.3d 1098 (9th Cir. 2008), is the only case the court could find in which a signature-match requirement was upheld against a procedural due process challenge based on the fundamental right to vote. There, the Ninth Circuit upheld Oregon's procedure for verifying signatures on referendum petitions. Oregon voters may approve legislation by referendum, and a referendum qualifies for statewide vote upon submission of a petition with a sufficient number of signatures. Lemons, 538 F.3d at 1100. In order to verify the petition signatures, the Secretary of State uses a statistical sampling method, whereby approximately five percent of the submitted signatures are cross-referenced with voter registration records. Id. If a petition signature is "genuine," it is counted. Id.

The plaintiffs argued that this procedure violated, among other things, their right to procedural due process. Id. at 1101. The Ninth Circuit disagreed, despite holding that Oregon's regulations on the referendum process "implicate the fundamental right to vote." Id. at 1102. The court determined that the value of additional procedures was negligible. Id. at 1105. The court reasoned that the verification process was "already weighted in favor of accepting questionable signatures," citing a number of elements of Oregon's procedure: (1) voters were notified on the referendum cover sheets that they must sign their name as they did on their voter registration; (2) the public could observe the process and object to signature-verification decisions; (3) all rejected signatures were subject to multiple layers of review; and (4) officials limited their review to a comparison between the petition signature and registration signature.

While the Ninth Circuit found additional procedures of negligible value, it assigned great weight to the administrative burden of additional procedures. Election officials might process more than 100,000 signatures in each election cycle, and it could take several minutes to "identify [each] signer, find the corresponding voter registration card, determine whether the signer is an active, registered voter, and then compare the signatures." Id. at 1104. The court therefore rejected the plaintiffs' claim, noting that "[r]equiring the state to provide thousands of petition signers with individual notice that their signatures have been rejected and to afford them an opportunity to present extrinsic evidence during the short thirty-day verification period would impose a significant burden on ... elections officials." Id. at 1104-05.

With these cases in mind, the court weighs the Mathews factors in the present case.

a. Private Interest

Plaintiffs argue that the individual interest at issue is the fundamental right to vote. Defendants respond that "the rights claimed by Plaintiff[s] should be afforded less weight than traditionally afforded the right to vote" because there is no right to vote by absentee ballot. Doc. no. 66 at 4.

The court accords this factor significant weight. It is beyond dispute that "[t]he right to vote is of the most fundamental significance under our constitutional structure." Ayers-Schaffner v. DiStefano, 37 F.3d 726, 727 (1st Cir. 1994) (internal quotation marks omitted); see also Griffin v. Burns, 570 F.2d 1065, 1075 (1st Cir. 1978) (noting that the right of suffrage is "a fundamental political right" because it is "preservative of all rights"). While "there is no corresponding fundamental right to vote by absentee ballot," Griffin v. Roupas, No. 02-C-5270, 2003 WL 22232839, at *4 (N.D. Ill. Sept. 22, 2003), the privilege of absentee voting is certainly "deserving of due process," Raetzel, 762 F.Supp. at 1358 ; accord Zessar, 2006 WL 642646, at *5 ; Doe v. Walker, 746 F.Supp.2d 667, 681 (D. Md. 2010). Having induced voters to vote by absentee ballot, the State must provide adequate process to ensure that voters' ballots are fairly considered and, if eligible, counted.

b. Risk of Erroneous Deprivation and Probable Value of Other Procedures

The parties dispute both elements of this factor. Plaintiffs assert that the risk of erroneous deprivation is "great" and that such risk could be easily remedied through additional procedures like a "simple telephone call." Doc. no. 48-1 at 34, 40. Defendants counter that the risk is "extremely slight" and that plaintiffs' alternative is not workable. Doc. no. 54 at 19, 22.

Defendants are correct that, based on the data available to the court, the overall rates of rejection due to a signature mismatch have been low in recent general elections. But those rates should be put into perspective. In the first place, even rates of rejection well under one percent translate to the disenfranchisement of dozens, if not hundreds, of otherwise qualified voters, election after election. See doc. no. 49-3 at 193-94 (deposition of David Scanlan) (stating that there is no indication that, in 2016 General Election, voters rejected due to signature mismatch were otherwise ineligible to vote). Given how close some races are in New Hampshire, that is a risk with real consequences. See, e.g., doc. no. 48-1 at 47 (noting one state senate race in 2016 that was decided by only seventeen votes).

