United Food and Com. Workers U. Loc. 120 v. Wal-Mart Stores

Full title: UNITED FOOD AND COMMERCIAL WORKERS UNION LOCAL 120, UNITED FOOD AND…

Court: United States District Court, N.D. California

Date published: Jun 21, 2004

222 F.R.D. 137 (N.D. Cal. 2004)

Opinion

[Copyrighted Material Omitted]

Charles V. Firth, Tinkler & Firth, Santa Fe, NM, Christine E. Webber, Deborah Vagins, Joseph Sellers, Cohen, Milstein, Hausefeld & Toll, Washington, DC, Debra Gardner, Jonathan Smith, Public Justice Center, Baltimore,

MD, Shauna Marshall, Hastings College of Law, San Francisco, CA, Brad Seligman, Jocelyn D. Larkin, The Impact Fund, Berkeley, CA, Doris Ng, Equal Rights Advocates, San Francisco, CA, Elizabeth Lawrence, Steve Stemerman, Davis, Cowell & Bowe, San Francisco, CA, Merit Bennett, Santa Fe, NM, Stephen Tinkler, Tinkler & Bennett, Santa Fe, NM, Sheila Thomas, Equal Rights Advocates, Oakland, CA, for Plaintiffs.

Nancy L. Abell, Katherine C. Huibonhoa, Paul, Hastings, Janofsky & Walker, LLP, Los Angeles, CA, Barbara L. Johnson, Neal D. Mollen, Paul, Hastings, Janofsky & Walker, LLP, Washington, DC, for Defendant.

ORDER GRANTING IN PART AND DENYING IN PART MOTION FOR CLASS CERTIFICATION

JENKINS, District Judge.

TABLE OF CONTENTS

INTRODUCTION

141

LEGAL STANDARD

143

ANALYSIS

144

I.

Rule 23(a) Factors

144

A.

Numerosity

144

B.

Commonality

144

1.

Company-Wide Policies and Practices

145

a.

Policies and Practices Governing Compensationand Promotion

145

(1)

Overview of Wal-Mart Store Structure

145

(2)

Policies Governing Compensation

146

(a)

Hourly Employees

146

(b)

Salaried Employees

147

(3)

Policies Governing Promotion Decisions

148

(a)

Subjectivity in the Selection Practice

148

(b)

Failure to Post PromotionalOpportunities

149

(4)

Whether Defendant's Compensation andPromotion Policies Support a Finding ofCommonality

149

b.

Corporate Culture

151

(1)

Wal-Mart's Emphasis on UniformCulture

151

(2)

Gender Stereotyping

153

2.

Statistical Evidence of Discrimination

154

a.

Statistical Evidence RegardingCompensation

155

(1)

Overview

155

(2)

Defendant's Challenges to Plaintiffs'Statistical Analysis

156

(a)

Aggregation at the Regional Level

156

(b)

Selection of Variables

159

b.

Statistical Evidence RegardingPromotions

160

(1)

Internal Data Analysis

160

(a)

Overview

160

(b)

Defendant's Challenge to Dr. Drogin'sChoice of Applicant Pools

161

(2)

External Benchmarking Data Analysis

164

3.

Anecdotal Evidence of Discrimination

165

4.

Conclusion re Commonality

166

C.

Typicality

166

1.

Whether the Named Plaintiffs Can Represent AllIn-Store Managers

166

2.

Degree of Individual Specificity for EachClaim

167

D.

Adequacy of Representation

168

1.

Conflicts of Interest

168

2.

Qualified Counsel

169

E.

Conclusion

169

II.

Maintainability Under Rule 23(b)

169

A.

Inclusion of Claim for PunitiveDamages

170

B.

Manageability of Liability and RemedyStages

173

1.

Manageability of Liability Stage

173

2.

Manageability of Remedy Phase

174

a.

Promotions Claim

175

(1)

Overview of Traditional and Formula Approach toBackpay Remedy

175

(2)

Application of Class-Wide FormulaApproach

178

(a)

Using a Formula to Calculate a Lump Sum BackpayAward

178

(b)

Determining Individual Eligibility

179

b.

Equal Pay Claim

183

(1)

Identification of victimized classmembers

183

(2)

Calculation of individual backpayawards

185

c.

The 1991 Civil Rights Act

186

CONCLUSION

187

INTRODUCTION

Plaintiffs have filed a Third Amended Complaint, brought on behalf of six named plaintiffs and all others similarly situated, asserting a claim against Wal-Mart Stores, Inc. (" Wal-Mart" ) for sex discrimination under Title VII of the 1964 Civil Rights Act, 42 U.S.C. § 2000e et seq. (" Title VII" ). Generally stated, plaintiffs allege that women employed in Wal-Mart stores (1) are paid less than men in comparable positions, despite having higher performance ratings and greater seniority; and (2) receive fewer promotions to in-store management positions than do men, and those who are promoted must wait longer than their male counterparts to advance.

Plaintiffs assert that the policies and practices underlying this discriminatory treatment are consistent throughout Wal-Mart, and that the discrimination of which they complain is common to all women who work or have worked in Wal-Mart stores. Plaintiffs seek class-wide injunctive and declaratory relief, lost pay, and punitive damages. They do not seek any compensatory damages on behalf of the class.

Wal-Mart is the largest private employer in the world. See Defendant's Opposition to Motion for Class Certification (" Def.'s Opp'n" ) at 1. It operates approximately 3,400 stores in the United States and currently employs well over a million people. See Declaration of Christine Webber (" Webber Decl." ) Ex. 70 at 3-5, 12. Wal-Mart is best known for its " Discount Stores," which offer a wide variety of discounted goods and services, and " Supercenters," which are similar to Discount Stores but also include full grocery departments. The company also operates " Sam's Clubs," which are membership-only stores that sell items in bulk or at deep discounts, and " Neighborhood Markets," which are smaller stores primarily selling food and drugs.¹

1 At the time this motion was briefed, Wal-Mart operated 1,568 Discount Stores nation-wide, each of which employed at least 150 individuals; 1,259 Supercenters (making Wal-Mart the nation's largest grocer), with 200 to 500 employees per store; 525 Sam's Clubs, with 120 to 225 employees per store; and 49 Neighborhood Markets, with 80 to 100 employees per store. Webber Decl., Ex. 70 at 3-5; Def.'s Opp'n at 13.

Currently before the Court is Plaintiffs' motion to certify a nation-wide class of women who have been subjected to Wal-Mart's allegedly discriminatory pay and promotions policies.² Specifically, plaintiffs propose that the Court certify the following class pursuant to Federal Rule of Civil Procedure 23(a) and 23(b)(2):

2 Also before the Court are various motions to strike submitted by both parties. The Court rules on these motions separately in its Order Granting in Part and Denying in Part Plaintiffs' and Defendant's Motions to Strike Expert and Non-expert Testimony (" Order re Motions to Strike" ), filed concurrently with the present order.

All women employed at any Wal-Mart domestic retail store at any time since December 26, 1998 who have been or may be subjected to Wal-Mart's challenged pay and management track promotions policies and practices.

Plaintiffs' Motion for Class Certification (" Pls.' Mot." ) at 37.³ In preparation for this motion, the parties undertook extensive discovery. They also filed extensive briefing along with volumes of documentary and testimonial evidence. On September 24, 2003, the Court heard oral argument over the course of seven hours.

3 In their briefs, Plaintiffs note that they are not challenging decisions on promotions into, or compensation for, the position of Store Pharmacist, which is subject to special educational and state licensing requirements. Pls.' Mot. at 37 n. 30. Accordingly, this position would be excepted from the class. Plaintiffs also propose to sever the racial discrimination claims alleged by two named plaintiffs and exclude these claims from the class case. Id. at 48 n. 36.

Before proceeding further, the Court wishes to acknowledge Defendant's characterization of this motion as historic in nature. Defendant emphasizes that the proposed class covers at least 1.5 million women who have been employed over the past five years at roughly 3,400 stores, thus dwarfing other employment discrimination cases that have come before. In its view, these numbers alone make this case impossible. Certainly, the size of the putative class raises concerns regarding manageability which this Court must, and does, carefully consider. Title VII, however, contains no special exception for large employers. Enacted in 1964 during the height of the civil rights movement, this Act forbids gender and race-based discrimination in the American workplace. Two years later, federal class action rules were amended to facilitate the vindication of these rights on a broader basis. SeeIn re Monumental Life Ins. Co., 365 F.3d 408, 417 n. 16 (5th Cir.2004) ( Fed.R.Civ.P. 23(b)(2) was added in 1966 primarily to facilitate the bringing of civil rights class actions). Insulating our nation's largest employers from allegations that they have engaged in a pattern and practice of gender or racial discrimination-simply because they are large-would seriously undermine these imperatives. Indeed, it is interesting to note, as a matter of historical perspective, that Plaintiffs' request for class certification is being ruled upon in a year that marks the 50th anniversary of the Supreme Court's decision in Brown v. Board of Education, 347 U.S. 483, 74 S.Ct. 686, 98 L.Ed. 873 (1954). This anniversary serves as a reminder of the importance of the courts in addressing the denial of equal treatment under the law wherever and by whomever it occurs. Id. at 495, 74 S.Ct. 686.

Notions of historical import aside, the Court also takes this opportunity to acknowledge two important factors. First, the Court, by this motion, is not called upon to make any determination on the merits of Plaintiffs' allegations of gender discrimination; rather, only procedural questions are presented. SeeEisen v. Carlisle & Jacquelin, 417 U.S. 156, 177, 94 S.Ct. 2140, 40 L.Ed.2d 732 (1974). Thus, resolution of this motion for class certification should not be construed in any manner as a ruling on the merits or the probable outcome of the case. Second, in spite of Defendant's concerns regarding the historic nature of this motion, the Court does not encounter the issues raised herein on a barren legal landscape. Rather, this Court is satisfied that ample legal precedent is available to guide the Court in determining the propriety of Plaintiffs' request for class certification. Indeed, while the size of the proposed class is unique, the issues are not novel, and Plaintiffs' claims are relatively narrow in scope.⁴ Having fully analyzed the pleadings, evidence, and oral argument presented by the parties, and the entire record herein, the Court grants in part and denies in part Plaintiffs' motion for class certification. Specifically, and as explained in greater detail below, the Court rules as follows:

4 Although Defendant characterizes this case as a " wall-to-wall" and " across-the-board" class action, Def.'s Opp'n at 1, it overstates the breadth of the case. While, as noted, the proposed class size is extremely large, the types of claims asserted are relatively limited in the context of other judicially sanctioned class actions. As described above, Plaintiffs challenge Defendant's alleged discriminatory actions solely pursuant to Title VII, without attempting to combine multiple legal claims through other statutes or constitutional provisions, as is common in other cases. The class is also limited to women; plaintiffs do not attempt to combine gender with race, age, disability, or other classifications. Further, Plaintiffs allege class-wide discrimination only in terms of pay and promotion-not hiring, hostile work environment, failure to train, retaliation, or other adverse employment actions. The class also is limited to certain categories of in-store employees, and does not seek to include other categories of workers; importantly, the proposed class does not reach into the upper levels of management. Nor does it seek compensatory damages. Thus, while this is a " wall-to-wall" case geographically speaking, in all other respects it is substantially more limited than Defendant suggests.

1) With respect to Plaintiffs' claim for equal pay, Plaintiffs' motion is granted and the proposed class is certified with respect to issues of liability and all forms of requested relief;

2) With respect to Plaintiffs' promotion claim, Plaintiffs' motion is granted in part and denied in part. Specifically, the Court certifies the proposed class with respect to issues of liability (including liability for punitive damages) and injunctive and declaratory relief. The Court finds, however, that with respect to the remedy of lost pay, it is manageable only with respect to those challenged promotions where objective data is available to document class member interest in the challenged promotion. Thus, the Court denies certification, on grounds of unmanageability, with respect to Plaintiffs' promotion claims for lost pay (and thus punitive damages) as to those class members for whom no such data is available.

LEGAL STANDARD

A party seeking to certify a class must demonstrate that it has met all four requirements of Federal Rule of Civil Procedure 23(a), and at least one of the requirements of Rule 23(b). Zinser v. Accufix Research Inst., Inc., 253 F.3d 1180, 1186 (9th Cir.2001). Rule 23(a) requires that all of the following four factors be met: " (1) the class is so numerous that joinder of all members is impracticable, (2) there are questions of law or fact common to the class, (3) the claims or defenses of the representative parties are typical of the claims or defenses of the class, and (4) the representative parties will fairly and adequately protect the interests of the class." In short, the class must satisfy the requirements of numerosity, commonality, typicality, and adequacy. Fed.R.Civ.P. 23(a).

Rule 23(b) requires, in relevant part, that one of three additional requirements be met. Here, plaintiffs assert that this case falls within Rule 23(b)(2), which provides that the " party opposing the class has acted or refused to act on grounds generally applicable to the class, thereby making appropriate final injunctive relief or corresponding declaratory relief with respect to the class as a whole." Fed.R.Civ.P. 23(b)(2). Rule 23(c) allows a court to maintain a class action as to particular issues only or to divide a class into subclasses. Fed.R.Civ.P. 24(c)(4).

The party seeking certification must provide facts sufficient to satisfy Rule 23(a) and (b) requirements. Doninger v. Pacific Northwest Bell, Inc., 564 F.2d 1304, 1308-09 (9th Cir.1977). In turn, the district court must conduct a rigorous analysis to determine that the prerequisites of Rule 23 have been met. Gen. Tel. Co. v. Falcon, 457 U.S. 147, 161, 102 S.Ct. 2364, 72 L.Ed.2d 740 (1982). If a court is not fully satisfied, certification should be refused. Id. Conditional certification is no longer proper. See Fed.R.Civ.P. 23, Advisory Committee Notes, 2003 Amendments.

While the court's analysis must be rigorous, Rule 23 confers " broad discretion to determine whether a class should be certified, and to revisit that certification throughout the legal proceedings before the court." Armstrong v. Davis, 275 F.3d 849, 872, n. 28 (9th Cir.2001); see alsoKnight v. Kenai Peninsula Borough Sch. Dist., 131 F.3d 807, 816 (9th Cir.1997). " ‘ Implicit in this deferential standard is recognition of the essentially factual basis of the certification inquiry and of the district court's inherent power to manage and control pending litigation.’ " In re Monumental Life Ins. Co., 365 F.3d at 414 (citation omitted); see alsoHartman v. Duffey, 19 F.3d 1459, 1471 (D.C.Cir.1994) (" In light of the fact that trial courts have the primary responsibility of ensuring the ‘ orderly management of litigation’ and that the purpose of class actions lies in ‘ advanc[ing] the efficiency and economy of multi-party litigation,’ trial courts ‘ are uniquely well situated to make class certification decisions.’ " ) (citations omitted).

In Falcon, the Court reiterated the well-recognized precept that " ‘ the class determination generally involves considerations that are enmeshed in the factual and legal issues comprising the plaintiff's cause of action.’ " Falcon, 457 U.S. at 160, 102 S.Ct. 2364 ( quotingCoopers & Lybrand v. Livesay, 437 U.S. 463, 469, 98 S.Ct. 2454, 57 L.Ed.2d 351 (1978)). Nevertheless, " [w]e find nothing in either the language or history of Rule 23 that gives a court any authority to conduct a preliminary inquiry into the merits of a suit in order to determine whether it may be maintained as a class action." Eisen, 417 U.S. at 177, 94 S.Ct. 2140. Thus, " [a]lthough some inquiry into the substance of a case may be necessary to ascertain satisfaction of the commonality and typicality requirements of Rule 23(a), it is improper to advance a decision on the merits to the class certification stage." Moore v. Hughes Helicopters, Inc., 708 F.2d 475, 480 (9th Cir.1983) (citation omitted); see alsoNelson v. United States Steel Corp., 709 F.2d 675, 679-80 (11th Cir.1983) (plaintiffs' burden at class certification " entails more than the simple assertion of [commonality and typicality] but less than a prima facie showing of liability" ) (citation omitted); Caridad v. Metro-North Commuter R.R., 191 F.3d 283, 292 (2d Cir.1999); Fed.R.Civ.P. 23, Advisory Committee Notes, 2003 Amendments (court review of the merits should be limited to those aspects relevant to making the certification decision on an informed basis).⁵ With these principles in mind, the Court turns to the requirements of Rule 23(a).

5 The restriction on conducting a merits inquiry applies equally to the Court's review of the expert testimony presented by the parties. Rather than resolving the " battle of the experts," and without conclusively ruling on admissibility, the Court's role at the class certification stage is to determine whether the expert evidence adds probative value to plaintiffs' claims. See Thomas & Thomas Rodmakers, Inc. v. Newport Adhesives and Composites, Inc., 209 F.R.D. 159, 162-63 (C.D.Cal.2002).

ANALYSIS

I. Rule 23(a) Factors

A. Numerosity

Under the first Rule 23(a) factor, the class must be " so numerous that joinder of all members is impracticable." Fed.R.Civ.P. 23(a)(1); Staton v. Boeing Co., 327 F.3d 938, 953 (9th Cir.2003). Plaintiffs need not state the exact number of potential class members; nor is a specific minimum number of class members required. Arnold v. United Artists Theatre Circuit, Inc., 158 F.R.D. 439, 448 (N.D.Cal.1994). Rather, whether joinder is impracticable depends on the facts and circumstances of each case. Id. Wal-Mart does not contest that numerosity is satisfied here, given that both parties estimate that the proposed class numbers over one million women. Indeed, it is beyond dispute that joinder would be impracticable in this case. Accordingly, the Court finds that this factor is satisfied.

B. Commonality

Rule 23(a)(2) requires that common questions of law or fact exist among class members. As the Supreme Court has explained, this requirement overlaps with the typicality requirement:

[T]he commonality and typicality requirements ... tend to merge. Both serve as guideposts for determining whether under the particular circumstances maintenance of a class action is economical and whether the named plaintiff's claim and the class claims are so interrelated that the interests of the class members will be fairly and adequately protected in their absence.

Falcon, 457 U.S. at 157 n. 13, 102 S.Ct. 2364. Nonetheless, each factor serves a discrete purpose-commonality focuses on the relationship of common facts and legal issues among class members, while typicality examines the relationship of facts and issues between the representatives and the class. See Conte & Newberg on Class Actions, Vol. I, at 411 (4th ed.) (hereafter " Newberg" ). In other words, commonality refers to the group characteristics of the class as a whole, and typicality refers to the individual characteristics of the named plaintiffs in relation to the class. As such, the Court will address each factor separately.

Rule 23(a)(2) does not require that all questions of law or fact be common. The test is qualitative rather than quantitative-one significant issue common to the class may be sufficient to warrant certification. Savino v. Computer Credit, Inc., 173 F.R.D. 346, 352 (E.D.N.Y.1997), aff'd, 164 F.3d 81 (2d Cir.1998); Newberg, Vol. I, at 272-74. Indeed, the necessary showing to satisfy commonality is " minimal." Hanlon v. Chrysler Corp., 150 F.3d 1011, 1020 (9th Cir.1998). Specifically, plaintiffs may demonstrate commonality by showing that class members have shared legal issues but divergent facts or that they share a common core of facts but base their claims for relief on different legal theories. Id. at 1019.

Plaintiffs present extensive evidence to support their contention that there are significant factual and legal questions common to all class members with respect to whether Wal-Mart has engaged in company-wide discrimination against female in-store employees in pay and promotions. This evidence can be grouped into three major categories: (1) facts and expert opinion supporting the existence of company-wide policies and practices; (2) expert statistical evidence of class-wide gender disparities attributable to discrimination; and (3) anecdotal evidence from class members around the country of discriminatory attitudes held or tolerated by management. The Court concludes that this evidence more than satisfies plaintiffs' burden to demonstrate commonality under Rule 23(a)(2). Each category is discussed in detail below.

1. Company-Wide Policies and Practices

This first category can be further divided into two discrete areas. First, Plaintiffs present evidence that Wal-Mart's policies governing compensation and promotions are similar across all stores, and build in a common feature of excessive subjectivity which provides a conduit for gender bias that affects all class members in a similar fashion. Second, Plaintiffs submit evidence that Wal-Mart cultivates and maintains a strong corporate culture which includes gender stereotyping. While the parties vigorously dispute certain aspects of this evidence, and certainly draw different inferences therefrom, it is sufficient at this juncture to support Plaintiffs' claim that there are common issues as to whether Wal-Mart engages in discriminatory policies and practices that affect putative class members in a similar manner.

a. Policies and Practices Governing Compensation and Promotion

Defendant argues that it utilizes a myriad of distinct systems for compensating and promoting in-store employees depending on the position and type of store (or store department), and that these differences preclude any finding of commonality. It further argues that decisions regarding pay and promotions are too decentralized to create any questions common to the class. As discussed below, however, while some variations exist, there is a basic organizational structure that is consistent across store types and throughout the company's domestic stores in important respects. Furthermore, the policies governing in-store compensation and promotions uniformly provide for managers to exercise significant subjectivity in making pay and promotion decisions. Indeed, although the parties disagree about many of the fine points, they essentially agree that Wal-Mart managers make pay and promotion decisions for in-store employees in a largely subjective manner.

(1) Overview of Wal-Mart Store Structure

As noted above, Wal-Mart operates four different types of stores: Discount Stores, Supercenters, Neighborhood Markets, and Sam's Clubs. Wal-Mart considers the first three to be part of a single operation, while Sam's Club operates as a separate division. All stores (except Sam's Clubs) are divided into six divisions, for a total of seven divisions including Sam's Clubs. Each division in turn is divided into regions for a total of 41 regions nationwide. Each region contains roughly 80-85 stores. K. Harper Depo. at 177, 215 (Webber Decl. Ex. 1); Butler Depo. at 39 (Webber Decl. Ex. 4).

Merchandise sold in the stores is generally segregated in up to 40-53 separate departments (e.g.-apparel, jewelry, hardware).⁶ Certain departments are designated " specialty departments" and are considered to operate as semi-autonomous units within the stores. The eight specialty departments are One-Hour Photo, Optical, Pharmacy, Shoes, Jewelry, Tire & Lube Express, Hearing, and Wireless Services. Grocery sections in the Supercenters also are considered to be semi-autonomous units within the stores. K. Harper Depo. at 90, 155-56 (Webber Decl. Ex. 1).