More importantly, the task of handwriting analysis by laypersons, as it is contemplated under RSA 659:50, III, is fraught with error. Dr. Mohammed opines that individuals will naturally vary their signatures as a result of a number of intentional and unintentional factors. To account for such variations when conducting handwriting analysis, a person needs sufficient knowledge, training, equipment, and experience. The procedure under RSA 659:50, III, however, imposes none of these safeguards. Among other things, Dr. Mohammed notes that neither state law nor any guidance from state agencies sets forth functional standards for comparing signatures and assessing variations; election officials are not required to undergo any training in handwriting analysis; moderators are not screened for disabilities that may impair the ability to make such comparisons; moderators are not required to have proper magnification or lighting equipment; and moderators do not have sufficient time to conduct each comparison. Dr. Mohammed's uncontroverted conclusion is that, as a result, election officials are likely to make erroneous signature-comparisons. In fact, laypersons are more likely "to wrongly determine that authentic signatures are not genuine than to make the opposite error." Doc. no. 49-21 at 8.

The absence of functional standards is problematic, and the likelihood of error resulting therefrom is only compounded by the lack of meaningful review or oversight. There is no feedback mechanism to ensure that moderators are applying appropriate standards: neither voters nor the general public may object to a determination; there is no appeal or review process; and the Secretary of State does not regularly monitor rates of rejection to ensure that moderators are properly applying RSA 659:50, III. Furthermore, the absence of direct notice to affected voters is not only troubling in itself, see Gorman v. Univ. of Rhode Island, 837 F.2d 7, 12 (1st Cir. 1988) (stating that notice is one of the "essential requisites of procedural due process"), it also impairs voters' ability to monitor the conduct of moderators. In law and in practice, the ultimate determination is left to the sole discretion of the moderator and is almost entirely insulated from meaningful scrutiny. As Dr. Mohammed's undisputed conclusions establish, that total reliance on untrained laypersons entails tangible risks. RSA 659:50, III is thus a far cry from the procedure upheld in Lemons, where the public could object to determinations and there were multiple layers of review. Lemons, 538 F.3d at 1104.

It cannot be emphasized enough that the consequence of a moderator's decision-disenfranchisement-is irremediable. See Zessar, 2006 WL 642646, at *8-9. The one caveat is the procedure by which moderators evaluate extrinsic evidence to determine whether a ballot should be accepted notwithstanding a signature mismatch. But the evidence before the court shows that this safety valve as it currently exists is haphazard at best, since it is limited to the personal knowledge of, and information immediately available to, election officials. The instance where a moderator confirmed a rejection based on an assistant's knowledge of the voter's signature, which she had seen on medical documents, is illustrative.

Defendants counter that the risk of rejection will be significantly lower in future elections because the new absentee ballot application and affidavit envelope provide notice to the voter that the signatures must match. In support of this conclusion, they rely on Dr. Mohammed's analysis of signature styles in the 2016 General Election, which shows that 94 of the rejected ballots had signatures with different styles. Defendants argue that the new forms will "put voters on notice that they must use the same signature style when signing each document." Doc. no. 54 at 22. This is beside the point, however, because even excluding those 94 ballots, it still means that dozens, if not hundreds, of rejected voters used the same signature style and were still rejected in the 2016 General Election. Nor does mere notice ameliorate any of the problems identified by Dr. Mohammed that contribute to the risk of erroneous deprivation. The natural variations in a person's handwriting-many of which are unintentional or uncontrollable, like mental or physical condition-when combined with the absence of functional standards, training, review, and oversight, create a tangible risk of erroneous deprivation. Cf. League of United Latin Am. Citizens of Iowa v. Pate, No. CVCV056403 (Iowa Dist. Ct. July 24, 2018), available at doc. no. 71-1 at 18 (stating that "there is potential for erroneous determinations of a mismatch" under Iowa signature-match requirement for absentee ballots, where election officials had "unbridled discretion to reject ballots based on signatures they find do not match," but did not have "official guidance or handwriting expertise"), aff'd in part, No. 18-1276, 2018 WL 3946147 (Iowa Aug. 10, 2018), available at doc. no. 71-2 (affirming temporary injunction of Iowa signature-match requirement based on Iowa Constitution).

On the other hand, additional procedures would provide a tangible benefit. Plaintiffs point out that the absentee-ballot application already provides sections to allow a voter to submit her phone number and email address. They contend that a procedure whereby a moderator simply reaches out to the voter in one form or another would be of great value. The court agrees.