6 Because of their relatively small size, Neighborhood Markets do not have departments but instead are " organized into areas." Haworth Decl. ¶ 57.

While the size and inventory of Wal-Mart stores vary, the evidence indicates that the personnel structure within each store operates in a basically similar fashion using similar job categories, job descriptions, and management hierarchies.⁷ At the top of the in-store management structure are the salaried positions beginning with the Store Manager. See K. Harper Depo. at 152 (Webber Decl. Ex. 1). In certain larger stores, there are Co-Managers who report to the Store Manager and often oversee the grocery departments. Seeid. at 35-36, 155-56; Haworth Decl. ¶ 50. Next in line are the Assistant Managers, with several per store, depending on store size. Specialty Department Managers report directly to District Managers as well as the Store Manager.

7 Some divisions may use slightly different nomenclature. For example, Sam's Club follows essentially the same hierarchy as the other Wal-Mart divisions, but uses different titles for some positions. Thus, the head of a Discount Store or Supercenter is called a Store Manager, while the head of a Sam's Club is called a General Manager; yet both positions have virtually identical duties. See Burner Depo. at 144:16-24 (Webber Decl. Ex. 5).

In order to transition into a salaried management position, employees must generally go through a four-to-five month training program as a " Management Trainee." See Def.'s Opp'n at 40. The Management Trainees are considered as being in a limbo phase between the hourly and salaried ranks, although they are paid on an hourly basis.

With respect to the hourly positions, the highest is that of Support Manager, which is considered a feeder into the Management Trainee position. Below Support Manager are the Department Managers and Customer Service Managers, who serve as lower-level hourly supervisors. At the bottom rung are the entry level Cashiers, Sales Associates, Stockers, and others.

In general, roughly 65 percent of hourly employees are women, while roughly 33 percent of management employees are women. Drogin Decl. ¶ 19. The approximate percentages of women in specific hourly and salaried management positions are as follows:

Salaried Positions

Store Manager-14%

Co-Manager (only in larger stores)-23%

Assistant Manager-36%

Hourly Positions

Management Trainee-42%

Support Manager-50%

Department Manager-78%

Customer Service Manager-85-90%

Other hourly positions-70% ⁸

8 This category comprises the four largest hourly jobs as follows: (1) Department Head (78%), (2) Sales Associate (68%), (3) Hardlines/Home Area Overnight Associates (57%), and (4) Cashier (93%). Drogin Decl. ¶ 26, Tables 10-11.

See Drogin Decl. ¶ 23, Table 7.

(2) Policies Governing Compensation

(a) Hourly Employees

All hourly employees at every Wal-Mart store are compensated pursuant to the same general pay structure. Each store has a minimum starting wage for each class of hourly jobs that is set by the Wal-Mart Home Office in Arkansas (hereafter " Home Office" ). Beyond that, Store Managers are granted substantial discretion in making salary decisions for hourly employees in their respective stores. Specifically, they are allowed to depart from the minimum start rates, within a two dollar per hour range, without being constrained by objective criteria and with limited oversight. See Def.'s Opp'n at 5-6 & n. 4. Indeed, " [i]n setting pay, a Store Manager makes the call based on his or her needs." Id. at 15; Haworth Decl. ¶ 4 (" This is a company that relies on its Store Managers to set the specific wage rates of each store's employees." ). Given that hourly workers typically earn approximately $18,000 per year (based on 2001 wages), see Drogin Decl. ¶ 19, the two dollar per hour discretionary range is significant. Store Managers also are allowed to increase pay for exceptional performance, again with limited guidance or oversight.

As a partial constraint on this local control Wal-Mart utilizes a system of oversight known as " management by exception." Under this system, District Managers and Specialty Group Regional Managers receive " exception reports" if Store Managers set an employee's pay rate at more than six percent above the minimum rate set by the Home Office, and must approve such rates. Defendant's national pay structure also specifies a 25 cent-per-hour gap between start rates in consecutive pay classes. See Drogin Reply Decl. ¶ 23; Webber Decl., Exhs. 93-94. This limited oversight, however, still leaves individual Store Managers with substantial discretion in setting pay rates for in-store employees. As one Store Manager explained, " There's [a presumptive limit of two dollars above the base], but I can do what I want. I mean, if I start throwing money around, I mean, eventually the phone is going to ring. But the store manager has the flexibility to do what he needs to do to run the building." Shatz Depo. at 66 (Weber Decl. Ex. 46).

(b) Salaried Employees

All types of Wal-Mart stores also compensate their salaried in-store management employees pursuant to common policies and practices. These decisions are made primarily by District Managers (the first level in the management hierarchy above Store Managers) and their superiors, the Regional Managers.

(1) Assistant Managers: For these lowest level salaried managers, Wal-Mart's Exempt Associate Pay Guidelines establish a broad salary range within which District Managers have discretion to set pay rates with little guidance and limited oversight. Assistant Managers receive a base salary and are eligible to receive annual performance and merit increases and a bonus. See Haworth Decl. ¶ 221; Def.'s Opp'n at 35 n. 21. Those working in areas with a high cost of living receive Geographical Assistance Pay (GAP). See Arnold Decl. ¶ 9. The base salary range for Assistant Managers starts at $29,500 and goes to between $42,000 to $47,000 depending on store size. See Haworth Decl. ¶ 222.⁹

9 Assistant Managers at Sam's Clubs have a similar pay structure, except that they are eligible for incentive plans based on store profitability. See Haworth Decl. ¶ ¶ 231-33.

(2) Co-Managers: Similarly, Co-Manager pay rates are set by District Managers who exercise complete discretion, with little guidance and limited oversight, to set salaries within a base salary range from $42,000 to $47,000 (with GAP adjustments). Co-Managers are also eligible to participate in an incentive plan based on store profitability. They do not receive merit or performance increases. Id. at ¶ 225.¹⁰

10 Co-Managers at Sam's Clubs have a similar pay structure. See Haworth Decl. ¶ 236.

(3) Specialty Departments: Specialty Department Managers also have similar, albeit not identical, pay structures. Defendant's expert, Dr. Joan Haworth, states that each Specialty Department has a " different compensation plan," but she only identifies four of the eight departments (Tire & Lube Express [TLE], Photo, Optical, and Pharmacy). Among those, all have base salaries in broad ranges (spanning from $24,000 to over $40,000 at the top, with GAP adjustments), all are eligible for incentive plans, and all are eligible for merit and performance increases. Id. at ¶ ¶ 241-45. Again, these managers' pay rates are set by higher level managers with complete discretion within the relevant range, with limited guidance or oversight.

(4) Store Managers: Store Managers are paid a base salary determined by store size, ranging from $44,000 to $50,000, with GAP adjustments. They also are eligible to receive incentives based on store size and profitability. Id. at ¶ 227. Again, their base salaries are determined by upper level managers with broad discretion within the established range.

As the above reflects, there is a basic compensation structure that applies similarly to all in-store salaried management positions across all types of Wal-Mart stores, in that the computation begins with a base salary within a range set by the corporation (with an allowance for GAP), with adjustments allowed for profit incentives and/or merit increases. While certain differences exist-e.g., some positions allow for profit incentives while others build in merit incentives or both-they do not fundamentally alter the common nature of the positions. Most importantly, all of the in-store salaried positions-like all of the hourly positions-share the common feature that there is a broad range of discretion built into the compensation structure for each position. As discussed further infra, this feature also provides a potential conduit for gender discrimination that is common to all class members. At this point, however, it suffices to say that Defendant's policies governing compensation of hourly and salaried employees do not preclude a finding of commonality; on the contrary, the evidence indicates that there is significant uniformity across stores, and that Defendant's policies all contain a common feature of subjectivity that is relevant to Plaintiffs' claims of class-wide gender discrimination thus supporting the existence of questions common to the class.

(3) Policies Governing Promotions

As with salary decisions, the parties agree that subjectivity is a primary feature of promotion decisions for in-store employees. In the words of Wal-Mart President and CEO Thomas Coughlin, " We push down to the manager of the facility level, [ sic. ] the responsibility to run those stores right." Haworth Decl. ¶ 7. The subjectivity in promotion decisions occurs in two fundamental ways: (a) a largely subjective selection practice hindered by only minimal objective criteria, combined with (b) a failure to post a large proportion of promotional opportunities.

(a) Subjectivity in the Selection Practice

The first promotion category challenged by Plaintiffs-the Support Manager position-is the primary feeder for the Management Training Program. It is undisputed that Wal-Mart allows Store Managers to apply their own subjective criteria when selecting candidates for the Support Manager position. Decisions regarding advancement into the Management Training Program are made by District and Regional Managers. Wal-Mart has minimum corporate guidelines for promotion into these positions, which include requirements that candidates have an " above average" evaluation, have at least one year in their current position, be current on training, not be in a " high shrink" department or store, be on the company's " Rising Star" list, and be willing to relocate. However, since the guidelines set forth only the minimum requirements for advancement, the decisions as to who will actually be selected for the Management Training Program are based largely on subjective criteria, resulting in what is fairly characterized as a " tap on the shoulder" process. See Butler Depo. at 135:4-11 (Webber Decl. Ex. 4) (" Each District Manager kind of has their own way of identifying talent within their district for development." ); Kintzele Depo. at 86:20-23 (Webber Decl. Exh. 13) (explaining that Store Managers nominate candidates for promotion); M. Miller Depo. at 65:2-13 (Webber Decl. Ex. 54); June 27, 2002 E-mail from Jarrells-Porter (Webber Decl. Ex. 100) (statement of Senior Vice President that Wal-Mart " [does] not have a poster, brochure, nothing that I am aware of" to inform hourly workers " how to get promoted into the management training program" ); Bielby Decl. ¶ 22 (citing testimony of Wal-Mart managers that promotions to in-store management positions are made based on largely discretionary criteria). Decisions regarding advancement to the highest in-store positions-Assistant Manager, Co-Manager, and Store Manager-are similarly based on subjective assessments beyond adherence to corporate minimum guidelines. See, e.g., Kintzele Depo. 130:10, 133:9-135:16, 165:22-167:1 (Webber Decl. Ex. 13); Bielby Decl. ¶ 37. The subjective nature of Defendant's promotion practices is further compounded by the fact that the company does not monitor the promotion decisions being made or otherwise systematically review the grounds on which candidates are selected for promotion. Bielby Decl. ¶ ¶ 37-38 & n. 53 (citing testimony of numerous Wal-Mart executives and managers).

(b) Failure to Post Promotional Opportunities

It is undisputed that, until January 2003, Wal-Mart did not post job vacancies for its Assistant Management Training Program, and it posted only a small number of vacancies for the Co-Manager position. See Def.'s Opp'n at 40 (" Wal-Mart does not have applicant flow data for the Trainee position before 2003" ); Drogin Decl. ¶ 46 & Table 20.¹¹ Also, despite a stated policy to post hourly Support Manager positions, roughly 80 percent of these openings were not in fact posted. Drogin Decl. ¶ 44 & Table 19. Because most positions were not posted, class members had no ability to apply for, or otherwise formally express their interest in, openings as they arose. As a result, Managers did not have to consider all interested and qualified candidates, thus further intensifying the subjective nature of the promotion process.¹²

11 From December 1998 through October 2002, Wal-Mart posted only 238 Co-Manager openings. Id. at 120 (chart identifying number of offers for " Co-Mgr-Supercenter" and " Co-Mgr-Div. 1" ).

12 Defendant argues that its wide-scale lack of posting is ameliorated by an aspect of its annual review process whereby employees " list what job they wished to be considered for in the ensuing year." Transcript of September 24, 2003 Oral Argument (" Oral Arg. Tr." ) at 199:22-200:5. However, there is no evidence in the record that this was a systematic process, that the company tracked these expressions of interest, or that the process substituted for a true posting system in any manner.

Wal-Mart did post openings for Store Manager positions during the relevant class period. This was not, however, an " open" application process; rather, candidates were required to obtain permission from their District Manager before being allowed to apply. In selecting who could apply, such managers have been free to rely upon subjective criteria beyond the minimum corporate guidelines. Thus the posting of Store Manager positions did not fully ameliorate the subjective nature of the promotion process for these positions.

(4) Whether Defendant's Compensation and Promotion Policies Support a Finding of Commonality

Having reviewed the extensive evidence submitted by the parties, this Court is satisfied that Wal-Mart's systems for compensating and promoting in-store employees are sufficiently similar across regions and stores to support a finding that the manner in which these systems affect the class raises issues that are common to all class members. Moreover, the fact that Wal-Mart's compensation and promotion policies consistently permit managers to utilize a great deal of subjectivity further supports a finding of commonality.

While some level of subjectivity is inherent in, and in fact a useful part of, personnel decisions, courts have long recognized that the deliberate and routine use of excessive subjectivity is an " employment practice" that is susceptible to being infected by discriminatory animus. SeeWatson v. Fort Worth Bank & Trust, 487 U.S. 977, 990-91, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988) (subjective decision-making is a " practice" subject to challenge under Title VII); Falcon, 457 U.S. at 159 n. 15, 102 S.Ct. 2364; Sengupta v. Morrison-Knudsen Co., 804 F.2d 1072, 1075 (9th Cir.1986) (citation omitted); Casillas v. United States Navy, 735 F.2d 338, 345 (9th Cir.1984) (citation omitted). In the same vein, a general practice of not posting promotional opportunities also supports a finding of commonality. SeeHemmings v. Tidyman's Inc., 285 F.3d 1174, 1187 n. 17 (9th Cir.2002) (" Failure to post vacancies and the use of subjective promotions practices ... may be evidence of discrimination" ).

And while the presence of excessive subjectivity, alone, does not necessarily create a common question of fact, where, as here, such subjectivity is part of a consistent corporate policy and supported by other evidence giving rise to an inference of discrimination, courts have not hesitated to find that commonality is satisfied. See e.g.,Shipes v. Trinity Industries, 987 F.2d 311, 316 (5th Cir.1993) (" Allegations of similar discriminatory employment practices, such as the use of entirely subjective personnel processes that operate to discriminate, satisfy the commonality and typicality requirements of Rule 23(a)." ) (citation omitted); Caridad, 191 F.3d at 292; Beckmann v. CBS, 192 F.R.D. 608, 613 (D.Minn.2000); Shores v. Publix Super Markets Inc., 1996 WL 407850, at *6 (M.D.Fla. Mar.12, 1996) (defendant's " policy of delegating hiring and promotion decisions to managers, who make those decisions on the basis of subjective criteria, is a common course of conduct ... adequate to meet the commonality requirement of Rule 23" ).¹³

13 See also Butler v. Home Depot, Inc. (Butler I), 1996 WL 421436 (N.D.Cal. Jan.25, 1996); Adams v. Pinole Point Steel Co., 65 Fair Empl.Prac.Cas. (BNA) 782, 1994 WL 515347, at *9 (N.D.Cal. May 18, 1994) (" [Plaintiffs] allege that [defendant's discriminatory] practices stem from a subjective hiring criteria and a high level of discretion afforded supervisory employees. A general policy of discrimination that affects individual class members provides a ‘ common nucleus of operative fact,’ and thus sufficient commonality to support class certification." ); Stender v. Lucky Stores, Inc., 803 F.Supp. 259, 335 (N.D.Cal.1992); Wofford v. Safeway Stores, Inc., 78 F.R.D. 460, 479 (N.D.Cal.1978).

The situation here is similar to Morgan v. United Parcel Service, Inc., 169 F.R.D. 349 (E.D.Mo.1996), in which the parties agreed that United Parcel Service had a subjective, decentralized decision-making process for promotions. The defendant claimed that this process should defeat commonality because individualized decisions are made at the local level, whereas plaintiffs argued that the policy of giving local managers unfettered authority was part of a nation-wide corporate policy to which all employees were subjected and which resulted in racial discrimination. Id. at 356. The court agreed with plaintiffs that the subjective decision-making practice was sufficient to satisfy the commonality requirement. Id.

Wal-Mart similarly argues that the fact that pay and promotion decisions are made locally by individual Store Managers necessarily defeats a finding of commonality. To support this argument, it attempts to align itself with those cases in which courts have denied certification based on geographically diverse corporations with localized decision-making. In the principal case relied on by Defendant, Reid v. Lockheed Martin Aeronautics Co., 205 F.R.D. 655 (N.D.Ga.2001), the court held that because the challenged employment practices were highly localized among many facilities, plaintiffs could not establish commonality. While Reid recognized that subjective decision-making may give rise to an inference of discrimination, it determined on the facts of the case that there was no evidence to provide a nexus between the subjective decision-making and discrimination. Id. at 670-72. Here, in contrast, plaintiffs have presented evidence of such a nexus by presenting evidence of gender stereotyping and a corporate culture of uniformity (discussed in greater detail below). Furthermore, Reid emphasized that all of the named plaintiffs in that case worked at the same facility, thereby leaving the court without any basis upon which to conclude that similar practices existed among diverse facilities. Id. Here, in contrast, the named plaintiffs work in different stores and Plaintiffs have submitted over a hundred declarations by designated class members showing common subjective practices across the country. The Court therefore finds Reid distinguishable.¹⁴ The Court also rejects Wal-Mart's suggestion that its practice of giving managers substantial discretion in making pay and promotion decisions should defeat commonality because it would embroil the court or jury in an endless series of individual fact scenarios regarding each pay or promotion decision. As discussed in section II.B.1, Plaintiffs' task at the liability stage of the trial is to prove a pattern and practice of discrimination. To do so, they will need to show sufficient evidence of excessive subjectivity to convince a jury that it is a wide-scale practice, which along with other evidence, is sufficient to create an inference of, and ultimately prove, discrimination. Because the focal point will be the practice of utilizing excessive subjectivity, rather than the facts concerning each individual decision, the Court is satisfied that the subjective nature of Defendant's personnel practices does not defeat commonality in this case.

14 Similarly, in the other case principally relied on by defendant, Rhodes v. Cracker Barrel Old Country Store, Inc., 2002 WL 32058462, at *56-58 (N.D.Ga. Dec.31, 2002), magistrate judge op. approved and adopted, 213 F.R.D. 619 (N.D.Ga.2003), the court denied class certification where plaintiffs failed to show any kind of nexus between subjective decision-making and discrimination. Defendant also attempts to analogize Wal-Mart's promotion system to that of Microsoft, as addressed in Donaldson v. Microsoft, 205 F.R.D. 558 (W.D.Wash.2001). There the court denied class certification in part based on a lack of commonality. Id. at 566. Wal-Mart's practices, however, are significantly more subjective than those described in Donaldson. Microsoft used a well-crafted combination of both objective and subjective measures, with features not present in Wal-Mart, such as bi-annual evaluations, advance mapping of goals and objectives, and an appeal process. Id. at 566.

b. Corporate Culture

Plaintiffs also present evidence that Wal-Mart has developed and continually reinforces a strong, centralized corporate culture. This evidence, taken together with evidence concerning Wal-Mart's company-wide subjective compensation and promotion practices discussed above, provides a further indicator of commonality. Specifically, Plaintiffs present evidence that Wal-Mart corporate culture includes two principal common aspects relevant to this case: (1) a strong emphasis on building and maintaining a uniform culture, and (2) an environment that may include gender stereotyping.

(1) Wal-Mart's Emphasis on Uniform Culture

There is no genuine dispute that Wal-Mart has carefully constructed and actively fosters a strong and distinctive, centrally controlled, corporate culture.¹⁵See, e.g., R. Harper Depo. at 121 (Webber Decl. Ex. 36) (describing Wal-Mart culture as a " very tight, deep culture" and " a very closed culture from a total corporation standpoint" ). This culture, labeled by Defendant as the " Wal-Mart Way," promotes and sustains uniformity of operational and personnel practices. This is important in the Rule 23 commonality analysis because, as Plaintiffs' expert, Dr. William Bielby, states, " [a] strong and widely shared organizational culture promotes uniformity of practices throughout an organization." Bielby Decl. ¶ 18.

15 Sam's Club shares the same culture as the Wal-Mart stores. See Swanson Depo. 140:16-19 (Webber Decl. Ex. 14); Reeves I Depo. 103:1-25 (Webber Decl. Ex. 2).

Every new employee nation-wide goes through the same orientation process and, as part of that process, is trained about the Wal-Mart culture. Thereafter, employees at Wal-Mart stores attend a daily meeting held at shift changes, where managers discuss the company culture and employees do the Wal-Mart cheer. See Muzingo Depo. at 85-86 (Webber Decl. Ex. 32). Employees also receive weekly training on culture topics at mandatory store meetings. ¹⁶ Wal-Mart Store Managers are provided with a detailed schedule of corporate culture lessons and accompanying training materials to present at these weekly meetings. The same Computer-Based Learning Modules are used in all stores, including required training on culture topics. Culture is also an integral part of all management training programs. Home Office managers also attend a meeting each Saturday morning which is dedicated once a month to presenting one of the company's culture lessons. See Muzingo Depo. 135-36 (Webber Decl. Ex. 32). In short, the inculcation of Wal-Mart culture is an ongoing process for all employees regardless of seniority; as one manager puts it, the company has a " continual process of learning culture." See Swanson Depo. at 141 (Webber Decl. Ex. 14).

16 The Culture Topic Index provided to store management employees states: " It is important for our Associates to understand how important culture is to their company. We have provided you with a culture topic for mandatory weekly discussions with all store Associates on each shift during store meetings." See Bielby Decl. ¶ 20, n.30 (and citation therein to document WMHO598668).

Additionally, Wal-Mart has a strong policy of " promoting from within," so that the culture lessons learned by junior-level employees contribute to building a foundation of common understanding and practice among the management team. See Peterson Depo. at 72:2-6 (Webber Decl. Ex. 16). The company also regularly moves store level managers from one retail facility to another, often from one state to another, and between Wal-Mart and Sam's Club divisions. On the average, each Store Manager is transferred to a different store 3.6 times after achieving that title. A majority of these transfers are into different districts and are often into different regions. See Drogin Decl. ¶ 35; Drogin Reply Decl. ¶ 24. This degree of mobility could only be efficient in a company with a high degree of store-to-store uniformity. These practices also help ensure that a uniform Wal-Mart Way culture operates consistently throughout all stores.

Wal-Mart also has a very advanced information technology system which allows managers in the Home Office to monitor the operations in each of its retail stores on a close and constant basis. For example, the Home Office is connected through a real-time computer link (called " Managers' Workbench" ) to every Store Manager throughout the country. The company uses the same Computer-Based Learning Modules in all stores, it broadcasts " Wal-Mart TV" into all stores, and employs a number of other uniform communications tools. Centralization at Wal-Mart extends to the point where the Home Office controls the temperature and music in each store throughout the country.

Finally, plaintiffs present evidence from Dr. Bielby, a sociologist, to buttress their claim that Wal-Mart has a strong corporate culture. After undertaking an extensive review of the depositions of Defendant's managers, Wal-Mart policies, and the professional research and literature in the field, Dr. Bielby gives the following opinion:

In sum, consistent with the organizational research on this topic, Wal-Mart's distinctive corporate culture is sustained by focused efforts of the firm through on-going training and socialization, communication specifically designed to reinforce its distinctive elements, promotion from within and relocating managers from store to store, and shared experiences among employees that build commitment to shared beliefs and values. As a result of these efforts, employees achieve a common understanding of the company's ways of conducting business.

Bielby Decl. ¶ 21.¹⁷

17 Additionally, Dr. Bendick, Plaintiffs' labor economics expert, concludes that Wal-Mart has a high management centralization ratio, in that 15.4 percent of its managers are located at the company headquarters, compared with an average of 8.1 percent for its twenty closest comparators. From this data, Dr. Bendick concludes that Wal-Mart has a strong corporate culture that is both perpetuated by and reflected in employment policies and practices that are common company-wide. Bendick Decl. ¶ ¶ 64-65.

Wal-Mart counters the suggestion that its strong corporate culture creates any commonality among class members in disparate locations by arguing that each of its stores is a virtual " main street" of stores within a store, all run by independent managers. See Def.'s Opp'n at 13. It also contends that each division of stores has its own unique hierarchical structure of reporting and supervision. Id. at 14. While it is true that the terminology and some details vary from division to division, and within store departments, there does not appear to be a significant difference in actual function. Rather, as discussed supra, Plaintiffs have presented evidence that the basic operational structure and staffing patterns of stores are quite uniform, and each individual store is subject to oversight from the company's Home Office, where all regional and higher level managers are based.

The Court recognizes that there is a tension inherent in characterizing a system as having both excessive subjectivity at the local level and centralized control. To clarify, the evidence indicates that in-store pay and promotion decisions are largely subjective and made within a substantial range of discretion by store or district level managers, and that this is a common feature which provides a wide enough conduit for gender bias to potentially seep into the system. These subjective decisions are not, however, made totally in isolation. Rather, the company maintains centralized corporate policies that provide some constraint on the degree of managerial discretion over in-store personnel decisions. The evidence suggests that the company relies also on its strongly imbued culture to guide managers in the exercise of their discretion. It would be premature and inappropriate for the Court to determine the precise degree to which the forms of centralized control at Wal-Mart keep managerial discretion in check. SeeEisen, 417 U.S. at 177, 94 S.Ct. 2140; Orlowski v. Dominick's Finer Foods, Inc., 172 F.R.D. 370, 373-74 (N.D.Ill.1997) (court found commonality in promotion process where degree of centralized decision-making could not be definitively determined without performing full merits inquiry). It is clear, however, that given the evidence regarding strong uniform culture and policies, the degree and impact of this practice is a significant question of fact common to the class as a whole. SeeMorgan, 169 F.R.D. at 356 (commonality requirement can be met where company combines " subjective, decentralized system of decisionmaking" with personnel policies that are " uniform throughout the country and are promulgated by the national corporate office" ).

(2) Gender Stereotyping

Plaintiffs also rely on the expert testimony of Dr. Bielby to support their contention that gender stereotyping is likely to exist at Wal-Mart, and that it persists to the present day. Dr. Bielby notes that there are many reasons why men and women can have different career trajectories, some of which are not generally considered discriminatory, such as differing job-related skill requirements. See Bielby Decl. ¶ 28. Employers create gender barriers, however, when they allow stereotypes to affect personnel decisions. Id. at ¶ ¶ 28-29.¹⁸

18 As another example, both parties recognize that family constraints limit women more than men, and that this is a broad phenomenon that is not entirely attributable to Wal-Mart. However, Dr. Bielby states that in an environment of stereotyping, people tend to act on assumptions that overstate the extent to which they are true. Without a systematic mechanism for determining availability and interest in promotion, stereotypes lead decision-makers to overlook or discount women's desire and qualifications for advancement. See Bielby Decl. ¶ 30.

Dr. Bielby opines that the social science research demonstrates that gender stereotypes are especially likely to influence personnel decisions when they are based on subjective factors, because substantial decision-maker discretion tends to allow people to " seek out and retain stereotyping-confirming information and ignore or minimize information that defies stereotypes." Id. at ¶ 34. With respect to Wal-Mart in particular, Dr. Bielby confirms that managers make decisions with considerable discretion and little oversight. Id. at ¶ ¶ 37-41. He concludes that subjective decisions such as these, as well as discretionary wage decisions, are likely to be biased " unless they are assessed in a systematic and valid manner, with clear criteria and careful attention to the integrity of the decision-making process." Id. at ¶ 39. Dr. Bielby further concludes that these systematic elements are missing from Wal-Mart's decision-making process. Id. Dr. Bielby also notes that the company's practice of requiring relocation across stores for salaried managers, which generally creates a greater burden for women, makes the promotion process " especially vulnerable to gender stereotyping." Id. at ¶ 45.

Lastly, Dr. Bielby reviewed Wal-Mart's diversity and equal opportunity policies, many of which have been implemented in just the last few years. He concluded that they have identifiable weaknesses that limit their effectiveness for identifying and eliminating discriminatory barriers. Id. at ¶ 62. For example, while Wal-Mart regularly compiles statistics on the gender composition of its workforce, it has not undertaken a systematic assessment to identify possible barriers to women's advancement. Id. at ¶ 52. As another example, while Wal-Mart sets diversity goals for female representation in management, the goals appear to be largely ad hoc. Rather than setting the goals based on an understanding of the number of qualified and interested women available for a given position, many of the managers simply make their own subjective determinations or set their goals as incremental improvements over the prior year. Id. at ¶ ¶ 54-55. On the whole, Dr. Bielby recognizes that Wal-Mart has increased its emphasis on diversity issues in the past few years, but because the company has not translated that emphasis into practical and effective measures, there has been little actual impact on gender differentials in pay and promotion. Id. at ¶ ¶ 56-57 (citing former Vice President's view that by failing to tie diversity achievement to manager's incentive pay, the company's diversity efforts remain " lip service" ). Additionally, although Wal-Mart regularly surveys its employees on a number of job-related issues (through its " Grass Roots" personnel program), it has never performed any kind of survey addressing diversity or gender issues. Id. at ¶ 61.

Defendant contests the accusation of gender stereotyping on the facts and by challenging Dr. Bielby's opinions. As a factual matter, Defendant points out a number of ways in which Wal-Mart culture promotes diversity. Wal-Mart has earned national diversity awards and its executives discuss diversity and include it in the company handbooks and trainings. See Porter Decl. ¶ ¶ 5-22. The company also has diversity goals, performance assessments, and penalties for EEO violations. See Def.'s Opp'n at 20.

Defendant also challenges Dr. Bielby's opinions as unfounded and imprecise. It is true that Dr. Bielby's opinions have a built-in degree of conjecture. He does not present a quantifiable analysis; rather, he combines the understanding of the scientific community with evidence of Defendant's policies and practices, and concludes that Wal-Mart is " vulnerable" to gender bias. See Bielby Decl. ¶ 63. Defendant rightly points out that Dr. Bielby cannot definitively state how regularly stereotypes play a meaningful role in employment decisions at Wal-Mart. See Def.'s Opp'n at 20:25-28. However, this is the nature of this particular field of science. See Fed.R.Evid. 702 (allowing " scientific, technical, or other specialized knowledge [that] will assist the trier of fact to understand the evidence" ) (emphasis added).

The appropriate question at this stage of the litigation is not whether Dr. Bielby can make a conclusive determination, but whether it could add probative value to the inference of discrimination that plaintiffs allege. Here, Dr. Bielby presents enough of a basis, both in his review of the scientific literature and on the facts of the case, to provide a foundation for his opinions. SeePrice Waterhouse v. Hopkins, 490 U.S. 228, 250-51, 109 S.Ct. 1775, 104 L.Ed.2d 268 (1989) (considering similar evidence by expert social psychologist). Whether those opinions, if presented to a jury, are ultimately worthy of belief will be for a jury to decide. For present purposes, Dr. Bielby's testimony raises an inference of corporate uniformity and gender stereotyping that is common to all class members.

2. Statistical Evidence of Discrimination

The second category of evidence presented by plaintiffs is statistical evidence of class-wide gender disparities. Use of statistical analysis to raise an inference of class-wide discrimination and satisfy commonality is well accepted. See, e.g.,Caridad, 191 F.3d at 292. Plaintiffs rely primarily on the expert testimony of Dr. Richard Drogin,¹⁹ a statistician who concludes, in essence, that there are statistically significant disparities between men and women at Wal-Mart in terms of compensation and promotions, that these disparities are wide-spread across regions, and that they can be explained only by gender discrimination. Plaintiffs bolster Dr. Drogin's analysis regarding promotion discrimination with the expert testimony of a labor economist, Dr. Marc Bendick, who performed a " benchmarking" study of twenty of Wal-Mart's competitors and concluded that Wal-Mart promotes a lower percentage of women than its retail counterparts.

19 Dr. Drogin is an emeritus professor in the Department of Statistics at California State University, Hayward and is a partner in a statistical consulting firm. See Drogin Decl. ¶ ¶ 1-2.

Defendant contends that Dr. Drogin's analysis is substantially flawed and that in fact there is no statistical pattern of discrimination at Wal-Mart. Defendant relies on the testimony of its own expert economist, Dr. Haworth, who provides both a critique of Dr. Drogin's analysis and a separate analysis that she performed.²⁰ Defendant also challenges Dr. Bendick's expert testimony as methodologically and factually flawed.

20 Dr. Haworth is the Chair and CEO of ERS Group, which specializes in the economic and statistical analysis of employment discrimination cases and other matters. Dr. Haworth is a former faculty member of the Department of Economics at Florida State University. See Haworth Decl. ¶ ¶ 5-6.

Defendant's arguments seek to engage the Court in a merits evaluation of the expert opinions. The Court rejects this approach, and views the statistical evidence and testimony through the proper lens of the standards applicable to a class certification motion. See Legal Standards, supra;Eisen, 417 U.S. at 177, 94 S.Ct. 2140 (admonishing courts not to engage in a preliminary assessment of the merits).²¹ Accordingly, the Court delves into the substance of the expert testimony only to the extent necessary to determine if it is sufficiently probative of an inference of discrimination to create a common question as to the existence of a pattern and practice of gender discrimination at Wal-Mart. SeeAvagliano v. Sumitomo Shoji America, Inc., 103 F.R.D. 562, 574 (S.D.N.Y.1984) (" While [plaintiffs] may have made a greater showing than is required on a motion for certification, certainly no more than an inference need be shown." ) (emphasis in original).

21 See also Caridad, 191 F.3d at 292 (" Though [defendant's] critique of the Class Plaintiffs' evidence may prove fatal at the merits stage, the Class Plaintiffs need not demonstrate at this stage that they will prevail on the merits. Accordingly, this sort of ‘ statistical dueling’ is not relevant to the certification determination." ) (citation omitted); EEOC v. Gen. Tel. Co., 885 F.2d 575, 579-82 (9th Cir.1989), cert. denied, 498 U.S. 950, 111 S.Ct. 370, 112 L.Ed.2d 332 (1990); Butler I, 1996 WL 421436, at *3 (N.D.Cal. Jan.25, 1996) ( " the evidence upon which plaintiffs propose to rely-statistical evidence of widespread discrimination-is common to the class as a whole" ).

a. Statistical Evidence Regarding Compensation

(1) Overview

Both parties' experts performed tests on an extensive array of Wal-Mart payroll and personnel data to determine whether gender disparities in salary exist and whether such disparities can be attributed to discrimination. Plaintiffs present largely uncontested descriptive statistics which show that women working in Wal-Mart stores are paid less than men in every region, that pay disparities exist in most job categories, that the salary gap widens over time even for men and women hired into the same jobs at the same time, that women take longer to enter into management positions, and that the higher one looks in the organization the lower the percentage of women. See Drogin Decl. ¶ ¶ 21, 23; Haworth Depo. at 67:14-70:16, 76 (Suppl. Webber Decl. Ex. 129); see also section I.B.1.a.

Evidence that certain disparities exist, however, does not, by itself, explain why they exist. Accordingly, Plaintiffs also present inferential statistical evidence in the form of multiple regression analysis to demonstrate that the above disparities can be explained only by gender discrimination and not by chance or other neutral variables.²² Defendant follows suit with a critique of Plaintiffs' regression analysis and with a separate regression analysis of its own. Thus, while they disagree as to application, the parties agree in principle that regression analysis is an appropriate and scientifically valid statistical technique which has gained general acceptance, indeed a general preference, by the courts. SeeHemmings, 285 F.3d at 1183-84 & n. 9 (" regression analysis is a common statistical tool" ); Lavin-McEleney v. Marist College, 239 F.3d 476, 482 (2d Cir.2001) (" It is undisputed that multiple regression analysis ... is a scientifically valid statistical technique for identifying discrimination." ).

22 Multiple regression analysis, in general terms, provides estimates of the effect of several independent variables on a single dependent variable. The purpose is to estimate the extent to which a particular independent variable (in this case, gender) has influenced the dependent variables of compensation and promotion. So long as the analysis includes enough relevant non-discriminatory independent variables (e.g.-education, experience, performance, etc.), the results will indicate whether any salary disparities are attributable to gender (thereby raising an inference of discrimination) or whether the disparities are attributable to other factors (and thereby refuting such an inference). See Hemmings, 285 F.3d at 1183-84 & n. 9.

With respect to compensation, Dr. Drogin performed separate regression analyses for hourly and salaried employees for each Wal-Mart region. In general, the regressions showed statistically significant ²³ gender-based disparities for all in-store job classifications in all 41 Wal-Mart regions. See Drogin Decl. ¶ 77. Specifically, total earnings paid to women ranged between 5 and 15 percent less than total earnings paid to similarly situated men in each year of the class period. Id. at ¶ 77. Dr. Drogin's regression analyses included seniority, turnover, performance, and other factors, none of which accounted for the disparities, thereby leaving him to conclude that gender is the only explanatory factor. In short, all of Dr. Drogin's regressions show that gender is a statistically significant variable in accounting for the salary differentials between female class members and male employees at Wal-Mart stores. Id. at ¶ 76.

23 Statistical significance is measured by standard deviations. The standard deviation is a number that quantifies the probability that chance is responsible for any difference between an expected outcome and the observed outcome in a sample containing two groups. The greater the number of standard deviations, the less likely it is that chance is the cause of any difference between the expected and observed results. " [A]s a general rule for ... large samples, if the difference between the expected value and the observed number is greater than two or three standard deviations, then the hypothesis that the [disparity] was random would be suspect to a social scientist." Castaneda v. Partida, 430 U.S. 482, 496 n. 17, 97 S.Ct. 1272, 51 L.Ed.2d 498 (1977).

Defendant's statistical expert, Dr. Haworth, rejected Dr. Drogin's focus at the regional level and conducted her regression analyses at the store sub-unit level. Consequently, while Dr. Drogin ran a separate regression analysis for each of the 41 regions, Dr. Haworth ran a separate regression analysis for (a) each of the Specialty Departments within each store, (b) each grocery division within each store, and (c) the remainder of each store. See Haworth Decl. ¶ 191.²⁴ Due to the focus on sub-units, Dr. Haworth performed approximately 7,500 separate regression analyses. Her results show a lack of broad-based gender differential in pay for hourly employees, although they show some gender disparities in limited instances. Dr. Haworth did not perform regression analyses for most salaried positions because, in her view, " total compensation for these jobs is at least in part a function of factors that are not available in the data provided to either side," and therefore " the regression models may not provide a meaningful explanation of variations in total compensation." Haworth Decl. ¶ ¶ 253-55.

24 As noted earlier, most stores have roughly 40 to 53 departments, such as automotive, jewelry, apparel, etc. Webber Decl. Ex. 70 at 8-9. Some of these departments are highly sex segregated (e.g. cashiers are 89.5% female, hardware sales associates are over 75% male), while others are roughly even in terms of gender (e.g. electronics is 47.2% female). Bielby Decl. ¶ 22.

(2) Defendant's Challenges to Plaintiffs' Statistical Analysis

Defendant does not dispute the accuracy of Dr. Drogin's calculations, but instead argues that his analysis fails to support an inference of class-wide discrimination because his methodology is fundamentally flawed in two respects: (1) Dr. Drogin improperly aggregated the data at the regional level, and (2) Dr. Drogin failed to account for certain variables. Each point is addressed in turn.

(a) Aggregation at the Regional Level

Defendant argues that because Dr. Drogin's regression analysis was conducted at the regional, rather than the store (or store department) level, his results are too generalized and fail to account for the significant differences in compensation practices that exist among the individual stores. In statistical language, Dr. Drogin's regional analyses suffer from " aggregation bias," and the only way to cure this bias, according to Defendant, would be to " disaggregate" the data and test it at the store-by-store level.²⁵

25 Defendant argues in its briefing that the statistical analysis should be performed at the store level, but Dr. Haworth went a level lower by analyzing store sub-units, as described above. This could be seen as a methodological inconsistency that would undermine Defendant's argument. However, Defendant's rationale for focusing on the store level is that most compensation decisions occur at the store level; Dr. Haworth's sub-unit analysis is logically consistent with that rationale because it is based on an understanding (disputed by Plaintiffs) that Specialty Department and Grocery unit employees are paid through a different structure than other store employees. The Court therefore concludes that the inconsistency between Defendant's focus on " stores" in its briefs versus Dr. Haworth's focus on store sub-units to be mostly semantic.

The proper test of whether workforce statistics should be viewed at the macro (regional) or micro (store or sub-store) level depends largely on the similarity of the employment practices, and the interchange of employees, at the various facilities. SeeKirkland v. New York State Dept. of Correctional Servs., 520 F.2d 420, 425 (2d Cir.1975) (focus of analysis depends on nature of defendant's employment practices); Lindemann & Grossman, Employment Discrimination, at 1598, 1723. Here, Dr. Drogin contends that it is proper to conduct the analysis on a regional level because the subjective decision-making in compensation and promotions takes place within parameters and guidelines that are highly uniform, and within a strong corporate culture. Plaintiffs also have shown, as discussed above, that the primary salary decision-makers, Store Managers, experience frequent relocation among the various stores.²⁶

26 Dr. Haworth attempts to frame this issue differently, by arguing that Plaintiffs assume that Wal-Mart is a company with " one decision-maker" who sets pay rates for all employees across the country. Haworth Decl. ¶ ¶ 4, 23. This is an argumentative distortion of Dr. Drogin's analysis. It is one thing to conclude that stores operate pursuant to common policies based on evidence that a number of core practices have a strong degree of similarity; it is another to infer the existence of a single decision-maker from the existence of common policies. Dr. Drogin did the former and not the latter.

Dr. Haworth challenges Dr. Drogin's regional focus largely based on a survey taken of Store Managers regarding factors individual managers utilize in setting compensation. Specifically, she argues that aggregating data on a regional level " is not consistent with the variety in the responses of a randomly selected group of Store Managers as to what they believe to be important factors in making pay decisions." Haworth Decl. ¶ 25. As discussed in this Court's Order re Motions to Strike, however, this survey is stricken from the record. Accordingly, Defendant's reliance on this survey to challenge Dr. Drogin's statistical methods is misplaced.

Defendant also attacks Dr. Drogin's reliance on regional level statistics because he failed to perform a " Chow" test. This test (named after the statistician who created it) can be used to analyze whether two or more sets of data may be aggregated as a single sample in a statistical model. Dr. Haworth contends that because Store Managers possess substantial discretion and latitude in setting pay rates, stores act as distinct units; therefore, Dr. Drogin should be required to successfully apply the Chow test before aggregating the data on a regional level.

Defendant does not, however, point to any authority that would require use of a Chow test in this case. Defendant relies on Coates v. Johnson & Johnson, 756 F.2d 524, 542 (7th Cir.1985), a decision on the merits, in which plaintiffs' expert aggregated (or " pooled" ) data over a number of years, rather than performing separate analyses on a yearly basis. The district court found that the weight of the competing expert testimony favored the defendant's position that the data was more reliable on a yearly basis. The Seventh Circuit emphasized that this was a close decision based on the particular facts of the case, id. at 542, and that " [p]ooling data is sometimes not only appropriate but necessary." Id. at 541. Defendant of course will be free to rely on the Chow test as a means of attempting to discredit Dr. Drogin at trial. Nothing in Coates, however, mandates a Chow test on the facts of this case.²⁷ Nor does the present factual record compel the conclusion that a Chow test must be performed. As noted, Dr. Haworth emphasizes that each store should be treated as a distinct unit, given that individual Store Managers possess substantial latitude in setting pay rates. Plaintiffs also present evidence, however, that the stores, while autonomous in certain respects, do not operate in isolation; rather, managerial discretion and subjectivity are exercised in the context of a strong corporate culture and overarching policies. In such circumstances, there is less justification for treating each store as a distinct unit for purposes of statistical analysis. As such, the Court is not persuaded that Dr. Drogin's aggregated statistical analysis should be rejected because he did not choose to utilize the Chow test.