As defendants stated at the hearing, moderators already engage in a process of considering extrinsic evidence to, in defense counsel's words, "salvage" a ballot that could otherwise be rejected. Necessarily, the premise of such a process is that the consideration of extrinsic evidence can be useful in determining whether the same person executed both the affidavit envelope and application. Plaintiffs seek no more than to open up that process to allow for consideration of evidence from the best source-the voter. To be sure, it may not be a perfect solution. Because many absentee voters are voting by absentee ballot due to work, religious commitment, or disability, they may not be available at the time the moderator is reviewing the signatures. Some may not be reachable by phone, and others may not have access to email. But with proper notice to voters that they may be contacted, a phone call or similar measure would make the process a more constructive exercise. As it stands currently, moderators consider limited and far less probative extrinsic evidence. In addition, there is evidence in the record that moderators consider evidence submitted with the affidavit envelope in deciding whether to accept an absentee ballot. That could provide an option to absentee voters who know they will be unreachable on Election Day and wish to ensure that their ballots will be accepted.

In short, based on the undisputed facts in the record, "[i]t is apparent that the risk of erroneous deprivation of the protected interest in absentee voting is not enormous, but the probable value of an additional procedure is likewise great in that it serves to protect the fundamental right to vote." Zessar, 2006 WL 642646, at *9.

c. Government's Interests

The third factor involves consideration of the government's interests, which may include "the function involved and the fiscal and administrative burdens that the additional or substitute procedural requirement would entail." Mathews, 424 U.S. at 335, 96 S.Ct. 893. Defendants argue that RSA 659:50, III furthers the State's interest in preventing voter fraud and protecting public confidence in the integrity of the electoral process. Further, defendants contend that, because the New Hampshire Constitution requires that all votes be counted on Election Day, it would be "extremely burdensome" to require moderators "to seek out and verify the identity of hundreds of absentee voters." Doc. no. 54 at 20.

The court agrees that the State has legitimate interests in preventing voter fraud and protecting public confidence in elections. See Crawford v. Marion Cty. Election Bd., 553 U.S. 181, 194-97, 128 S.Ct. 1610, 170 L.Ed.2d 574 (2008). That being said, in comparison to the two instances of absentee-voter fraud that defendants cite as support, one in the 2012 General Election and the other in the 2016 General Election, hundreds of voters (approximately 740 by plaintiffs' estimate) were disenfranchised under RSA 659:50, III in the 2012, 2014, and 2016 General Elections. See doc. no. 48-1 at 21-22. Furthermore, plaintiffs point out that neither instance of absentee-voter fraud was uncovered through the signature-matching process.

In any case, the court fails to see how additional procedures would harm these interests. Moderators already consider limited forms of extrinsic evidence. Additional procedures would simply allow for more probative extrinsic evidence to be considered. Thus, if anything, additional procedures further the State's interest in preventing voter fraud while ensuring that qualified voters are not wrongly disenfranchised. See Fla. Democratic Party v. Detzner, No. 4:16cv607, 2016 WL 6090943, at *7 (N.D. Fla. Oct. 10, 2016) ("[L]etting mismatched-signature voters cure their vote by proving their identity further prevents voter fraud-it allows supervisors of elections to confirm the identity of that voter before their vote is counted."). Likewise, improving the currently opaque, unreviewable process by which moderators compare signatures and consider extrinsic evidence would only serve to enhance voter confidence in elections. Two of the plaintiffs have expressed their anger and frustration at the treatment of their votes under the current system. See doc. no. 48-2 at 2-3; doc. no. 48-3 at 3.

Next, contrary to defendants' argument, plaintiffs' proposed procedures would not entail significant administrative burdens. Moderators already engage in a practice of considering extrinsic evidence before rejecting a ballot due to a signature mismatch. Consequently, this is a case not of foisting wholly novel procedures on state election officials, but of simply refining an existing one to allow voters to participate and to ensure that the process operates with basic fairness.

While the current procedure would need to be expanded so that moderators could reach out to voters, no individual district is likely to be materially impacted. In the 2016 General Election, for example, 74% of polling places would not have been impacted at all, because they did not reject any ballots due to a signature mismatch. Only 5 of New Hampshire's 318 polling places had ten or more such rejections. That procedures already exist which could be readily extended to provide basic guarantees of due process to voters militates against defendants' argument. See Detzner, 2016 WL 6090943, at *7 (finding statute requiring rejection of absentee ballot based on signature mismatch unconstitutional, where, among other things, voters who failed to sign their ballot received opportunity to cure, but voters whose signatures did not match received no such opportunity); La Follette, doc. no. 49-4, at 4 (same).