27 See also Rossini v. Ogilvy & Mather, 615 F.Supp. 1520, 1522-23 (S.D.N.Y.1985), vacated and remanded on other grounds, 798 F.2d 590 (2d Cir.1986) (in decision on merits regarding salary discrimination, the court found that the Chow test showed statistically significant differences in regressions run on data before and after limitations date; nonetheless, the court held that " categorical rejection" of a single regression combining all data before and after the limitations date was " not warranted" ); Vuyanich v. Republic Nat'l. Bank, 505 F.Supp. 224, 299, 314 (N.D.Tex.1980), vacated on other grounds, 723 F.2d 1195 (5th Cir.1984) (" Because the controversy here [regarding the Chow test] appears to center on an issue on the frontier of econometrics, and there seems, at least to a court unschooled in the intricacies of econometrics, to be genuine conflict between the experts as to the proper approach, we do not decide the issue." ).

Moreover, the Ninth Circuit recently held that plaintiffs may rely on aggregated data to present claims of employment discrimination in Paige v. California, 291 F.3d 1141 (9th Cir.2002). In that case, the plaintiffs' statistician aggregated data from various promotion examinations rather than viewing each examination on its own. Id. at 1148. All non-white plaintiffs were also aggregated as a single group rather than viewing each race separately. Id. The court began by stating that " it is a generally accepted principle that aggregated statistical data may be used where it is more probative than subdivided data." Id. (citation omitted). After noting that plaintiffs' theory was that defendant's employment practices had an identical discriminatory effect upon all members of the class, the court stated that " [r]ight or wrong, [plaintiffs] are entitled to attempt to prove their case" and " plaintiffs may aggregate data ... for comparison purposes." Id. at 1149.

Clearly, the specific forms of aggregation in Paige are quite different than the geographical aggregation at issue here. The general issue, however, is the same: whether a particular level of generalization should be declared the only legally relevant focal point for statistical analysis, or whether statistical results may be presented at various levels of generality or specificity with the ultimate determination being made after all the evidence is evaluated by the finder of fact. The Ninth Circuit answered this question by holding that statistical analyses run on aggregated data should not be excluded, and may in fact be highly relevant.

Finally, the Court notes that Abram v. United Parcel Service of America, Inc., 200 F.R.D. 424 (E.D.Wis.2001), relied upon by Defendant, does not undermine the use of regional statistics in this case. In Abram, the court denied class certification in part because plaintiffs' statistics failed to support commonality. Id. at 431. The parties agreed that there was a statistically significant pay gap between African-Americans and Caucasians when the data was considered in the aggregate at the nation-wide level. Id. However, at the district-by-district level, plaintiffs' expert conceded that there was no statistically significant difference. Id. Furthermore, the court found that plaintiffs' expert failed to control for non-racial variables, such as education and employment history, which, when properly accounted for even at the nation-wide level, led to the conclusion that there was no statistically significant difference among the races. Id. at 431 n. 3. Thus, the Abram decision is essentially a rejection of plaintiffs' statistical evidence on the merits of that particular case. To the extent that a general proposition can be extracted from Abram, it would be that aggregated data at the nation-wide level is highly suspect, but that aggregation at a district level could be appropriate. Notably, United Parcel Service divides the country into sixty-five districts, some of which extend to an entire state or bloc of states. Id. at 426. Wal-Mart uses a similar organizational scheme with forty-one regions nation-wide. Since Dr. Drogin analyzed the data at a level similar to the United Parcel Service district range, and controlled for a variety of relevant non-gender factors, his regional analyses is actually consistent with the Abram decision. ²⁸

28 Defendant also relies on Lott v. Westinghouse Savannah River Co., 200 F.R.D. 539 (D.S.C.2000), in which the court denied class certification. In Lott, plaintiffs presented mostly " simple grade-school level analyses" of the data through lay witnesses. Id. at 546. Plaintiffs also presented the testimony of an expert witness, which the court does not specifically describe, but which appears to have been a break-down of racial disparities in employment in each EEO job category. Id. at 547. While understandably concluding that " [p]laintiffs' statistical reports fail to demonstrate satisfaction of the commonality criteria because of critical deficiencies in their composition," the court also criticized plaintiffs for aggregating the data across divisions and classes of employees, reasoning that " [s]uch aggregated statistics ‘ fail to ‘ reveal whether there is a pattern of disparities across the various decisionmaking units or a reliable sample of them.’ ' " Id. at 560-61 ( quoting Stastny v. Southern Bell Tel. & Tel. Co., 628 F.2d 267, 278 (4th Cir.1980)). The court provides no further explanation for its decision regarding the statistics. Its single conclusory statement can not be extended to justify Defendant's proposed rejection of Dr. Drogin's aggregated analysis, which as discussed above, is based on a substantial factual record.

In sum, the Court is not persuaded that Dr. Drogin's regional analysis should be rejected for purposes of this motion because it is lacking in probative value. The ultimate question of whether subjective decision-making and a uniform culture contribute to a nation-wide pattern of gender discrimination will, of course, be for a jury to decide. At this stage, however, these factors are apparent enough to support Dr. Drogin's regional approach as at least a reasonable means of conducting a statistical analysis. See, e.g.,Barefield v. Chevron U.S.A., Inc., 44 FEP 1885, 1987 WL 65054 (N.D.Cal.1987) (rejecting challenge to plaintiffs' statistics at class certification stage when objection went to the merits of the case).²⁹

29 See also Caridad, 191 F.3d at 292-93 (in employment class action where defendant had subjective decision-making practices, plaintiffs' expert conducted multiple regression analyses on both a company-wide and a department-by-department basis; the court concluded that the statistical disparities were " not insignificant," that the " statistical dueling" in which defendant sought to engage the Court was " not relevant to the certification determination," and that while " [m]ore detailed statistics might by required to sustain the Plaintiffs' burden of persuasion [at trial]," they " satisfie[d] the Class Plaintiffs' burden of demonstrating commonality for purposes of class certification." ).

(b) Selection of Variables

Defendant also contends that Dr. Drogin's statistical analysis should be rejected because it fails to account for a variety of factors, or control elements, that could be responsible for the disparities in question-referred to as " omitted variable bias." The question of which factors a statistician should account for in any given case is not governed by any bright-line rules; rather, it depends upon the theory and facts of each particular situation.

Here, Dr. Drogin controlled for a number of major variables, including: gender, length of time with the company, number of weeks worked during the year, whether the employee was hiring or terminated during the year, full-time or part-time, which store the employee worked in, whether the employee was ever hired into a management position, job position, and job review ratings. See Drogin Decl. ¶ ¶ 67-71. Defendant argues, however, that these variables do not fully reflect its compensation decision-making structure, thereby leaving open the possibility that one or more missing variables could explain the gender disparities in question. In particular, Dr. Haworth identifies eleven variables that Dr. Drogin allegedly omitted: hours worked, seniority, leave of absence, full-time/part-time status at hire, recent promotion or demotion, prior grocery experience, pay group, night shift, department, store size, and store profitability. See Haworth Decl. ¶ ¶ 280-92.

Dr. Haworth's methodology for determining which variables are relevant was formed in large part by selecting certain decision-making factors identified by Store Managers in a survey. See Haworth Decl. ¶ 20. As previously noted, however, Dr. Haworth can not properly rely on this survey. Even assuming that Dr. Haworth could show that Dr. Drogin omitted one or more relevant factors, this would not justify disregarding Dr. Drogin's findings for purposes of this motion.

First, as the Ninth Circuit recently emphasized in Hemmings, 285 F.3d at 1188, " the law does not require the near-impossible standard of eliminating all possible nondiscriminatory factors." Id. (emphasis in original). Rather, " it is clear that a regression analysis that includes less than ‘ all measurable variables' may serve to prove a plaintiff's case." Id., quotingBazemore v. Friday, 478 U.S. 385, 400, 106 S.Ct. 3000, 92 L.Ed.2d 315 (1986). Given that Dr. Drogin clearly considered a broad array of significant factors, whether he omitted some other, relevant factors-and if so, whether that failure is of sufficient magnitude to discredit his analysis-are questions that should await resolution at trial Hemmings, 285 F.3d at 1183-84, 1187 (multiple regression analysis that controlled for experience and seniority, but did not account for employee qualifications, education, motivations, preferences, and individual fields was relevant and admissible); SeeAdams v. Ameritech Serv., Inc., 231 F.3d 414, 427-28 (7th Cir.2000) (defendant's objections that analysis did not control for outside factors went to the weight of the evidence, not its admissibility); Bouman v. Block, 940 F.2d 1211, 1225 (9th Cir.1991) ( " Whether the statistics are undermined or rebutted in a specific case would normally be a question for the trier of fact." ).³⁰ Second, Defendant fails to show, as it must, that " ‘ curing the alleged flaws would also cure the statistical disparity.’ " Hemmings, 285 F.3d at 1188 (citation omitted). ³¹

30 See also Mangold v. Cal. Pub. Utils. Comm'n, 67 F.3d 1470, 1476 (9th Cir.1995); EEOC v. Gen. Tel. Co., 885 F.2d 575, 582 (9th Cir.1989) (concluding that plaintiff's statistical analysis was probative of discrimination despite its failure to include employment interests); Bruno v. W.B. Saunders Co., 882 F.2d 760, 767 (3d Cir.1989) (affirming district court's admission of statistical studies, despite the failure of the studies to take " account of the minimum qualifications of the jobs into which promotion or transfer occurred" ).

31 Defendant's citation to Sheehan v. Purolator, Inc., 839 F.2d 99 (2d Cir.1988) does not further its cause. In Sheehan (1) the plaintiffs' expert failed to account for critical factors, (2) the court found that these factors could in fact explain the gender disparities at issue, and (3) plaintiffs failed to adduce other significant evidence to support their claims. Id. at 103. The court also recognized that even with the serious flaws infecting plaintiffs' regression analysis, it would have been improper to exclude it from evidence. Id.; see also EEOC v. IBM, Inc., 583 F.Supp. 875, 898 (D.Md.1984) (the court conducted a merits inquiry at trial, finding that plaintiffs' expert failed to account for critical non-discriminatory variables, including seniority, job level, and education).

In short, Defendant has shown only that its choice of variables presents an alternative approach; it has not discredited or nullified Dr. Drogin's results, at least on the record thus far presented. On the contrary, Dr. Drogin's selection of variables appears to go well above the minimal threshold established by the courts, and thus his analysis is sufficient to raise an inference of discrimination for purposes of this motion.³²

32 This is not to say that the omitted variables would have no impact on the regression results. For example, Defendant properly notes that prior grocery experience is a distinguishing factor which could warrant Store Managers to use their discretion to award higher pay, and that more men have this experience than women (57.4% men versus 46.6% women). See Def.'s Opp'n at 26 n. 15; Oral Arg. Tr. at 190:7-25. On the other hand, Wal-Mart's compensation structures make no distinction between grocery and non-grocery. The Court simply is not in a position at this stage of the litigation to definitively resolve this issue. For present purposes, as discussed above, it is enough that Dr. Drogin's analyses account for a sufficient number of the principal variables to support Plaintiffs' position.

b. Statistical Evidence Regarding Promotions

(1) Internal Data Analysis

(a) Overview

Dr. Drogin conducted a statistical analysis of Wal-Mart's internal promotion data during the class period, and concluded that, on a company-wide basis, there is a statistically significant shortfall of women being promoted into each of the in-store management classifications over the entire class period. He further found that the pattern of under-promotion into each of the in-store managerial classifications is " consistent in nearly every geographic region at Wal-Mart." Drogin Decl. ¶ 63. His data essentially supports this conclusion in statistically significant terms for Support Managers and Management Trainees. Id. & Table 26.³³

33 The Court observes, however, that Dr. Drogin's data shows that while female Co-Managers received fewer promotions than similarly situated men in 37 of 40 regions across the country, the disparity was of a statistically significant value in only 22 regions. Furthermore, while the data shows that for Store Manager positions women received fewer promotions than men in 34 of 40 regions, the disparity was of statistically significant value in only 13 regions. Id.

Dr. Drogin also studied the time that it has taken employees to reach in-store management positions. He found, and Defendant does not dispute, that it consistently takes women longer than comparable men to reach the higher management levels. On average, it took women 4.38 years from date of hire to be promoted to assistant manager, while men took 2.86 years. It took 10.12 years for women to reach Store Manager, compared with 8.64 years for men. See Drogin Decl. ¶ 29.

Dr. Haworth rejects Dr. Drogin's approach and concludes, using her own methodology, that women have been selected for all in-store management positions at a rate substantially equivalent to the rate at which they apply (or the rate at which they can be deemed to have applied). In fact, she concludes that women are promoted at a higher rate than predicted by their representation among those who applied. See Haworth Decl. ¶ 139. To the extent that some gender differentials exist, she contends that they are attributable to differing job aspirations and interests between men and women which exist in the general labor force and cannot be blamed on Wal-Mart.

(b) Defendant's Challenge to Dr. Drogin's Choice of Applicant Pools

Defendant's principal challenge to Dr. Drogin's analysis is that it fails to rely solely on actual applicant flow data. This defect, Defendant argues, renders Dr. Drogin's methodology fatally flawed. As explained below, however, this Court concludes that Dr. Drogin's analysis is sufficiently probative to support a finding of commonality.

Both parties' experts attempt to compare the number of women hired into different positions with the number of women applicants seeking advancement to those positions. In circumstances where (a) the employer notifies employees of openings through a neutral job posting or other system, (b) allows all interested employees to apply, and (c) maintains complete records, the actual applicant flow data (i.e. the number of employees actually applying for the job) exists and is recognized as the best basis for making this comparison. SeePaige, 291 F.3d at 1147. Where, however, openings are not posted, or where the system is poorly documented or biased, it is impossible to determine with certainty how many employees would have applied for openings; accordingly, statisticians look for the next best source available to estimate what the actual applicant flow would have been. Seeid. at 1145.

The parties agree that the actual applicant flow data for promotions at Wal-Mart during the class period is limited to the following: (1) from 1998 to 2002 the company conducted numerous postings for the lowest level hourly management position (which is not part of the class), and a very limited number of postings for Support Manager,³⁴ (2) from 1998 to 2002, most Store Manager positions were posted,³⁵ and (3) in January and April-May 2003, Wal-Mart conducted a limited posting for the Management Trainee position. ³⁶ As such, the data contains the following substantial gaps:

34 From the inception of the class period on December 27, 1998 through September 30, 2002, Wal-Mart posted roughly 6,800 job openings for Support Manager. Approximately 41,000 applications were submitted for these openings, and 46.1% of the applicants were women. See Haworth Decl. ¶ ¶ 12, 84.

35 Between December 27, 1998 and October 17, 2002, Wal-Mart Management Career Selection (MCS) system documented over 46,000 applications to roughly 4,400 salaried management positions. See Haworth Decl. ¶ 119. Most of these documented moves were to Store Manager positions. Id. at ¶ 128 & Table entitled " Comparison of Moves, Management Career Selection Data versus Global PeopleSoft Data from December 27, 1998 through March 8, 2002." Most of the moves to other salaried management positions were not documented by the MCS system. Id.

36 Wal-Mart conducted a seven-day company-wide posting in January 2003 for admission to the Management Trainee position. Roughly 30,000 candidates applied, of which 44% were female. See Haworth Decl. ¶ ¶ 16, 72, 96. Approximately 1,400 positions were filled through this posting. See Drogin Reply Decl. ¶ 11. Wal-Mart posted job vacancies for the Management Trainee position again in April-May 2003, although this posting was limited to Wal-Mart Discount Stores and SuperCenters. This posting generated approximately 11,500 applicants, of which 41% were female. See Haworth Decl. ¶ 72 & n. 33, 103.

(1) There is no documented job applicant data for 80 percent of the actual promotions to the position of Support Manager. See Drogin Decl. ¶ 44; Drogin Reply Decl. ¶ 6.

(2) Prior to January 2003, there is no documented job applicant data for entry level salaried management positions.

(3) The amount of Management Career Selection (MCS) information on Assistant Manager vacancies (less than 1% of the promotions are documented) and Co-Manager vacancies (less than 3% are documented) is negligible. See Haworth Decl. ¶ 128. Furthermore, the MCS system contains no data for any management positions in Sam's Club stores. See Drogin Decl. ¶ 45.

Predictably, the experts draw opposite conclusions from the limited amount of actual applicant-flow data available. Dr. Haworth argues that the data, while limited, is nonetheless sufficient to justify an extrapolation for all job openings during the entire class period. In contrast, Dr. Drogin concludes that Wal-Mart's job posting systems " provide no useful data for analyzing promotions into the upper level hourly supervisory, and store management positions." Drogin Decl. ¶ 43. Therefore, he looked to what he considered the next best source to estimate actual applicant flow by tabulating " the incumbents in historical feeder jobs for [each] promotion." Id. at ¶ 48.

While Defendant vigorously attacks Dr. Drogin's approach, it is well recognized that where actual applicant flow data is inadequate or unavailable, other measures of applicant flow-including but not limited to " feeder pools" -are deemed acceptable so long as they are used in a reliable manner. See R. Paetzold and S. Willborn, The Statistics of Discrimination Using Statistical Evidence in Discrimination Cases, §§ 4.02-4.04 (Oct.2002). In Hemmings v. Tidyman's, Inc., for example, the Ninth Circuit recently affirmed using feeder pools where the defendant fills its higher management positions through internal promotions. 285 F.3d at 1185-86 (finding that " the potential applicant pool-and thus the appropriate comparison pool-for promotions to upper and middle management jobs at Tidyman's is comprised of the current employees in lower management positions" ).³⁷

37 Similarly, in Stender, 803 F.Supp. at 333-34, the court accepted the same type of feeder pool methodology used by plaintiffs here. In fact, the Stender plaintiffs employed Dr. Drogin, and the defendant employed Dr. Haworth. The court found that the defendant's bid data was incomplete, and that it was therefore appropriate for Dr. Drogin to take into account the historical pattern of movement between jobs at Lucky Stores by constructing feeder pools for each target job. Id. at 295-96. Specifically, in Stender, Dr. Drogin compared the movement of male and female employees out of low-level clerk positions and into various mid-level apprentice positions, and found a shortfall of women for each set of promotions. Id. at 296. Dr. Drogin used the same control variables as in his Wal-Mart analysis, i.e. year of promotion, feeder job, and store (although the geographic variable in Wal-Mart is sometimes the district rather than store). Id. at 296 n. 8. Dr. Haworth revised the feeder pool analysis by using a reference adjustment factor, based on defendant's analysis of gender differentials in interest levels, and found no evidence of discrimination. Id. at 313. Judge Patel credited Dr. Drogin's analysis and rejected Dr. Haworth's. Id.

Notably, Wal-Mart fails to cite any authority that would require Dr. Drogin to rely solely on extrapolations from grossly incomplete actual applicant flow data. Defendant cites to General Electric Co. v. Joiner, 522 U.S. 136, 146, 118 S.Ct. 512, 139 L.Ed.2d 508 (1997) for the proposition that " [c]ourts permit extrapolation for the liability period from recent applicant flow data, absent proof that the level of interest had materially changed." Def's Opp'n at 40 & n. 25. General Electric, however, merely states the general proposition that experts commonly extrapolate from data, and it made this unremarkable statement in the context of a personal injury case in which the expert extrapolated cancer causality from studies on mice to causation in humans. 522 U.S. at 146, 118 S.Ct. 512.

Defendant's reliance on Stout v. Potter, 276 F.3d 1118, 1123 (9th Cir.2002) and Paige, 291 F.3d at 1147, is equally misplaced. In Stout, an employment case, the court was presented with actual applicant flow data for the entire time period at issue, and therefore never had occasion to address the situation in which such data is incomplete. In Paige, the court had access to both internal and external applicant flow data, and it stated a preference, given the particular facts of the case, for the internal data. Significantly, Paige recognized that the preference for actual applicant flow holds true " only if there is not ‘ a characteristic of the challenged selection device that makes use of the actual pool of applicants or eligible employees inappropriate.’ " 291 F.3d at 1145 ( quotingMoore v. Hughes Helicopters, Inc., 708 F.2d 475, 482 (9th Cir.1983)). A widespread failure to use the " selection device" of job posting could certainly make the preference for applicant flow " inappropriate." ³⁸ At bottom, Defendant fails to grapple with the basic fact that any approach-whether it is extrapolation from limited data (as Defendant prefers) or use of comparative data (as Plaintiffs prefer)-will have a degree of uncertainty given the admitted lack of actual data for most of the positions at issue due to the company's wide-scale lack of job posting. Defendant's assertion that its approach is necessarily superior does not withstand scrutiny. Rather, Defendant's arguments, which go to the weight of the evidence, merely highlight the presence of a significant issue affecting all class members which supports, rather than defeats, granting class certification. Moreover, while a jury may or may not ultimately find Dr. Drogin's approach more probative than Dr. Haworth's, it is at least reasonable. As such, Dr. Drogin's method is sufficient to create an inference of discrimination for purposes of this motion.

38 Defendant also cites Capacchione v. Charlotte-Mecklenburg Schools, 57 F.Supp.2d 228, 247 n. 16 (W.D.N.C.1999), aff'd in part, rev'd in part, vacated in part, by Belk v. Charlotte-Mecklenburg Bd. of Educ., 269 F.3d 305 (4th Cir.2001), a school desegregation case in which the court was presented with over twenty years of complete enrollment data. The court approved an extrapolation from that data to estimate the enrollment during a previous eight year period. Id. The use of extrapolation in that case was unchallenged and the court merely found it to be a reasonable method that was the best means available in the situation. Id.

(2) External Benchmarking Data Analysis

Plaintiffs also buttress their position that there is a common question regarding Defendant's alleged discriminatory policies toward the class with the benchmarking analysis of a labor economist, Dr. Marc Bendick. In contrast to Dr. Drogin and Dr. Haworth, who analyzed internal Wal-Mart data to determine whether promotion disparities exist and, if so, whether they are attributable to gender discrimination, Dr. Bendick looked outside of Wal-Mart to other large retailers for his data source. Specifically, Dr. Bendick compared, or " benchmarked," Wal-Mart against twenty other large general merchandise retailers by comparing workforce data provided by the companies to the Equal Employment Opportunity Commission (" EEOC" ). The practice of benchmarking one company's performance against its competitors is a standard management technique used throughout the private sector and by Wal-Mart itself. See Bendick Decl. ¶ 14 (citing Peterson Depo.).

The type of benchmarking applied here addresses the issue of labor supply, i.e. the availability of female employees who are qualified and interested in management positions. The purpose of this analysis is to determine the extent to which women in the relevant market sought promotion, so that an inference could be made that roughly the same percentage of women would have applied at Wal-Mart if given the opportunity. As Dr. Bendick explains: " The logic in benchmarking is that, if retail chains comparable to Wal-Mart are successfully employing women at some rate, then women are presumably available, interested, and qualified to hold comparable positions at Wal-Mart at a similar rate." Id. at ¶ 16.

Dr. Bendick used data from the EEOC to compare Wal-Mart against twenty large nation-wide retailers.³⁹ He found that a shortfall in female managers was present in 79.5 percent of all Wal-Mart stores. Id. at ¶ 43. He concluded that " [w]ith a shortfall present in four out of five stores, it is virtually impossible for the pattern to be geographically localized." Id. He further concluded that while the in-store managerial workforce at the comparison stores was 56.5 percent female, it was only 34.5 percent female at Wal-Mart. Dr. Bendick determined that this differential is highly statistically significant (47 standard deviations).

39 The data is compiled by the EEOC from a mandatory survey of the nation's largest private firms, and the results are contained in EEO-1 reports, which essentially form a census of the workers at those firms. Dr. Bendick selected the comparator firms by looking for similar large retail companies in the EEO-1 reports. He used two criteria: (1) the firm was in a retail industry focusing on building material and garden supplies, general merchandise, food store, automotive, apparel, furniture, and miscellaneous retail, excluding eating and drinking places; (2) the firm had at least 100 separately reporting establishments with at least 100 employees each. Twenty firms met his criteria. See Bendick Decl. ¶ 21. The identities of these companies in the EEO-1 reports are anonymous, but based on the limited number of companies of such large size in the country, Dr. Bendick was able to surmise that the list of twenty includes Target, Costco, K-Mart, and others. Id. at ¶ 23.

Defendant argues that Dr. Bendick was biased in conducting his benchmarking analysis by " cherry-picking" the comparator companies. However, his criteria for selecting the comparators appears to be reasonable, and there is no evidence that he left out companies that fit his criteria.⁴⁰ Defendant also points to its own alternative benchmarking analysis, which adjusted the criteria to include small and medium sized companies. Defendant will be free to present this alternative model to the jury. It fails to demonstrate, however, that Dr. Bendick's approach is so flawed as to lack probative value. Again, the question of which approach should ultimately prevail is for the jury, and Defendant's attempt to litigate this issue now is premature.

40 Defendant also argues that Dr. Bendick's analysis is methodologically flawed. It contends that Dr. Bendick's sample size is too small, in that he excludes all small employers. However, the analysis covers twenty firms with over 1.8 million workers in nearly 10,000 establishments nation-wide. See Bendick Rebuttal Decl. ¶ 12. Furthermore, if Dr. Bendick had compared Wal-Mart to small employers, he would have been subject to criticism for comparing apples to oranges.

Based on his determination that Wal-Mart has a shortfall of women being promoted to in-store management positions, Dr. Bendick also concludes that the shortfall cannot be explained in terms of lack of qualifications, interest, or availability among female employees. This conclusion appears to follow logically from the analysis of the data. While it is not for the Court at this stage to find whether or not the analysis and conclusion are correct, the inference is appropriately raised. Accordingly, the Court finds that Dr. Bendick's analysis further supports Plaintiffs' showing of commonality by providing additional evidence which is sufficient, for purposes of this motion, to raise an inference of class-wide discrimination.

3. Anecdotal Evidence of Discrimination

Circumstantial and anecdotal evidence of discrimination is commonly used in Title VII pattern and practice cases to bolster the statistical proof by bringing " the cold numbers convincingly to life." Int'l Brotherhood of Teamsters v. United States, 431 U.S. 324, 339, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977); Rudebusch v. Hughes, 313 F.3d 506, 517 (9th Cir.2002). Here, plaintiffs have submitted declarations from each of the class representatives, as well as 114 declarations from class members around the country. The women testify to being paid less than similarly situated men, being denied promotion or being delayed in promotion in a disproportionate manner compared with similarly situated men, working in an atmosphere with a strong corporate culture, and being subjected to various individual sexist acts. See, e.g., Scott Decl. ¶ ¶ 8, 9, 12 (male Store Manager told declarant that " [m]en are here to make a career and women aren't. Retail is for housewives who just need to earn extra money" ); S. Hall Decl. ¶ 2 (after seeking transfer to Hardware, male Support Manager stated: " We need you in toys ... you're a girl, why do you want to be in Hardware?" ); Page Decl. ¶ 10 (Store Manager gave sporting goods department manager position to male because she " needed a man in the job" ). This anecdotal evidence, in combination with the other evidence previously discussed, further supports an inference that Defendant's policies and procedures have the effect of discriminating against Plaintiffs in a common manner.

4. Conclusion re Commonality

In sum, Plaintiffs have exceeded the permissive and minimal burden of establishing commonality by providing: (1) significant evidence of company-wide corporate practices and policies, which include (a) excessive subjectivity in personnel decisions, (b) gender stereotyping, and (c) maintenance of a strong corporate culture; (2) statistical evidence of gender disparities caused by discrimination; and (3) anecdotal evidence of gender bias. Together, this evidence raises an inference that Wal-Mart engages in discriminatory practices in compensation and promotion that affect all plaintiffs in a common manner. While Defendant raises numerous challenges to this evidence, they are insufficient to defeat Plaintiffs' showing of commonality; rather, the objections are predominantly of the type that go to the weight of the evidence, and thus should properly be addressed by a jury considering the merits.

C. Typicality

Typicality requires that the named plaintiffs be members of the class they represent and " ‘ possess the same interest and suffer the same injury’ " as class members. Falcon, 457 U.S. at 156, 102 S.Ct. 2364 (citation omitted).⁴¹ The named plaintiffs' claims need not be identical to the claims of the class to satisfy typicality; rather, the claims are typical if they are " reasonably co-extensive with those of absent class members." Staton, 327 F.3d at 957, quotingHanlon, 150 F.3d at 1020. It is sufficient for plaintiffs' claims to " arise from the same remedial and legal theories" as the class claims. Arnold, 158 F.R.D. at 449 (citations omitted); see alsoMcClain v. Lufkin Indus., Inc., 187 F.R.D. 267, 281 (E.D.Tex.1999) ( quotingLightbourn v. County of El Paso, Texas, 118 F.3d 421, 426 (5th Cir.1997)) (typicality is met where " [i]n the event the class members in this case were to proceed in a parallel action, they would advance legal and remedial theories similar, if not identical, to those advanced by the named plaintiffs" ). The Ninth Circuit interprets typicality, like it does commonality, permissively. Staton, 327 F.3d at 957; Hanlon, 150 F.3d at 1020.

41 Just as commonality and typicality tend to merge, so do typicality and adequacy of representation. See Newberg, Vol. I at 411-12. The adequacy test is broader, however, in that a representative may have typical claims but otherwise have an irreconcilable conflict with the class. Id.

Defendant contends that the named plaintiffs are not sufficiently " typical" for two reasons: (1) the named plaintiffs have not worked in the upper levels of in-store management and therefore cannot represent those managers, and (2) each of the plaintiffs' claims is too individual-specific.

1. Whether the Named Plaintiffs Can Represent All In-Store Managers

There is no dispute that the named plaintiffs are reasonably co-extensive with the hourly class members since almost all of the named plaintiffs hold hourly positions. Defendant, argues, however, that the named plaintiffs are not typical of all female in-store managers because only one named plaintiff-Christine Kwapnoski-holds a salaried management position, and it is a lower level position at a Sam's Club. As such, Defendant emphasizes, " [n]o Plaintiff ever occupied a salaried position at a Wal-Mart Discount Store, Neighborhood Market or Supercenter." Def.'s Opp'n at 9.⁴² As discussed earlier, however, the management structure at Sam's Club is very similar to the structure at the other Wal-Mart divisions. Therefore, there is very little practical import to this argument.

42 After the hearing on the Motion for Class Certification, Defendant filed two separate ex parte applications further challenging plaintiff Kwapnoski's status as a class representative for female in-store managers. See Defendant's Ex Parte Applications filed January 5 and March 25, 2004. In both Applications, defendant points out that Ms. Kwapnoski participated as a claimant in a separate lawsuit filed in state court challenging Wal-Mart and Sam's Club's overtime policies. In the context of the state suit, she filed a claimant response form indicating that in both of her managerial positions she spent " more than one-half of [her] time performing the same type of non-managerial tasks as [her] hourly subordinates." Keane Decl. in Support of Defendant's Ex Parte Application, Exh. A at 4. The Court finds that this statement does not undermine or limit Ms. Kwapnoski's representative status. Whether Wal-Mart chose to have plaintiff (and other managers) performing hourly tasks is of no import to the typicality analysis, regardless of its relevance to a wage and hour analysis. The single relevant fact here is that during the span of this action, Ms. Kwapnoski held jobs that Wal-Mart itself designated as salaried managerial positions. Further, Wal-Mart's recent re-designation of plaintiff's positions as non-exempt does not alter, and cannot erase, this historical fact. Thus, Defendant's ex parte applications are denied.

Moreover, there is no requirement that plaintiffs have a class representative for each management category that they seek to represent. SeeTaylor v. Union Carbide Corp., 93 F.R.D. 1, 6 (S.D.W.Va.1980). It is sufficient that one of the named plaintiffs is a manager, even if her position is lower than the top rung. SeeHartman, 19 F.3d at 1471 (employee can challenge discrimination in " different job categories where the primary practices used to discriminate in the different categories are themselves similar. While it may be prudent to have the class divided into sub-classes represented by a named plaintiff from each of the differing job categories, it would not be necessary to the validity of the class certification to do so." ); Meyer v. Macmillan Publ'g Co., 95 F.R.D. 411, 414 (S.D.N.Y.1982) ( " [T]he fact that the jobs performed by the named plaintiffs are, in some sense, unique, is not a bar to their being class representatives. If it were, no class of professional employees could ever be certified." ); Paxton v. Union Nat'l Bank, 688 F.2d 552, 562 (8th Cir.1982) (holding that " [t]ypicality is not defeated because of the varied promotional opportunities at issue, or the differing qualifications of plaintiffs and class members." ). Furthermore, the class representatives include several women who claim that they were denied positions into management. If Plaintiffs sought to represent all levels of management, including high-level managers in the Home Office, there would be greater cause for concern. However, since the range of managers in the class is limited to those working in the stores, it is not a very broad class and a named plaintiff occupying a lower level salaried in-store management position is sufficient to satisfy the " permissive" typicality requirement. Staton, 327 F.3d at 957.

2. Degree of Individual Specificity for Each Claim

Defendant also argues that the named representatives' claims are not typical of the class because each claim in this case is too " individual-specific." To support this contention, Defendant cites various class member declarations to show the fact-specific nature of each of their claims. See Def.'s Opp'n at 7 n. 9. Some degree of individualized specificity must be expected in all cases, however, and it does not necessarily defeat typicality. SeeStaton, 327 F.3d at 957 (" representative claims are ‘ typical’ if they are reasonably coextensive with those of absent class members; they need not be substantially identical." ); Adams v. Pinole Point Steel Co., 1994 WL 515347, at *7 (N.D.Cal. May 18, 1994); Wofford v. Safeway Stores Inc., 78 F.R.D. 460, 488 (N.D.Cal.1978). Rather, the court must consider whether the named plaintiffs suffered injury from a specific discriminatory practice of the employer in the same manner that the members of the proposed class did, and whether the named plaintiffs and the class members were injured in the same fashion by a general policy of employment discrimination. Staton, 327 F.3d at 957. Thus, even where individual employees in different stores with different managers received different levels of pay and were denied promotion or promoted at different rates, so long as the discrimination they allegedly suffered occurred through an alleged common practice-e.g. excessively subjective decision-making in a corporate culture of uniformity and gender stereotyping-their claims are sufficiently typical to satisfy Rule 23(a)(3).⁴³

43 Another approach to this issue is to ask whether the adjudication of the class representatives' claims will necessarily involve the determination of common questions affecting the class. See Newberg, Vol. I at 387. Here, while the particular disparities in levels of pay and denial or delay in promotion will be specific to each class representative, each one will have to link those disparities to discrimination. The linkage for each and every plaintiff is the common theory of excessive subjectivity affected by a biased corporate culture and the common statistical data showing a pattern and practice of discrimination. The Court finds that the legal claims made by the representative plaintiffs are not idiosyncratic; rather, they are typical of the claims available to all class members.

D. Adequacy of Representation

The fourth and final Rule 23(a) requirement-adequacy-requires (1) that the proposed representatives do not have conflicts of interest with the proposed class and (2) that plaintiffs are represented by qualified counsel. Hanlon, 150 F.3d at 1020. The adequacy of representation requirement serves two functions. Like the commonality and typicality requirements, it assures that shared interests between the representative plaintiffs and the class as a whole are present, but it also functions to ensure the absence of conflicts. Amchem Products, Inc. v. Windsor, 521 U.S. 591, 625-26, 117 S.Ct. 2231, 138 L.Ed.2d 689 (1997). To serve as representatives, the named plaintiffs must have common interests with class members and a lack of interests adverse to class members. They must also vigorously prosecute the interests of the class through qualified counsel. East Texas Motor Freight System Inc. v. Rodriguez, 431 U.S. 395, 403-05, 97 S.Ct. 1891, 52 L.Ed.2d 453 (1977). This standard requires similarity, not identity, of interests. Nor does it preclude some unique interests; it only precludes adverse interests.

1. Conflicts of Interest

Defendant argues that certification should be denied because there is a conflict of interest among female in-store managers who are both plaintiff class members and decision-making agents of the Defendant. As the Ninth Circuit made clear in Staton, however, courts need not deny certification of an employment class simply because it may include both supervisory and non-supervisory employees. In Staton, objectors to a class-wide settlement similarly argued that a conflict arose from the fact that " the class cuts across the levels of authority of [the company] employees and that, in particular, some class members supervise some of their fellow class members." 327 F.3d at 958. The court rejected a per se rule, noting that the issue must be resolved in the context of the particular case and circumstances. Id. The court then concluded there was no conflict in that case given that (1) the objectors had failed to identify a " substantive issue for which there is a conflict of interest between two or more sets of employees," (2) the named plaintiffs included both supervisors and non-supervisors, and (3) the requested relief applied throughout the class. Id. at 958-59.

Here, as in Staton, the named plaintiffs include persons who have worked as both supervisory and non-supervisory employees, and the requested injunctive and lost pay relief applies throughout the class. Further, the alleged discriminatory policies affect supervisory and non-supervisory employees alike. Accordingly, the Court is not persuaded that there are any substantive conflicts between supervisory and non-supervisory employees that would preclude certification. See alsoShores, 1996 WL 407850, *8 (finding no inherent conflict where class included supervisory and non-supervisory employees since " there is nothing to suggest that the elimination of sex discrimination for non-managerial employees would adversely affect managerial employees and vice versa" ); Butler I, 1996 WL 421436, at *4 (noting that any potential conflict created by supervisors and non-supervisors being members of the same class is cured by bifurcating the trial into liability and remedial phases).

Wal-Mart also conjures a scenario where " at trial we [would] have the very real specter of class counsel cross-examining a woman he represents, trying to rebut her testimony that her subordinates were fairly compensated, attacking her credibility." Def.'s Opp'n at 31 (emphasis in original). It is unlikely, however, that many, if any, such scenarios would occur given that the liability phase of the trial properly focuses on the existence of a general pattern and practice of discrimination, not individual employment decisions. See section II.B.1. Defendant's speculation on this point does not warrant a finding that the named plaintiffs can not adequately represent the class.

More fundamentally, the fact that some individual female managers may disagree with the allegations in Plaintiffs' complaint, and testify in favor of Wal-Mart, does not create any significant conflict between supervisory and non-supervisory employees. Rather, it just reflects the fact that in any employment class there may be individuals that support the employer's position. Such individual predilections, however, do not create a substantive conflict between supervisory and non-supervisory employees as a group. ⁴⁴

44 The Court is aware of a few district court decisions in which class actions have been rejected or narrowed on the ground of a conflict of interest where the class included supervisors who might have made discriminatory decisions. However, none of these cases provide a persuasive analysis of how this conflict would cause a problem in actuality. See Donaldson, 205 F.R.D. at 568 (class action denied in part because of an " insurmountable" conflict between supervisor class representatives who made the decisions and the other members); Abram, 200 F.R.D. at 433 (in denial of class certification, court mentioned that three of the named plaintiffs were promoted to central managers and had a stake in proving they do not discriminate).

Finally, Defendant points to Wagner v. Taylor, 836 F.2d 578, 595 (D.C.Cir.1987), a Title VII case in which the court found inadequate representation because the named plaintiff, who was a supervisor, purported to represent a broad class that included non-supervisory employees. In Staton, the Ninth Circuit distinguished Wagner because " Wagner was both a senior executive and the only representative of an attempted class of all grade GS-9 and above African-American employees of, and applicants to, the Interstate Commerce Commission." Staton, 327 F.3d at 958. Thus, while Wagner cautions against having a single senior executive representing a broad range of supervisory and non-supervisory employees, the multiple named plaintiffs here provide significantly broader representation for the class.

2. Qualified Counsel

Plaintiffs have submitted declarations that sufficiently establish their counsel's abilities to handle a case of this nature and magnitude. Defendant does not contest the adequacy of class counsel. Based on the vigorous and skilled prosecution of this action thus far, as well as counsel's past experience with complex employment litigation, the Court finds that Plaintiffs' counsel have demonstrated their willingness, capacity, and dedication to pursuing these claims. Pursuant to the recently added subsection (g)(1)(C)(I) to Rule 23 (effective December 1, 2003), the Court also specifically finds as follows: (1) Plaintiffs' counsel have been diligent in identifying and investigating all potential claims in this litigation; (2) Plaintiffs' counsel individually and collectively have a wealth of extensive experience in handling class actions and other complex civil rights litigation; (3) plaintiffs' counsel have proved themselves to have an excellent grasp of employment discrimination and class action law; and (4) Plaintiffs' counsel have the resources to commit to fully representing the class throughout this litigation.

E. Conclusion

Without making any findings on the merits, the Court concludes, having reviewed the voluminous evidence provided by the parties, that Plaintiffs have come forth with sufficient evidence to satisfy the requirements of numerosity, commonality, typicality, and adequacy of representation, as set forth in Rule 23(a). Accordingly, the Court now considers whether plaintiffs can also satisfy the elements of Rule 23(b).

II. Maintainability Under Rule 23(b)

In addition to satisfying the prerequisites of Rule 23(a), Plaintiffs must also demonstrate that their proposed class is maintainable under one of the subsections of Rule 23(b). In this case, Plaintiffs rely on subsection (b)(2), which applies if " the party opposing the class has acted or refused to act on grounds generally applicable to the class, thereby making appropriate final injunctive relief or corresponding declaratory relief with respect to the class as a whole."

As noted in the introduction, Congress added subsection (b)(2) in 1966 " ‘ primarily to facilitate the bringing of class actions in the civil rights area.’ " In re Monumental Life, 365 F.3d at 417 n. 16 (citation omitted). Indeed, this provision was written with employment discrimination specifically in mind. " [T]he drafters of Rule 23 specifically contemplated that suits against discriminatory hiring and promotion policies would be appropriately maintained under Rule 23(b)(2)." Domingo v. New England Fish Co. (Domingo II), 727 F.2d 1429, 1443 n. 11 (9th Cir.1984); see alsoColey v. Clinton, 635 F.2d 1364, 1378 (8th Cir.1980); Wetzel v. Liberty Mutual Ins. Co., 508 F.2d 239, 250 (3rd Cir.1975).

Here, Plaintiffs allege that Wal-Mart has acted or refused to act-through its discriminatory practices-on grounds generally applicable to the proposed class, which Plaintiffs seek to remedy through injunctive relief and the recovery of lost pay and punitive damages. Wal-Mart does not dispute that the claims for injunctive relief and lost pay readily fall within the ambit of a(b)(2) class action. Indeed, it is well established that lost pay is recoverable as an equitable, make-whole remedy in employment class actions notwithstanding its monetary nature. See, e.g.,Gotthardt v. Nat'l R.R. Passenger Corp., 191 F.3d 1148, 1152-55 (9th Cir.1999); Eubanks v. Billington, 110 F.3d 87, 92 (D.C.Cir.1997) (" [I]t is not uncommon in employment discrimination [ (b)(2) ] cases for the class also to seek monetary relief in the form of back pay or front pay." ); Salinas v. Roadway Express, 735 F.2d 1574, 1576 (5th Cir.1984).

Defendant argues, however, that inclusion of the claim for punitive damages renders this case unsuitable for certification under section (b)(2). Defendant further contends that this case is not maintainable as a class action under (b)(2)-or any other subsection-because the size of the class makes the case wholly unmanageable. Each of these issues is addressed in turn.

A. Inclusion of Claim for Punitive Damages

In 1991, Congress amended Title VII to permit plaintiffs to recover punitive damages in certain circumstances. Specifically, such damages are recoverable in cases of intentional discrimination if the plaintiff proves that the employer " engaged in a discriminatory practice or discriminatory practices with malice or with reckless indifference to the federally protected rights of an aggrieved individual." 42 U.S.C. § 1981a(b)(1).

While plaintiffs have foregone compensatory damages in this case, they do seek punitive damages to punish Wal-Mart for its allegedly " reckless disregard of the rights of its women employees to equal employment opportunity, and to deter similar misconduct by Wal-Mart and other large retailers in the future." Pls.' Mot. at 44. Defendant argues, however, that Plaintiffs' claim for punitive damages is incompatible with a(b)(2) action because such damages are not incidental to, and would in fact overwhelm, Plaintiffs' claims for injunctive relief.