Defendants respond that the analysis is not so simple. They posit that, "[i]f plaintiffs' argument was extended to its logical conclusion, ... then notice and an opportunity to cure could be required for every absentee ballot rejected for any reason," further burdening moderators. Doc. no. 66 at 5. The court is not persuaded. The court's conclusion rests on a careful, context-sensitive balancing of the Mathews factors. See, e.g., Mathews, 424 U.S. at 334, 96 S.Ct. 893 ("(D)ue process is flexible and calls for such procedural protections as the particular situation demands."). The need for additional process with respect to RSA 659:50, III does not inevitably lead to the conclusion that other reasons for rejection will demand similar process.

Defendants make two final arguments that merit only brief comment. First, on the basis of their statistical evidence allegedly showing the success of their trainings, defendants argue that at most moderators merely applied the statute unconstitutionally in 2016. They argue that the proper remedy would be to require them to conduct additional training for those moderators who have higher rejection rates. The court is not persuaded. Defendants' statistical evidence is insufficient for the reasons articulated by plaintiffs in their briefing. Furthermore, additional training would not be a useful exercise, since defendants merely reiterate the limited guidance set forth in the Election Procedure Manual. See note 9, supra.

Second, defendants argue that plaintiffs cannot succeed on a facial challenge because they do not dispute that the absentee ballots of some voters-specifically, those who (1) omit signatures, (2) use the wrong name on a document, or (3) use a digital signature-were correctly rejected. But "[t]he proper focus of the constitutional inquiry is the group for whom the law is a restriction, not the group for whom the law is irrelevant." City of Los Angeles, California v. Patel, ––– U.S. ––––, 135 S.Ct. 2443, 2451, 192 L.Ed.2d 435 (2015). Voters in those categories would be rejected regardless of RSA 659:50, III, as they would fail to meet independent requirements of RSA 659:50. See RSA 659:50, II (requiring that the affidavit be properly executed); RSA 659:50, IV (requiring that the signatures appear to be the signatures of a duly qualified voter). For the same reason, defendants are incorrect when they argue that striking down RSA 659:50, III would prevent moderators from rejecting ballots that are unsigned or signed by a person with a different name than the voter.

d. Conclusion

"Procedural due process guarantees fair procedure, not perfect, error-free determinations." Aurelio v. R.I. Dep't of Admin., Div. of Motor Vehicles, 985 F.Supp. 48, 56 (D.R.I. 1997) (internal quotation marks omitted). One could, by taking in isolation various facets of the current process, conclude that RSA 659:50, III passes constitutional muster. But taken as a whole, and in light of the fundamental importance of the right to vote, the current process for rejecting voters due to a signature mismatch fails to guarantee basic fairness. The infirmity with the statute begins with vesting moderators with sole, unreviewable discretion to reject ballots due to a signature mismatch. Such discretion becomes constitutionally intolerable once other factors are taken into account: the natural variations in voters' signatures combined with the absence of training and functional standards on handwriting analysis, and the lack of any review process or compliance measures. And there is an easy fix-an existing procedure that, with minor refinement, could reduce the risk that qualified voters are wrongly disenfranchised and bolster the State's interests in preventing voter fraud and safeguarding voter confidence in elections.

Therefore, in light of the undisputed, material facts in the record, plaintiffs are entitled to summary judgment on their procedural due process claim. The court grants plaintiffs' request for declaratory relief insofar as RSA 659:50, III is unconstitutional under the Fourteenth Amendment of the United States Constitution.

The court also grants plaintiffs' request for permanent injunctive relief. Outside of their arguments on the merits, defendants do not argue that the elements for a permanent injunction are not satisfied. See Esso Standard Oil Co. v. López-Freytes, 522 F.3d 136, 148 (1st Cir. 2008) (listing elements for permanent injunction). The court concludes that plaintiffs have demonstrated an entitlement to permanent injunctive relief. Because "a successful facial attack means the statute is wholly invalid and cannot be applied to anyone," Ezell, 651 F.3d at 698, the court enjoins defendants from enforcing RSA 659:50, III. Although the enforcement of the provision falls primarily on local election officials, i.e., nonparties, the court is confident that the Secretary of State will take appropriate steps to ensure that this injunction is enforced.

II. Remaining Claims (Counts II, III, IV)

Having resolved Count I in plaintiffs' favor, the court declines to go further and address plaintiffs' remaining claims. This is because plaintiffs are afforded complete relief by virtue of their success on their procedural due process claim: they will receive the declaratory relief they request and corresponding injunctive relief.