Resolution of this issue is governed by Molski v. Gleich, 318 F.3d 937 (9th Cir.2003), which holds that (b)(2) class actions can include claims for monetary damages so long as such damages are not the " predominant" relief sought, but instead are " secondary to the primary claim for injunctive or declaratory relief." Id. at 947, 950; see alsoProbe v. State Teachers' Ret. Sys., 780 F.2d 776, 780 (9th Cir.1986). In making this determination, courts should examine the " specific facts and circumstances of each case," and attempt to ascertain the " intent of the plaintiffs in bringing suit." Molski, 318 F.3d at 950. In Molski, the injunctive relief " appeared to be the primary goal in the litigation," and thus the district court properly found that injunctive relief was the predominant remedy sought by the class. Id.

Notably, Molski expressly rejects using a bright line distinction between incidental ⁴⁵ and non-incidental damages to determine if monetary relief predominates, reasoning that such an approach would " nullify the discretion vested in the district courts through Rule 23." Id. Rather, it endorses the case-by-case approach described above. Id. Accordingly, Defendant's contention that this case cannot be certified under (b)(2) because the punitive damages sought are not " incidental" is misplaced. As Molski also makes clear, the predominance test turns on the primary goal of the litigation, not the potential size of a punitive damage award. Id. Accordingly, Defendant's focus on the possible size of a such an award is also misplaced. Indeed, focusing on the potential size of a punitive damage award would have the perverse effect of making it more difficult to certify a class the more egregious the defendant's conduct or the larger the defendant. Such a result hardly squares with the remedial purposes of Title VII.

45 Incidental damages are defined as damages that " ‘ flow directly from liability to the class as a whole on the claims forming the basis of the injunctive or declaratory relief.’ " Molski, 318 F.3d at 949 (emphasis in original) (citation omitted).

This Court has little difficulty concluding that here the equitable relief sought predominates over the claim for punitive damages. Plaintiffs' claims for injunctive and declaratory relief, if successful, would achieve very significant long-term relief in the form of fundamental changes to the manner in which Wal-Mart makes its pay and promotions decisions nationwide that would benefit not only current class members, but all future female employees as well. These benefits would be both monetary-in the form of increased class-wide remuneration-and non-monetary-in terms of increased job opportunities. SeeYoung v. Pierce, 544 F.Supp. 1010, 1028 (E.D.Tex.1982) (" [W]hen the relief sought is injunctive relief, the benefits...would inure not only to known class, but also to a future class of indefinite size." ). Recovery of lost pay for all injured class members would result in additional substantial equitable relief to the class. Against this backdrop, Plaintiffs' claim for punitive damages-which, by its nature, is subject to more demanding proof requirements, and thus inherently more speculative-appears secondary in nature.

The named class representatives also affirm that their central motivation for participating in this action is to improve opportunities for women at Wal-Mart. Betty Dukes, for example, states that " [m]y primary goal [in this litigation] is to ensure that the employment practices at Wal-Mart which hinder the progress of women wishing to enter management be changed to ensure fair and equitable treatment of female employees, and to ensure women receive equal pay." Dukes Decl. ¶ 20. Edith Arana similarly avers that " [m]y main concern is to end all those employment practices at Wal-Mart that have prevented women from obtaining management positions and to ensure equal pay for comparable work and equal access to the training and mentoring necessary to advance in the Company." Arana Decl. ¶ 3. See also Surgeson Decl. ¶ 13; Gunter Decl. ¶ 3; Kwapnoski Decl. ¶ 23.

Given all of the above, the Court is satisfied that Plaintiffs' claim for punitive damages is secondary to their primary goal of achieving broad equitable relief. Accordingly, inclusion of this claim does not preclude certification of this case under Rule 23(b)(2). Molski, 318 F.3d at 947-50. See alsoRobinson v. Metro-North Commuter R.R. Co., 267 F.3d 147, 164 (2nd Cir.2001) (district court may certify class under (b)(2) if it finds in its discretion that the positive weight or value of the injunctive relief sought is predominant even though punitive damages are claimed); Butler I, 1996 WL 421436, *1, *5 (rejecting defendant's argument that punitive damage claim would " overwhelm" claims for injunctive relief as conclusory and speculative); Barefield v. Chevron, 1988 WL 188433, *3 (N.D.Cal.1988) (class claim for punitive damages did not render monetary aspect of case predominant where the injunctive and declaratory relief sought was the heart of the action).

The Court further notes that a claim for punitive damages does not detract from the homogeneity or cohesiveness of the class. Rather, it is " consistent with the notion that the focus of a (b)(2) action is the defendant's conduct toward persons sharing a common characteristic." Barefield, 1988 WL 188433, *3. Indeed, since the purpose of punitive damages is not to compensate the victim, but to punish and deter the defendant, a punitive damage claim focuses " not on facts unique to each class member, but on the defendant's conduct toward the class as a whole." Id., citingJenkins v. Raymark Industries, 782 F.2d 468, 474 (5th Cir.1986), reh'g denied, 785 F.2d 1034 (1986); see alsoKolstad v. Am. Dental Ass'n, 527 U.S. 526, 536, 119 S.Ct. 2118, 144 L.Ed.2d 494 (1999) (focus of punitive damage claim is on defendant's conduct). Accordingly, the " cohesive core of the (b)(2) action is ... maintained." Barefield, 1988 WL 188433, *3.

Wal-Mart also argues that Plaintiffs' class claim for punitive damages is foreclosed by the Supreme Court's recent decision in State Farm Mut. Auto. Ins. Co. v. Campbell, 538 U.S. 408, 123 S.Ct. 1513, 155 L.Ed.2d 585 (2003). Nothing in State Farm, however, supports this supposition. In State Farm the Court held that a punitive damage award in an individual action improperly punished the defendant for conduct that " bore no relation to the [plaintiff's] harm." Id. at 422, 123 S.Ct. 1513. Specifically, it found that the jury improperly considered conduct by State Farm that occurred in other states and did not directly affect the plaintiff. Id. at 422-23, 123 S.Ct. 1513. As such, it underscored the basic proposition that a " defendant should be punished for the conduct that harmed the plaintiff, not for being an unsavory individual or business." Id. at 423, 123 S.Ct. 1513. Such a principle is not, as Wal-Mart suggests, incompatible with the recovery of punitive damages in a class action.

First, courts can ensure that any award of punitive damages to the class is based solely on evidence of conduct that was directed toward the class. Second, as Plaintiffs propose here, courts can limit recovery of any punitive damages to those class members who actually recover an award of lost pay, and thus can demonstrate that they were in fact personally harmed by the defendant's conduct. Finally, courts also can ensure that any punitive damage award is allocated among the lost pay class in reasonable proportion to individual lost pay awards. Accordingly, this Court is satisfied that procedures exist that permit Plaintiffs' punitive damage claim to be managed in a manner fully consistent with the principles of State Farm.⁴⁶

46 In Jenkins, 782 F.2d at 474-75, for example, the Fifth Circuit affirmed a district court's certification of a punitive damage claim in a 23(b)(3) class action, noting that the jury could determine the defendants' liability to the class for punitive damages in one phase while still ensuring that only class members who are harmed recover such damages. " While no plaintiff [class member] may receive an award of punitive damages without proving that he suffered actual damages .... the allocation need not be made concurrently with an evaluation of the defendant's conduct." Id. at 474 (internal citations omitted). Rather, such determinations could be handled in later proceedings. Id. The court further suggested that jurors " could be allowed to award an amount of money that each class member should receive for each dollar of actual damages awarded." Id.

While plaintiffs' claim for punitive damages does not preclude certifying this case under Rule 23(b)(2), minimum due process requirements do apply where the monetary relief sought, albeit not " predominant," is nonetheless " substantial." Molski, 318 F.3d at 947-48. Recognizing the Supreme Court's " growing concerns regarding the certification of mandatory classes when monetary damages are involved," the Molski court held that " notice and the right to opt-out [ ] must be provided to bind absent class members when substantial monetary damages are involved." Id. at 948. The Court further indicated that any claim for punitive damages is " substantial." Id. at 951 (" Because the statutory damages ... provide for treble [i.e. punitive] damages, the remedy must be considered substantial" ); cf.In re Monumental Life, 365 F.3d at 416-17 (due process requires provision of notice where Rule 23(b)(2) class seeks monetary damages, while provision of opt-out rights is optional).⁴⁷

47 The type of notice utilized in a(b)(2) action, however, need not always be " equivalent to that required in (b)(3) actions." Johnson v. Gen. Motors Corp., 598 F.2d 432, 438 (5th Cir.1979).

Molski further notes that notice and the opportunity to opt-out can be provided in a(b)(2) class action. 318 F.3d at 951 n. 16. Indeed, the district court's discretion in this regard is well established. See, e.g.,In re Monumental Life, 365 F.3d at 417 (district courts have discretion to order notice and opt-out rights when certifying a Rule 23(b)(2) class); Robinson, 267 F.3d at 165-67 (notice and opt out can be afforded (b)(2) class members with respect to non-incidental damage claims); Jefferson v. Ingersoll Int'l Inc., 195 F.3d 894, 898-99 (7th Cir.1999) (same); Eubanks, 110 F.3d at 94. Accordingly, notice and an opportunity to opt-out shall be provided to the plaintiff class with respect to Plaintiffs' claim for punitive damages.

B. Manageability of Liability and Remedy Stages

A critical issue raised by Plaintiffs' motion is whether a class action of the size proposed-which is huge both in sheer numbers of class members and in its nationwide geographic scope-is manageable. While Rule 23(b)(2) does not expressly refer to manageability-in contrast to Rule 23(b)(3)-such a requirement is implicit in any type of class certification. See, e.g.,Robinson, 267 F.3d at 164 (District court may allow (b)(2) certification if " class treatment would be efficient and manageable, thereby achieving an appreciable measure of judicial economy." ); Piva v. Xerox Corp., 70 F.R.D. 378, 388 (N.D.Cal.1975) (noting that proposed 23(b)(2) class of employees in Western Region " directly involves important considerations of manageability " ). Certainly, no court would certify a class unless it believed that the case could proceed in a manageable fashion.

Defendant contends that this case would be completely unmanageable in both the liability and remedial phases. Indeed, as Defendant repeatedly emphasizes, no court has ever confronted a motion for class certification in an employment discrimination case of this size-or anything close to it. This fact alone, however, does not preclude the Court from acting. Nonetheless, the Court agrees that it must very carefully assess the manageability issues presented by this unique case. While courts possess wide discretion to flexibly respond to manageability issues that may arise during the course of a class action, see, e.g.,Blackie v. Barrack, 524 F.2d 891, 906, n. 22 (9th Cir.1975), this Court must be confident that such issues will not be of such a magnitude as to defy its ability to oversee this case in a responsible and reasonable manner.

Having given these matters considerable thought and deliberation, this Court is satisfied, for the reasons expressed below, that the liability stage of this case is not rendered unmanageable by the size of the proposed class. With respect to the remedial phase of this case, this Court also is amply satisfied that the size of the class would not present undue obstacles to managing Plaintiffs' requests for injunctive relief on both their unequal pay and promotion claims. The Court similarly finds that fashioning a lost pay remedy for Plaintiffs' unequal pay claim would be manageable. Finally, with respect to Plaintiffs' promotion claim, the Court concludes that a lost pay remedy would be manageable only for a portion of the class. The manageability of each of these aspects of this case is discussed in turn below.

1. Manageability of Liability Stage

The standard approach in Rule 23(b)(2) employment discrimination class actions is to bifurcate the case into two stages: liability and remedy. Robinson, 267 F.3d at 158, Arnold, 158 F.R.D. at 458-59. In the Stage I liability proceeding, plaintiffs are required to prove that the defendant engaged in a pattern and practice of discrimination against the class. Robinson, 267 F.3d at 158. If punitive damages are at issue, plaintiffs also must prove liability for such damages by showing that the pattern and practice of discrimination was undertaken maliciously or recklessly in the face of a perceived risk that defendant's actions would violate federal law. SeeKolstad, 527 U.S at 535, 119 S.Ct. 2118, 144 L.Ed.2d 494 (1999); Butler I, 1996 WL 421436, *1. Finally, defendant's liability for the class representatives' individual claims usually is determined in Stage I as well.

Defendant argues that determining its liability for the alleged pattern and practice of discrimination against the class would be unmanageable because " due process" would necessarily require 3,244 individual mini-trials, including testimony from 3,244 Store Managers, to determine whether each individual Wal-Mart store discriminated against class members employed at that store. Defendant cites no authority for this argument. See Def.'s Opp'n at 45. Nor is it persuasive since it is inconsistent with the fundamental character of the class action proceeding.

As courts have long recognized, the Stage I liability determination in a systemic disparate treatment case focuses on the existence (or not) of a company-wide pattern or practice of discrimination against the class. While some acts of discrimination against individual class members are usually introduced as anecdotal evidence to bring statistical evidence " convincingly to life," Teamsters, 431 U.S. at 339, 97 S.Ct. 1843, the class is not required to prove that each member suffered discrimination. Id. at 360, 97 S.Ct. 1843. Indeed such an approach would defeat the very efficiencies that the class action device is designed to achieve. Instead, it is plaintiffs' burden to prove that " discrimination was the company's standard operating procedure." Id. at 336, 97 S.Ct. 1843.

As such, the evidence introduced during Stage I should properly focus on matters relevant to the class as a whole, such as statistical analysis and evidence of system-wide policies and practices. " [T]he stress at Stage I is upon demonstration of the defendant's broad employment policies and practices ...." United States v. United States Steel Corp., 520 F.2d 1043, 1053 (5th Cir.1975); see alsoCooper v. Fed. Reserve Bank, 467 U.S. 867, 876, 104 S.Ct. 2794, 81 L.Ed.2d 718 (1984); Teamsters, 431 U.S. at 339-40, 97 S.Ct. 1843 (statistics are valid tool for proving policy of employment discrimination); Robinson, 267 F.3d at 158-59; Thiessen v. General Electric Capital Corp., 267 F.3d 1095, 1106 (10th Cir.2001). Likewise, defendants are expected to respond to plaintiffs' pattern and practice claim on a class-wide basis. Teamsters, 431 U.S. at 360 n. 46, 97 S.Ct. 1843; Thiessen, 267 F.3d at 1107 (" common defenses" not " individualized defenses" were relevant at liability stage). " The point is that at the liability stage of a pattern-or-practice trial the focus often will not be on individual hiring decisions, but on a pattern of discriminatory decision-making." Teamsters, 431 U.S. at 360 n. 46, 97 S.Ct. 1843 (emphasis added). In short, once the court determines that there are sufficient common questions to justify proceeding as a class, the liability phase properly focuses on evidence that affects the class generally-not on individuals or other minor segments of the class.

Given that the scope of the instant class is company-wide, Defendant is no more entitled to 3,244 individual store-by-store trials than it would be entitled to try each class member's individual claim in a case of smaller scope. Notably, Teamsters involved " nationwide operations" at 51 terminals in 26 states, 431 U.S. at 328, 329 n. 2, 97 S.Ct. 1843; yet, the Court nowhere suggests that the employer in that case was entitled to defend itself on a terminal-by-terminal basis. Wal-Mart of course can defend against Plaintiffs' claim of a nationwide pattern and practice by arguing that its more decentralized store sub-unit by store sub-unit statistical analysis refutes the existence of any company-wide policy of discrimination. It also can introduce evidence to rebut any evidence Plaintiffs present of centralized, nationwide policies regarding Defendant's corporate culture, personnel policies or gender stereotyping. It is not, however, entitled to circumvent or defeat the class nature of the proceeding by litigating whether every individual store discriminated against individual class members.

A Stage I liability trial in this case would be of course a substantial undertaking-albeit no more, and probably less, so than some other complex litigation, such as certain anti-trust or patent cases. Defendant's contention that it would necessarily be unmanageable, however, is simply not persuasive.

2. Manageability of Remedy Phase

In the event that a plaintiff class establishes liability in Stage I, and the case advances to the remedy phase, courts will first " proceed to fashion class-wide injunctive relief." Robinson, 267 F.3d at 159. Defendant does not argue that this aspect of the case would present any manageability issues. Second, the court must address the remedy of lost pay (backpay and/or front pay), which usually is awarded " in order to effectuate the statutory goal [of Title VII] of compensating the victimized employee and placing him in as good a position as he would have been had he not been subject to discrimination," as well as to deter recalcitrant employers. Stewart v. General Motors Corp., 542 F.2d 445, 451 (7th Cir.1976). Finally, if liability to the class for punitive damages has been established, this issue must be addressed as well.

As previously noted, Wal-Mart contends that any attempt to provide a lost pay remedy to a class of the size proposed here would mire the Court in a completely unmanageable endeavor. Plaintiffs respond that such problems can be avoided by employing a formula approach to streamline the process. Wal-Mart replies that formula approaches are discredited, and moreover, such an approach would still leave the Court with severe manageability issues.

In advancing their respective positions, both parties lump together Plaintiffs' promotion and equal pay claims. Consequently, both neglect to address the fact that the remedial challenges each claim presents are actually quite distinct. Having evaluated each claim separately, this Court concludes that while a formula approach is still a perfectly valid remedial tool in the appropriate case, it would not resolve all of the manageability problems presented by the promotion claim. The Court is convinced that a lost pay remedy with respect to the promotion claim can be effectively managed for only a limited subset of class members, as defined below. With respect to Plaintiffs' pay discrimination claim, the Court is well satisfied that, should Plaintiffs prevail, a process for awarding lost pay can be devised that is entirely manageable for all class members, albeit without reliance on a formula approach. Each claim is discussed in turn below. In addition, the Court addresses Defendant's contention that both the promotion and equal pay claims are rendered unmanageable by the Civil Rights Act of 1991.

a. Promotions Claim

(1) Overview of Traditional and Formula Approach to Backpay Remedy

As a general matter, courts do not begin the remedy stage on a blank slate. " ‘ [T]he force of [the proof from the liability phase] does not dissipate.’ " Teamsters, 431 U.S. at 361, 97 S.Ct. 1843. Rather, all class members benefit from a presumption in their favor " that any particular employment decision, during the period in which the discriminatory policy was in force, was made in pursuit of that policy." Id. at 362, 97 S.Ct. 1843; see alsoRobinson, 267 F.3d at 159; Domingo II, 727 F.2d at 1444-45 (" ‘ proof of a discriminatory pattern and practice creates a rebuttable presumption in favor of relief’ " for individual class members) (citation omitted).

While proof of a pattern of discrimination establishes a presumption in favor of class members, it does not in and of itself entitle each class member to back (or front) pay. Pettway v. Am. Cast Iron Pipe Co. (Pettway III), 494 F.2d 211, 259 (5th Cir.1974). Rather, only those class members who can make a showing that they were either actually harmed by the discriminatory policy or were at least " a potential victim of the proved discrimination" are eligible to recover an award of lost pay. Teamsters, 431 U.S. at 362, 97 S.Ct. 1843; Robinson, 267 F.3d at 159. Thus, courts awarding lost pay face two primary tasks at the remedy stage: (1) identifying the specific class members that were either actually, or at least potentially, harmed by the employer's discriminatory policies, and (2) determining the specific amount of back or front pay each such person is owed.

In cases involving hiring or promotions, such determinations are usually made during the course of additional proceedings. In Teamsters, for example-a hiring and promotions case-the Court explained that " a district court must usually conduct additional proceedings after the liability phase of the trial to determine the scope of individual relief." Id. at 361, 97 S.Ct. 1843. These proceedings typically consist of mini-hearings presided over by a special master or the court. While the burden on individual class members at this point is minimal, they must at least identify themselves and make some showing (less than a prima facie case) that they suffered an adverse employment decision. Robinson, 267 F.3d at 159. The burden then shifts to the employer to prove that the class member was denied the job or promotion for lawful reasons. Teamsters, 431 U.S. at 362, 97 S.Ct. 1843; Robinson, 267 F.3d at 159-60. The advantages of such an approach are obvious. It protects both the defendant and the class by ensuring that each victim, or at least potential victim, of the employer's discriminatory policy is correctly identified, and that each such victim is paid the correct amount to make him or her whole.⁴⁸

48 See Stewart, 542 F.2d at 452 (Backpay awards should be made on an individualized basis whenever possible " because it will best compensate the victims of discrimination without unfairly penalizing the employer." ); United States v. United States Steel Corp., 520 F.2d at 1055-1056 (" Apart from protecting the defendant, this method has the virtue of distributing the recovery to the victims who, by the greater likelihood, are entitled to it .... [W]e reemphasize that alternative methods possessed of superior certainty should be exhausted before the court resorts to racially-drawn classwide comparisons ...." ); Smith v. Union Oil Co., 1977 WL 77 (N.D.Cal. Aug.22, 1977), decision supplemented by, 1978 WL 13884 (N.D.Cal. June 15, 1978) (in most cases and where feasible, there should be a separate determination on an individual basis as to who is entitled to recover and the amount of such recovery).

As Plaintiffs implicitly concede, holding individual hearings for the number of women potentially entitled to backpay in this case is impractical on its face, and thus the traditional Teamsters mini-hearing approach is not feasible here. Plaintiffs contend, however, that a lost pay remedy can manageably be afforded the class through the use of a formula approach that obviates the need for individualized Teamster hearings for each class member. See Pls.' Mot. at 49 (" The lost wages due to the class can be determined by formula" ).

As Plaintiffs correctly point out, courts have, in certain circumstances, employed a formula to calculate a lump sum amount that represents the employer's total liability for backpay to the class. In Domingo II, 727 F.2d at 1443, for example, the Ninth Circuit held that departure from the preferred Teamsters approach is justified when subjective employment practices, including lack of objective hiring criteria and word-of-mouth recruitment, " make[ ] it difficult to determine precisely which of the claimants would have been given a better job absent discrimination, but it is clear that many should have." Domingo II, 727 F.2d at 1444. As the court further explained:

" [W]hen the class size or the ambiguity of promotion or hiring practices or the illegal practices continued over an extended period of time calls forth [a] quagmire of hypothetical judgment ... a class-wide approach to the measure of back-pay is necessitated."