This is not to say that plaintiffs' remaining constitutional claims are "moot" in the technical sense. A case does not become moot for purposes of Article III merely because a court finds that the plaintiff is entitled to relief under one of a number of alternative theories. See Novella v. Westchester Cty., 661 F.3d 128, 149 (2d Cir. 2011) ("[I]t is cases rather than reasons that become moot with the meaning of Article III." (internal quotation marks omitted) ); Air Line Pilots Ass'n, Int'l v. UAL Corp., 897 F.2d 1394, 1397 (7th Cir. 1990) ("Whether a court gives one or ten grounds for its result is not a question to which Article III prescribes an answer."). But courts also use the term "moot" to "refer to an issue that need not be decided in light of the resolution in the same opinion of another issue." UAL Corp., 897 F.2d at 1397. The decision to reach or avoid an unnecessary issue falls within the court's discretion, and will depend on the circumstances presented. See Novella, 661 F.3d at 149 ; Clark v. Dep't of Army, 997 F.2d 1466, 1470 (Fed. Cir. 1993), superseded by statute on other grounds as recognized in Kewley v. Dep't of Health & Human Servs., 153 F.3d 1357, 1361-62 (Fed. Cir. 1998). "Courts recognize that it may be valuable to decide alternative grounds, even though not all are necessary, and also understand that there may be excellent reasons to avoid alternative grounds." 13B Charles Alan Wright, Arthur R. Miller & Edward H. Cooper, Federal Practice & Procedure § 3533 (3d ed.) (collecting cases).

In this case, the court concludes that it is neither necessary nor prudent to reach plaintiffs' other constitutional claims. In Count II, plaintiffs target the same basic defects in the statute as they do in Count I, albeit under the auspices of a different test. And although Count III presents a distinct question, as it rests on the alleged lack of uniformity wrought by RSA 659:50, III, it raises sensitive legal questions, including the applicability of Bush v. Gore to this context, the standard by which such a claim should be evaluated on a facial challenge, etc. While these claims may not be moot under Article III, the same considerations undergirding the constitutional rule-judicial economy and the reluctance to opine on abstract propositions-support the court's decision here. See Thomas R.W. v. Mass. Dep't of Educ., 130 F.3d 477, 479 (1st Cir. 1997) ; Soto v. City of Cambridge, 193 F.Supp.3d 61, 69 (D. Mass. 2016).

By contrast, plaintiffs' ADA claim is moot under Article III. As a result of the 2017 amendments to RSA 659:50, III, an absentee voter who receives assistance "because the voter is blind or has a disability" is exempt from the requirement. RSA 659:50, III. In her deposition, Ms. Saucedo testified that she will "definitely" rely on her husband's assistance when she votes in the future. Doc. no. 49-5 at 7 (brackets omitted). Therefore, Ms. Saucedo-the only plaintiff with a disability-is and will be exempt from RSA 659:50, III, and consequently no longer has a legally cognizable interest in the outcome of the ADA claim. See Horizon Bank & Trust Co. v. Massachusetts, 391 F.3d 48, 53 (1st Cir. 2004) ("A case is moot when the issues are no longer live or the parties no longer have a legally cognizable interest in the outcome."). Put differently, because the court can no longer give any effectual relief to Ms. Saucedo on this claim, the claim is moot and the court may not entertain it. See id.; cf. Steir v. Girl Scouts of the USA, 383 F.3d 7, 16 (1st Cir. 2004) ("A federal court may not entertain a claim by any or all citizens who no more than assert that certain practices of officials are unlawful." (internal brackets omitted) ).

For these reasons, the court does not address Counts II, III, and IV.

CONCLUSION

For the reasons stated herein, plaintiffs' motion for summary judgment (doc. no. 48) is granted with respect to Count I, and is otherwise denied. Summary judgment on Counts II and III is denied in light of the complete relief afforded to plaintiffs on Count I, and summary judgment on Count IV is denied because the claim is moot. Counts II, III, and IV are therefore dismissed without prejudice. Defendants' cross-motion for summary judgment (doc. no. 52) is granted on Count IV to the extent that Count IV is dismissed without prejudice as moot, and is otherwise denied. Plaintiffs' motion to strike (doc. no. 56) is denied.

As to relief under Count I, the court grants plaintiffs' request for declaratory relief insofar as RSA 659:50, III is unconstitutional under the Fourteenth Amendment of the United States Constitution. The court also grants plaintiffs' request for permanent injunctive relief, and defendants are hereby permanently enjoined from enforcing RSA 659:50, III.

SO ORDERED.