Id., quotingPettway III, 494 F.2d at 261 (alterations in original); see alsoEEOC v. O & G Spring & Wire Forms Spec. Co., 38 F.3d 872, 879-880 & n. 9 (7th Cir.1994) (approving district court's use of formula approach); Hameed v. Int'l Ass'n of Bridge Workers, Local 396, 637 F.2d 506, 520, 521 & n. 18 (8th Cir.1980) (same) (and cases cited therein); Stewart, 542 F.2d at 452-53 (7th Cir.1976) (same) (and cases cited therein); EEOC v. Chicago Miniature Lamp Works, 668 F.Supp. 1150, 1151-52 (N.D.Ill.1987); Green v. United States Steel Corp., 640 F.Supp. 1521, 1525-26 (E.D.Pa.1986), aff'd in part and vacated in part on other grounds, 843 F.2d 1511 (3rd Cir.1988), vacated in part on other grounds, 490 U.S. 1103, 109 S.Ct. 3151, 104 L.Ed.2d 1015 (1989).

In short, while a formula approach is certainly not the norm, it is a potential option where the employer uses largely subjective criteria for hiring or promotion decisions, objective requirements are minimal, and many more class members qualified for the positions than would have been hired or promoted even absent discrimination. Because it is virtually impossible in such cases to determine which class members would actually have been hired or promoted (and thus which class members were the actual victims of the defendant's discriminatory policy), there is little point in going through the exercise of individual hearings. As the Seventh Circuit explained in Stewart:

In determining the appropriate award to be made to black employees who were unfairly denied promotions to salaried positions ... the utilization of an individualized method of calculation is impossible. Because General Motors had no objective standards by which to measure whether a given employee deserved a promotion, deciding in individual cases whether a particular person would have been promoted but for racial discrimination would lead the district court into a " quagmire of hypothetical judgments," in which any supposed accuracy in result would be purely imaginary.

542 F.2d at 452 (internal citation omitted); see alsoPettway III, 494 F.2d at 261 (In cases such as those described above, it is " not a choice between one approach more precise than another. Any method is simply a process of conjectures." ).

Even in those circumstances, however, where the court properly employs a formula to calculate a lump sum backpay award to the class, it still must undertake the additional steps of determining which individual class members are eligible to share in the lump sum award and allocating the award among those eligible members. Given that the actual victims cannot be identified, courts have focused instead on ascertaining which class members were at least potentially victimized by the employer's discriminatory policy. In a promotion case, this would include any class member who met the minimum qualifications for the job and was interested in a promotion, and either applied for (and was denied) a promotion, was deterred from applying, or was unable to apply because of lack of application procedures.

The cases acknowledge that awarding backpay to all potential victims of the employer's policy (as opposed to just the actual victims) has the effect of generating a " windfall for some employees who would have never been promoted had vacancies been filled on a [non-discriminatory] basis and undercompensat[ing] the genuine victims of discrimination by forcing them to share the award with their undeserving brethren." Stewart, 542 F.2d at 452-453. For example, in O & G Spring & Wire Forms, backpay for 17 positions that would have gone to class members absent discrimination was spread over 450 persons who identified themselves as having applied (or having been deterred from applying) for the positions at issue, 38 F.3d at 879 & n. 9; in Hameed, backpay for 45 apprentice positions was spread over a much larger group of " potential discriminatees," 637 F.2d at 519-20. The cases conclude, however, that this " rough justice" is better than the alternative of no remedy at all for any class member. Stewart, 542 F.2d at 453 (" Given a choice between no compensation for black employees who have been illegally denied promotion and an approximate measure of damages, we choose the latter" ); see alsoHameed, 637 F.2d at 521 n. 19; Segar v. Smith, 738 F.2d 1249, 1291 (D.C.Cir.1984) (" Though [Title VII] generally does not allow for backpay to those whom discrimination has not injured, this section should not be read as requiring effective denial of backpay to the large numbers of agents whom [defendant's] discrimination has injured in order to account for the risk that a small number of undeserving individuals might receive backpay." )

Defendant attempts to dismiss Domingo and other formula cases as inconsistent with Teamsters. Yet Domingo, O & G Spring & Wire Forms, Hameed, Chicago Miniature Lamp Works, and Green all post-date Teamsters. As these cases implicitly recognize, nothing in Teamsters precludes calculating a total backpay award that is allocated among potential victims, where the actual victims cannot realistically be identified even were the court to undertake individual hearings. On the contrary, Teamsters supports the notion that class members who can demonstrate that they have potentially been discriminated against are eligible to receive backpay. SeeTeamsters, 431 U.S. at 362, 97 S.Ct. 1843 (" The Government need only show that an alleged individual discriminatee ... was a potential victim of the proved discrimination" ); see alsoRobinson, 267 F.3d at 159 (" a class member at the remedial stage of a pattern-or-practice claim need only show that he or she suffered an adverse employment decision ‘ and therefore was a potential victim of the proved [class-wide] discrimination’ " ), quotingTeamsters, 431 U.S. at 362, 97 S.Ct. 1843; McKenzie v. Sawyer, 684 F.2d 62, 75 (D.C.Cir.1982), abrogated on other grounds byBerger v. Iron Workers Reinforced Rodmen, Local 201, 170 F.3d 1111, 1125 (D.C.Cir.1999) (To share in relief, class members " must make some showing that they were potential victims of the discriminatory employment practices" ). Accordingly, this Court rejects Defendant's contention that Domingo and the other formula cases cited above lack weight. Indeed, Domingo is binding precedent on this Court.

Furthermore, this case presents the type of situation a formula approach was intended to address. As discussed above in section I.B.1.a(2), Wal-Mart uses largely subjective criteria for its in-store promotion decisions, and objective requirements are minimal. Furthermore, it is likely that many more class members would qualify for the challenged positions than would have been promoted absent any proven discrimination, making it virtually impossible to determine which class members were actual victims. In short, the subjectivity built into Wal-Mart's promotion system would simply call forth the " quagmire of hypothetical judgments" warned of in Domingo. In addition, the class size is large and the alleged practices have " continued over an extended period of time." Domingo II, 727 F.2d at 1444. As such, this is precisely the kind of case in which " ‘ a class-wide approach to the measure of back-pay is necessitated.’ " Id.

Nonetheless, the Court still must have confidence that a formula approach can be applied in a manageable manner in this particular case. Here, Plaintiffs have suggested that it is unnecessary at this early juncture to either determine the specifics of the formula that would be used to calculate Defendant's backpay liability to the class or detail how the Court would identify the potential victims of Defendant's alleged discriminatory promotion policy. Indeed, Plaintiffs' briefs only discuss the formula process in the most conclusory terms. While a detailed blueprint certainly is not required, the general contours should be sufficiently defined, particularly in a case of this unique magnitude, such that the Court has confidence that it is not being called upon to embark upon a process that will ultimately prove unmanageable. It is to these issues the Court turns next.

(2) Application of Class-Wide Formula Approach

(a) Using a Formula to Calculate a Lump Sum Backpay Award

While the methods for constructing backpay formulas vary, most involve the following three steps: (1) identifying the number of positions that would have gone to class members absent discrimination, (2) estimating the total amount of earnings that would have accrued to class members in those positions, and finally (3) subtracting a mitigation amount that represents what the class members earned or would have earned during the period in question in any event. See, e.g., Domingo II, 727 F.2d at 1445 (" Determination of the [formula] award could proceed along any of several avenues, all of which are designed to estimate the difference between what non-whites actually earned and what they would have earned but for the discrimination" ); O & G Spring & Wire Forms, 38 F.3d at 880, n. 9 (detailing actual formula); Hameed, 637 F.2d at 520 (same); Stewart, 542 F.2d at 454; Pettway III, 494 F.2d at 262-63; Green, 640 F.Supp. at 1526.

While the development of a formula in this case plainly would be a complex undertaking, Defendant does not argue that it would be unmanageable (rather, Defendant focuses on the manageability of the eligibility issues discussed below). Moreover, courts may enlist the assistance of an expert or special master under Federal Rule of Civil Procedure 53. See, e.g., Pettway v.Am. Cast Iron Pipe Co. ( Pettway V), 681 F.2d at 1261 (11th Cir.1982) (formula could be decided by special master after a hearing and guidance from the court). At least at this stage of the litigation, the Court is satisfied that a formula for calculating Defendant's lump sum backpay liability to the class could be developed in a manageable manner given the Court's limitations on eligibility set forth below.⁴⁹

49 The Court also notes that where a defendant's total backpay liability to the class is set by formula in the form of a lump sum award, there is no need for the defendant to participate further in the issue of which class members are eligible to share in the award. Although Domingo cites Teamsters for the proposition that " the employer has the burden of proving that the applicant was unqualified or showing some other valid reason why the claimant was not, or would not have been, acceptable," 727 F.2d at 1445, it appears to have simply imported this procedure from Teamsters without considering whether it is necessary in a formula case. Clearly, in a traditional Teamsters model, the employer has a strong interest in contesting each individual claimant's qualifications and expressed interest in a promotion since the fewer the number of persons found to be victims of the employer's discriminatory policy, the less backpay the employer will owe. Thus, Teamsters logically provides that the employer has the opportunity to demonstrate that each claimant seeking backpay was not in fact discriminated against but rather was denied the position for lawful reasons. 431 U.S. at 362, 97 S.Ct. 1843.

(b) Determining Individual Eligibility

As noted above, even where a " formula approach" such as that endorsed in Domingo is invoked, the formula only advances the backpay remedy to the point of calculating the defendant's total backpay liability to the class. Courts still face the " eligibility phase" -i.e. the practical problem of identifying which class members were at least potential victims of the employer's discriminatory policy and thus are eligible to share in the lump sum backpay award. In a promotion case, such potential victims consist of those class members who were both (1) qualified for, and (2) interested in, a promotion but were denied that opportunity. These are two distinct factors that must be analyzed separately for manageability purposes. A class member who was qualified for, but not interested in, a promotion, would by definition not have been even potentially injured by defendant's discriminatory promotion policies, and thus would not be eligible to participate in the backpay award.

These potential victims typically constitute some subset of the class, since courts generally will not presume that every class member is both qualified for, and interested in, every promotional opportunity. See, e.g.,Teamsters, 431 U.S. at 369, 370 n. 55, 97 S.Ct. 1843 (" the desirability of the [promotion positions] is not so self-evident as to warrant a conclusion that all employees would prefer [the promotion] if given a free choice" ). McKenzie, 684 F.2d at 76 (" The benefits illegally denied to the plaintiffs as a class were opportunities neither automatically sought nor automatically bestowed .... [It is not] reasonable to assume that all journeymen would wish to assume supervisory responsibilities." ).

In the promotion formula cases relied upon by Plaintiffs, courts typically resolve the eligibility issue through additional proceedings wherein class members come forward, self identify, and provide some minimum information regarding their (1) qualifications for, and (2) interest in the job promotions at issue. For example, in Domingo, the Ninth Circuit stated as follows:

In order to be eligible for back-pay [from the formula-derived lump sum fund], claimants need only prove they applied for a position or would have applied if not for Nefco's discriminatory practices. They may be required to show what their qualifications were, but do not have the burden of proving they were qualified for the position sought .... In determining whether claimants applied or were deterred from applying for a better position, the district court should not put an unrealistic burden on claimants .... All uncertainties should be resolved against the employer.

727 F.2d at 1445; see alsoStewart, 542 F.2d at 453; Pettway V, 681 F.2d at 1266 (employee need only come forth with statement of employment history, jobs denied, and qualifications, not conclusive proof of such matters).

As Plaintiffs concede, such additional proceedings would be unmanageable in this case. They argue, however, that these eligibility determinations could be handled solely by reference to objective data contained in corporate records, which would eliminate the need for additional, individualized eligibility proceedings. " The determination of which class members would be entitled to backpay ... could be drawn from the economic models that each side's experts create" which account for factors such as performance evaluation scores, tenure, and positions held. See Pls.' Reply at 23-24.

The Court agrees that the objective data from Wal-Mart eliminates the need for individualized hearings with respect to the qualifications component of the eligibility determination. As Plaintiffs assert, and Defendant does not dispute, Wal-Mart maintains an extraordinarily sophisticated information technology system. See Pls.' Opp. at 8; Bielby Decl. ¶ ¶ 16. A significant part of this system is Wal-Mart's " PeopleSoft" database, which is one of the most extensive data collection systems in the country. The database contains information on each employee individually with respect to job history, seniority, job review ratings, and many other factors, thereby enabling a sophisticated user to create detailed reports of individual work histories and qualifications. Id.; see also Drogin Decl. ¶ ¶ 6-8, 10 (describing PeopleSoft database as including (1) identifying information including name, employee ID, social security number, and gender, (2) complete job histories including all jobs held and at which stores, salary changes, promotions, transfers, etc., and (3) performance review information for all employees during class period); Cynn Decl. (discussing detailed job history and payroll data derived from Peoplesoft database for a sample of fifty employees). As such, it is plain that ample corporate data exists that would permit the Court to objectively determine which class members were minimally qualified for a given promotion. Accordingly, this aspect of the promotion backpay process is also manageable. ⁵⁰

50 The Court could instruct the parties to complete the task of deriving qualification-related information from the database, especially since the experts already have mined the database in constructing their analyses, or it may appoint a special master and/or expert. See Labor/Community Strategy Center v. Los Angeles County Metro. Transp. Auth., 263 F.3d 1041, 1049-50 (9th Cir.2001) (approving district court's appointment of special master with broad powers to resolve issues for purpose of enforcing court's remedial order).

With respect to the interest component of the eligibility determination, however, the Court faces a more complex situation. The objective data relied on by Plaintiffs-performance evaluation scores, tenure, and positions held-goes only to the issue of qualification, not interest in a position. As such, the Court finds that they are insufficient proxies for interest. Moreover, the objective data that would document interest-i.e. past applications for positions-is simply not available for many class members because, as discussed in section I.B.1.a.(2)(b), Wal-Mart did not utilize a system of posting and accepting applications for many positions during the class period. Indeed, without this data it is difficult to discern how the Court could identify which class members were interested in unposted promotions absent some type of individualized proceedings, whether it be by declaration or otherwise.

Plaintiffs respond by urging the Court to dispense with the " interest" factor altogether in this case. Specifically, Plaintiffs argue that individual interest in a promotion should be irrelevant to the remedy phase because at Wal-Mart expression of interest was not a requirement for promotion in the first place. Rather, Wal-Mart operated largely under a " tap on the shoulder" method, as discussed above. Given the wide-scale absence of application procedures, Plaintiffs conclude, eligible class members should simply consist of all qualified women in the relevant " feeder pools" (i.e. all qualified women in a job category and geographical location from which promotional candidates are drawn), all of whom are identifiable from the corporate records.

While expanding eligibility to include all qualified women in the relevant feeder pools, regardless of interest, would solve the manageability issue for unposted promotional opportunities, the Court is not convinced that such an approach is justified by either the facts or current law. First, the fact that an expression of interest is not a requirement for promotion does not mean that interest is irrelevant to determining eligibility. The fact that no expression of interest is required for promotion goes to the question of which employees were qualified for promotion. Thus, a female employee need not show she expressed interest in a promotion in order to establish that she was qualified for a promotion. This fact has no logical bearing, however, on the question of whether the employee was potentially victimized by the employer's discriminatory policies because the employee was interested in (but denied) a promotion.

Second, while courts commonly rely on feeder pools in making liability determinations,⁵¹ Plaintiffs have not pointed to any precedent that would permit this Court to rely on feeder pools to determine eligibility for a backpay award. On the contrary, Teamsters raises serious questions as to the propriety of this approach. There, the Court rejected the argument that the group of " deterred employees" eligible for make-whole relief should automatically include all employees in the " feeder pool" (i.e. all employees who were qualified to apply for a better " line-driving" job) without the need for any individualized showing that each employee was actually interested in a promotion. 431 U.S. at 368-369, 97 S.Ct. 1843. The Court made this holding even though the district court found that line-driver jobs are generally considered the most desirable driving job, and that applying for the job would have been a futile act:

51 See, e.g., Hemmings, 285 F.3d at 1185-86; Shidaker, 833 F.2d at 631; Stender, 803 F.Supp. at 326; Butler II, 1997 WL 605754.

While the scope and duration of the company's discriminatory policy can leave little doubt that the futility of seeking line-driver jobs was communicated to the company's minority employees, that in itself is insufficient. The known prospect of discriminatory rejection shows only that employees who wanted line-driving jobs may have been deterred from applying for them. It does not show which of the nonapplicants actually wanted such jobs .... [T]he desirability of the [line-driving jobs] is not so self-evident as to warrant a conclusion that all employees would prefer to be line drivers if given a free choice.

⁵²TeamstersSeeid.⁵³

Finally, Plaintiffs' attempt to factually distinguish Domingo is not persuasive. Plaintiffs argue that in Domingo, the employer utilized an informal application process, seeDomingo II, 727 F.2d at 1445, and therefore it was reasonable to require class members to demonstrate their earlier interest on a post hoc basis, since an objective manifestation of such interest-rejected applications-was available. The employer in Domingo, however, did not in fact advertise job openings or post vacancies. Domingo v. New England Fish Co. (Domingo I), 445 F.Supp. 421, 435 (W.D.Wash.1977). Rather, " [m]ost positions [were] filled by word-of-mouth recruitment." Id. In fact, the informal application procedures referred to in the Domingo appellate opinion appear to have been targeted toward non-class members through " [w]ord-of mouth recruitment, directed along racial lines, [which] made it especially difficult for [class members] to become aware of openings as they occurred." Domingo II, 727 F.2d at 1445. Significantly, despite the lack of job posting, Domingo still required some, albeit minimal, showing that the class member had desired a promotion before being eligible to participate in the backpay award to the class. Id. (noting for example that " any request for a better position" would suffice, and that " [a]ll uncertainties should be resolved against the employer" ).

The Court acknowledges that backpay is a significant aspect of any make-whole remedy that should be granted whenever possible. See, e.g.,Albemarle Paper Co. v. Moody, 422 U.S. 405, 421-422, 95 S.Ct. 2362, 45 L.Ed.2d 280 (1975). Nonetheless, the Court concludes that this aspect of the remedy can not be reasonably managed on a class basis in this particular case for promotions that were not posted, such that no objective applicant data exists. Plaintiffs have not persuaded the Court that it can properly identify which class members are eligible for backpay solely by reliance on feeder pools or other objective corporate data serving as a proxy for interest. As Plaintiffs have not proposed any other manageable alternative in this case, the Court declines to certify a claim for lost pay with respect to the portion of Plaintiffs' promotion claim where no objective applicant data exist. SeeWilliams v. Owens-Illinois, Inc., 665 F.2d 918, 929 (9th Cir.1982) (trial judge has discretion to limit issues in class action to " ‘ those parts of a lawsuit which lend themselves to convenient use of the class action motif’ " ) (citations omitted); Morgan, 169 F.R.D. at 358 (limiting certification to liability and request for injunctive relief); Fed.R.Civ.P. 23(c)(4).

Where, however, promotional opportunities were posted, Wal-Mart's advanced personnel system does contain objective applicant data documenting which class members were interested in each such promotion. In particular, two corporate electronic databases-known as Job Posting Data for hourly job vacancies, and Management Career Selection (MCS) for salaried vacancies-contain the identities of the individual applicants for posted positions. See Drogin Decl. ¶ ¶ 8, 44-45; Haworth Decl. ¶ ¶ 72, 78, 118. Thus, with respect to all posted positions, the Court finds that it can readily identify through objective data those class members who were both qualified and interested in promotional opportunities-and thus were potential victims of any proven discriminatory policy. Accordingly, for this subset of the class, the Court is confident that it could manageably determine which class members would be eligible to share in a formula-derived, lump sum backpay award to the class.⁵⁴

54 While the Court does not, and need not, at this time attempt to identify precisely which challenged positions were in fact posted, it notes that this subset of positions covers a sizeable raw number of promotions. For example, from 1998 to 2002, well over 3,000 Store Manager openings were posted and applicants were tracked in the database. See Haworth Decl. ¶ 120 (see Jobs 2 and 6 in chart entitled " Analysis of Selections from Management Job postings by Job" ); Drogin Decl. ¶ 46 & Table 21. Haworth Decl. ¶ ¶ 72, 118 (" Wal-Mart has systematically retained records of about 580,000 actual applications received for hourly promotions and about 56,000 electronic expressions of interest in promotion to salaried management jobs since 1997" ; additionally, the 2003 Assistant Manager Trainee postings generated over 40,000 applicants for promotion); Bielby Decl. ¶ ¶ 43-44.

The ability to identify the specific class members who are both qualified and interested would also permit the Court to calculate through a formula Defendant's total backpay liability to this subset of the class. As discussed above, most formulas involve (1) identifying the number of positions that would have gone to class members absent discrimination, (2) estimating the total earnings that would have accrued to class members in those positions, and (3) some method to account for mitigation. There are, of course, various ways to approach these steps, and the Court does not intend to limit any options at this juncture. With the data discussed above, however, the Court would be able to determine if the percentage of women selected for certain positions was commensurate with, or fell short of, the percentage of qualified women who actually applied for such positions. In the event the percentage selected falls short, this information could be used to identify the number of " shortfall" positions-i.e., the number of positions that would have gone to class members absent discrimination. This number of shortfall positions could then be used for the second step of the formula, which would involve estimating the total earnings that would have accrued to class members in those shortfall positions (based, for example, on the earnings of men in the shortfall positions). Finally, the Court would deduct an amount to account for mitigation.

In sum, the Court rejects Defendant's contention that a lost pay remedy on Plaintiffs' promotion claim is entirely unmanageable. First, the Court finds that this is the type of case in which a formula approach may be appropriate. Second, the Court concludes that a formula-based remedy would be manageable, however, only with respect to those positions for which objective applicant data is available to document class member interest. In such cases, the Court would have sufficient data to identify those class members who would be eligible to participate in a lump sum backpay award, and to fashion a formula to calculate such an award. Where no such objective data exists, however, manageability concerns preclude a lost pay remedy in this class action. Accordingly, any lost pay remedy on Plaintiffs' promotion claim would be limited to that subset of the class for whom objective applicant data exists. SeeWilliams, 665 F.2d at 929; Morgan, 169 F.R.D. at 358; Fed.R.Civ.P. 23(c)(4).

b. Equal Pay Claim

As indicated above, Plaintiffs' equal pay claim presents issues that are quite distinct from plaintiffs' promotion claim. Having considered these issues, the Court concludes that, should plaintiffs prevail on the merits, a lost pay remedy for the class would be manageable. As previously discussed, Plaintiffs would enter the remedy phase with a presumption in their favor that any unequal pay was due to Defendant's discriminatory policy and practice SeeTeamsters, 431 U.S. at 361, 97 S.Ct. 1843. In order to award lost pay, however, the Court still must be able to (1) identify which class members were actually, or at least potentially, victimized by the employer's discriminatory pay policies, and (2) calculate the specific amount of lost pay that each such person is owed. As discussed below, the Court is well satisfied that it could accomplish both tasks in a manageable fashion and without resort to a formula approach.

(1) Identification of victimized class members

With respect to the first task, the primary manageability obstacle posed by the promotion claim simply is not present here. As discussed above, courts are not generally willing to assume, for purposes of granting make-whole relief, that all employees are uniformly interested in every promotional opportunity, given the various considerations that go into such a decision. This necessitates some examination of the interest levels of individual class members, as discussed above in terms of identifying objective information from the promotion databases. In contrast, courts can safely assume that all employees uniformly desire equal pay for equal work. Indeed, equal pay for equal work is a benefit that automatically flows to all employees. Thus, there is simply no need to assess individual interest levels with respect to equal pay claims. Rather, virtually all of the potential victims of an employer's discriminatory pay policy-i.e. those class members who were paid less than male employees for comparable work-can be identified by reference to objective criteria such as pay rates and job positions held.

It is also clear that identifying these potential victims would be a manageable task in this particular case. As discussed above, Defendant's PeopleSoft database contains a wide range of personnel and payroll data enabling a sophisticated user to create detailed reports of individual work and earnings histories as well as comparisons between workers. In addition, Wal-Mart maintains computerized bi-weekly payroll information for each and every employee working in the United States. See Drogin Decl. ¶ 7 (noting that the payroll data is capable of being merged into the PeopleSoft database for the creation of earnings reports). Thus, the objective data necessary to identify potential equal pay victims is readily available in this case.

Moreover, there is sufficient objective data available to identify not only the potential victims but also the actual victims of any proven discriminatory pay policy-i.e. those class members who not only were paid less for comparable work but who also were at least equally qualified as their male counterparts. This requires a more sophisticated use of the data because determining qualifications requires an analysis of additional factors. Again, however, Wal-Mart's database contains the critical information necessary to perform such an analysis for each employee individually, including job history, seniority, job review ratings, weeks worked, full-time versus part-time status, regular-time versus over-time, and store location. See Drogin Decl. ¶ ¶ 67, 69, 71; Cynn Decl.

It would be premature of course for the Court to determine at this juncture precisely which factors should be relied upon to compare qualifications. Nor is it necessary to account for every possible factor that might be relevant to qualifications. As the case law repeatedly emphasizes, when calculating backpay awards " unrealistic exactitude is not required, and [ ] uncertainties in determining what an employee would have earned but for the discrimination should be resolved against the discriminating employer." Shipes, 987 F.2d at 317 (citations omitted); see alsoTeamsters, 431 U.S. at 372, 97 S.Ct. 1843; McKenzie, 684 F.2d at 77; EEOC v. Chicago Miniature Lamp Works, 640 F.Supp. 1291, 1298-99 (N.D.Ill.1986) (accepting lack of precision at remedy phase due to subjectivity in defendant's hiring method). The Court thus flatly rejects Defendant's argument that any comparison of male and female employees would be impossible because the available data does not include dozens of factors identified in a survey of Store Managers as being relevant to pay decisions. Oral Arg. Tr. at 105, 125-26.⁵⁵ Rather, the Court is satisfied that it will have access to sufficient data to meaningfully compare the qualifications of female and male employees doing comparable work for purposes of calculating lost pay. SeeNational Treasury Employees Union v. United States, 54 Fed. Cl. 791, 795, 800 (2002) (in computing backpay for class of 210,000 employees court relied on " data concerning the class member's agency, occupation, geographic location, and grade," along with the use of " a few simplifying rules and assumptions to allow for the computation[ ] to be completed with reasonable accuracy, and in a reasonable amount of time" ).

55 As noted previously, the Court has stricken references to Defendant's Store Manager survey in Dr. Haworth's declaration for failure to satisfy the standards of Fed.R.Evid. 702 and 703. See Order re Motions to Strike. Moreover, even this survey indicates that the most important factors identified by the Store Managers declarations actually are in Defendant's database, e.g., " minimum pay established for the job classification by Wal-Mart's pay guidelines," and performance. See Haworth Decl., Appendix Vol. 2, Tab 16 (chart entitled " Store Manager Declarations: Percentage of Decisions in Which Factors Play a Role in Determining Starting Hourly Pay" ); see also Drogin Reply Decl., Ex. 2 (Wal-Mart " Field Associate Compensation Guidelines" ). While the data does not capture pre-Wal-Mart experience, the Court concludes that this one factor does not prevent it from making sufficient comparisons between class members and male employees for purposes of calculating lost pay. Segar, 738 F.2d at 1274-76 (affirming trial court's finding that plaintiffs' evidence was sufficient for backpay award despite failure to account for aspects of prior experience, especially where the past experience at issue was not used by defendant as a minimum objective qualification, and where it was defined subjectively and amorphously). Other factors noted by defense counsel at oral argument, such as willingness to work weekends or rotating shifts, or live cricket experience, are not significant, even according to Defendant's own data. See Haworth Decl. ¶ 187.

(2) Calculation of individual backpay awards

Once the actual victims of a proven discriminatory pay policy have been identified, Defendant's PeopleSoft data system also can facilitate the calculation of individualized damage awards. Since the database contains payroll information, it could be used to calculate the differential between each individual victim and the comparable male pay rate. Again, the Court need not decide at this time exactly how such calculations would be made. Rather, suffice it to say that the Court is confident that a fair and manageable method can be devised. SeeAlbemarle Paper Co., 422 U.S. at 419, 421, 95 S.Ct. 2362 (Congress' purpose in vesting discretionary powers in the courts to provide relief under Title VII was " to make possible the ‘ fashion[ing][of] the most complete relief possible’ " ) (citation omitted).

The Court's approach here-with respect to both the identification and calculation issues-is similar to that taken by the Fifth Circuit in Shipes, 987 F.2d 311. In Shipes, the plaintiff class of African-American workers alleged, similar to this case, that the employer used " entirely subjective personnel processes" to set discriminatory pay rates. Id. at 316. After the plaintiffs proved class-wide liability, the employer urged the Court to employ a Domingo -type formula approach to awarding backpay. The district court declined and instead appointed an expert to determine the amount of backpay, if any, due each class member. The expert developed a model for assessing backpay by comparing the starting pay of class members and white employees with similar qualifications, and then awarding the differential amount to those class members who suffered an actual loss. Id. at 319. The district court's individualized procedure was affirmed on appeal. Id. at 315-19 (" The fact that a class is large does not mean that pro rata [formula] relief should automatically be ordered.... [T]here should be, if possible, a determination on an individual basis as to which class members are entitled to damages and the amount of such recovery." ).

Notably, Shipes makes no mention of any need for Teamsters hearings or any kind of employer participation in the process of determining which employees are entitled to backpay, instead relying exclusively on corporate records. Shipes, 987 F.2d at 317-19. This is consistent with the fact that Teamsters was not an equal pay (nor a formula) case, but a hiring/promotion case that required determinations that could not be capably captured in that case by objective data. Indeed, where, as here, the actual victims of any discriminatory pay policy, and the amount of lost pay due, can be determined based on objective data, a remedy-phase Teamsters mini-hearing is simply unnecessary. See alsoLiberles v. County of Cook, 709 F.2d 1122, 1127, 1136 (7th Cir.1983) (where backpay is awarded for salary discrimination alone, and where evidence of pay rates and comparative qualifications exists, " individualized hearings are unnecessary" ).

In sum, the Court is satisfied that the actual victims of a proven discriminatory pay policy can be sufficiently identified and compensated through reference to corporate records, and that this can be done on an individualized basis without resort to a Domingo -type formula and without the need for Teamsters hearings.⁵⁶ Accordingly, the Court rejects Defendant's contention that a manageable remedial process for awarding backpay as compensation for equal pay violations cannot be fashioned in this case.⁵⁷

56 While Plaintiffs in their briefing assumed the use of a formula to calculate backpay for both promotion and equal pay claims, at oral argument Plaintiffs' counsel was " fully supportive" of the approach taken herein. Oral Arg. Tr. at 70.

57 Defendant notes that it took the Shipes expert several years to complete the process for a much smaller class, and suggests that such a process would therefore be unmanageable in a case of this scope. See Shipes, 987 F.2d at 315. While the reason for the length of time in Shipes is not apparent, this Court is confident that advances in technology since the Shipes decision in 1993, combined with the sophistication of Defendant's database, will allow for a backpay award process to proceed at a manageable pace in this case.

c. The 1991 Civil Rights Act

In a final attempt to demonstrate the unmanageability of Plaintiffs' promotion and equal pay claims, Defendant invokes the Civil Rights Act of 1991, Pub.L. No. 108-198, 105 Stat. 1071 (1991); 42 U.S.C. § 2000e. Specifically, Defendant argues that this Act entitles it to present a " mixed-motive" defense with respect to each class member, which would necessitate individualized hearings. See 42 U.S.C. § 2000e-5(g)(2)(B). While neither party has cited cases applying the " mixed motive" defense in a class action context, Defendant's argument completely lacks merit since, in any event, Plaintiffs have elected " to pursue their claims solely under [a 42 U.S.C.] § 2000e-2(a) [single-motive theory]." Plaintiffs' Reply at 25 (emphasis in original).

Title VII makes it an " unlawful employment practice for an employer ... to discriminate against any individual ... because of such individual's race, color, religion, sex, or national origin." 42 U.S.C. § 2000e-2(a)(1) (emphasis added). Courts commonly view the causation issue in employment discrimination cases by separating them into two categories. In " single-motive" cases, causation is seen as a binary question where the true basis for the adverse employment action is either legal or illegal. SeeCosta v. Desert Palace, Inc. (Costa II), 299 F.3d 838, 856 (9th Cir.2002), aff'd, 539 U.S. 90, 123 S.Ct. 2148, 156 L.Ed.2d 84 (2003). In " mixed-motive" cases, the factfinder concludes that the adverse action was the result of a combination of both legal and illegal factors. Id.

In Price Waterhouse, the Supreme Court held that a finding of mixed-motives would provide a complete defense. 490 U.S. at 244, 261, 109 S.Ct. 1775. In response, Congress amended Title VII in 1991, allowing plaintiffs to prevail in a mixed motive case. See 42 U.S.C. § 2000e-2(m). Congress further provided, however, that in such cases, the employer can limit the plaintiff's remedies to declaratory relief, certain types of injunctive relief, and attorney's fees and costs, if the employer successfully demonstrates that it would have taken the same action in the absence of the impermissible motivating factor. 42 U.S.C. § 2000e-5(g)(2)(B). This defense is generally referred to as the " same decision affirmative defense." See, e.g.,Costa II, 299 F.3d at 857.

Nothing in the " mixed-motive" provisions of the 1991 Civil Rights Act-which were enacted to strengthen and expand employee rights under Title VII ⁵⁸-or the governing case law, create any manageability concerns in this case. First, in practical terms, the 1991 amendments do not affect the presentation of evidence at the liability phase. As the Ninth Circuit explained in Costa II:

58 " The Civil Rights Act of 1991 ... was enacted to restore civil rights limited by then-recent Supreme Court decisions and to ‘ strengthen existing protections and remedies available under federal civil rights laws to provide more effective deterrence and adequate compensation for victims of discrimination.’ " E.E.O.C. v. Luce, Forward, Hamilton & Scripps (" Luce" ), 345 F.3d 742, 747 (9th Cir.2003), quoting H.R.Rep. No. 102-40(II), at 1 (1991), reprinted in 1991 U.S.S.C.A.N. 549; Duffield v. Robertson Stephens & Co., 144 F.3d 1182, 1192 (9th Cir.1998), overruled on other grounds by Luce, 345 F.3d 742 (" The purpose of the Act was uniformly to expand employees' rights and ‘ to increase the possible remedies available to civil rights plaintiffs.’ " ) (citation omitted) (emphasis in original).

[W]e emphasize that there are not two fundamentally different types of Title VII cases. In some cases, the employer may be entitled to the " same decision" affirmative defense instruction. In others, it may not. The employee's ultimate burden of proof in all cases remains the same: to show by a preponderance of the evidence that the challenged employment decision was " because of" discrimination.

Costa II, 299 F.3d at 857. Thus, at trial both parties will present evidence on the issue of whether any disparities between male and female employees with respect to pay and promotion are the result of a pattern and practice of intentional gender discrimination or legitimate factors. As discussed above, the liability phase of the case focuses on the class as a whole, not individualized employment decisions. Thus, the " same decision" defense with respect to individual class members will play no role in the liability phase of this case.

If a jury finds, in the liability phase, that Plaintiffs have proven a pattern of intentional gender discrimination, and that Wal-Mart's evidence has failed to show that one or more legitimate motives accounted in part for that discrimination, then Plaintiffs will prevail on their single motive theory. In that event, the Act would provide no basis for the assertion of a " same decision" defense at the remedy stage since, by the explicit statutory language, such defense applies only where the illegal conduct was the result of mixed-motives. See 42 U.S.C.2000e-5(g)(2)(B) (limiting " same decision" defense to " claim[s] in which an individual proves a violation under section 2000e-2(m) [the mixed-motive provision]" ); Desert Palace, Inc. v. Costa (Costa I), 539 U.S. 90, 123 S.Ct. 2148, 156 L.Ed.2d 84 (2003).

If the jury finds that the challenged conduct results either from entirely legitimate factors or a mix of legitimate factors and illegal discrimination, then the claims for lost pay and punitive damages will conclude in Defendant's favor, given Plaintiffs' election to forego seeking the full panoply of remedies that would require adjudication of a " same-decision" defense. Thus, Plaintiffs are not attempting to unilaterally manipulate the trial proceedings, as Defendant suggests; rather, they simply have elected to forego a certain potential remedies to which they might otherwise be entitled. Given all of the above, it is clear that the " same decision" defense-which only pertains to the scope of the remedy-will not be an issue in this case regardless of whether Plaintiffs or Defendant prevails at the liability phase. Accordingly, the Court rejects Defendant's argument that the 1991 Civil Rights Act prevents certification of a class in this case.

CONCLUSION

For all of the foregoing reasons, this Court finds that Plaintiffs' motion for class certification should be granted in part, and denied in part, consistent with this decision and as set forth below. This Court retains, of course, the discretion to revisit this certification should unexpected circumstances arise. As the Ninth Circuit has explained, " [b]ecause class actions vary so widely in their circumstances, the trial judge is vested with broad discretionary control over the conduct of such actions enabling the presiding judge to respond fluidly to the varying needs of particular cases." Officers for Justice v. Civil Service Comm'n, 688 F.2d 615, 633 (9th Cir.1982). Indeed, district courts have the inherent discretion to review class certification decisions at any time. Armstrong, 275 F.3d at 872, n. 28; Barefield, 1988 WL 188433 (N.D.Cal.1988) (class certification orders are " inherently tentative and may be modified as circumstances require" ).

Based on the current record, however, the Court is well satisfied that Plaintiffs have met their burden of demonstrating that each of the Fed. R. Civ. Proc. 23(a) requirements are satisfied, as follows:

1. Numerosity: This factor is not contested, and the Court finds joinder would be impracticable in this case.

2. Commonality: Plaintiffs have satisfied their burden of showing the existence of significant legal and factual issues concerning Defendant's alleged company-wide discriminatory pay and promotion policies that are common to the entire class. In particular, Plaintiffs have provided:

(a) evidence of common company policies and practices, including subjective decision-making, a culture of corporate uniformity, and gender stereotyping; (b) expert opinion establishing statistical evidence of gender disparities which support an inference of company-wide discrimination in both pay and promotions; and

(c) anecdotal evidence from class members of discriminatory attitudes held or tolerated by management.

3. Typicality: Plaintiffs have satisfactorily shown that the claims of the named plaintiffs are reasonably co-extensive with those of the class as a whole.

4. Adequacy: Plaintiffs have satisfactorily shown that the interests of the class representatives do not conflict with those of the class members. Class counsel are qualified to conduct this litigation.

Plaintiffs also have met their burden of demonstrating that the requirements of Fed. R. Civ. Proc. 23(b)(2) are satisfied here in that Defendant " has acted or refused to act on grounds generally applicable to the class, thereby making appropriate final injunctive relief or corresponding declaratory relief with respect to the class as a whole." As discussed above, however, the class shall be afforded notice and an opportunity to opt-out in order to satisfy due process considerations.

The Court further concludes that Plaintiffs' claim for equal pay is manageable with respect to issues of liability and all forms of requested relief. Finally, the Court finds that Plaintiffs' promotion claim is manageable with respect to issues of liability and injunctive and declaratory relief. The Court finds, however, that with respect to the remedy of lost pay, it is manageable only with respect to those challenged promotions where objective data is available to document class member interest in the respective challenged promotion. Thus, the Court denies certification, on grounds of unmanageability, with respect to Plaintiffs' promotion claims for lost pay (and hence punitive damages) as to those class members for whom no such data is available.

Accordingly, and good cause appearing, it is HEREBY ORDERED as follows:

1. Plaintiffs' Motion for Class Certification is GRANTED IN PART and DENIED IN PART consistent with this decision.

2. The Court certifies the following class for purposes of liability, injunctive and declaratory relief, punitive damages,⁵⁹ and lost pay, except that class members for whom there is no available objective data documenting their interest in challenged promotions shall be limited to injunctive and declaratory relief with respect to plaintiffs' promotion claim:

59 The Court emphasizes that in the event Plaintiffs are able to establish liability, punitive damages would not flow automatically from such a finding. Rather, only those class members who establish that they are entitled to an award of lost pay, and thus demonstrate individual harm from Defendant's conduct, would be eligible to receive such damages. See section II.A. The Court further notes that the fact that this subset of the class is not identifiable now does not preclude class notice at this juncture. Class action notices are typically sent to all class members who may be entitled to recover, not to those whose right to recover has already been established.

All women employed at any Wal-Mart domestic retail store at any time since December 26, 1998, who have been or may be subjected to Wal-Mart's challenged pay and management track promotions policies and practices.

3. This action shall be bifurcated into separate liability and remedy phases.

4. The individual racial discrimination claims asserted by named plaintiffs Betty Dukes and Cleo Page shall be severed from the class case.

5. The parties shall appear on Wednesday, July 28, 2004 at 2:00 p.m. for a status conference to discuss the issue of notice to the class, and other scheduling and case management issues. A Joint Status Statement shall be filed no later than 10 days in advance of the status conference.

IT IS SO ORDERED.

Since the parties rely mostly on their regression analyses in this case, the Court is not called upon to decide whether plaintiffs' descriptive statistics alone suffice to support a finding of commonality or a prima facie case. See Reference Manual on Scientific Evidence, p. 148-49 n. 224 (2d ed. 2000) (" The conditions under which a simple disparity between two groups amounts to a prima facie case that shifts the burden of proof to the defendant in Title VII and other discrimination cases have yet to be articulated clearly and comprehensively." ) (citations omitted).

Although Defendant has not made an issue of the number of regions lacking a statistically significant gender differential at the higher level in-store managerial jobs, the Court became concerned that this evidence raises the question of whether a nation-wide class should be narrowed to lower level in-store managers (Support Managers and Management Trainees) with respect to promotions. The Court raised this issue with counsel at oral argument, and Plaintiffs responded by arguing that the wide-spread discrimination at the lower levels carries through to the upper levels, especially where most promotions are made from within. See Oral Arg. Tr. at 48:23-49:11. As Plaintiffs also stated in their briefs: " Obviously, as women have been under-promoted to the first-level management positions, their representation in the feeder pools for store manager is commensurately diminished." Pls.' Mot. at 29.

The Court is persuaded, at least for purposes of class certification, that the evidence is sufficient to support an inference that the statistically significant disparities identified at the lower levels affect the experience of women seeking promotion at the higher levels. This evidence consists of: (1) the sizable number of regions with statistically significant disparities, (2) descriptive statistics showing a steep drop in female presence the higher one goes in the in-store managerial hierarchy ( see section I.B.1.a.(1); Drogin Decl. ¶ 23, Table 7), (3) the significantly longer time it takes women to achieve promotions than men generally (Drogin Decl. ¶ 29); (4) Defendant's practices affecting promotions-such as broad subjectivity, wide-scale lack of posting, and relocation policies-as discussed above ( see sections I.B.1.a.(3) & I.B.1.b.(1)), and (5) anecdotal evidence of discrimination ( see section I.B.3).

The Court's determination is further supported by the case law in this area. As the court explained in ruling on the merits in Stender:

The statistical evidence is significant for some positions, but not for others. However, the court finds that the lack of statistically significant disparities in promotions of women to [upper-level jobs] is caused by women being blocked from upper management positions at the lower rungs of the promotional ladder. In addition, the court has considered direct and indirect evidence of discriminatory intent on the part of Lucky managers. In considering plaintiffs' evidence cumulatively, the court concludes that Lucky's promotion process is tainted by discretionary decision making and the use of variable and subjective criteria. Accordingly, the court finds that the evidence is sufficient to sustain plaintiffs' burden of proving disparate treatment in Lucky's promotion practices as a whole.

Stender, 803 F.Supp. at 333; see alsoMcClain v. Lufkin Indus., Inc. 187 F.R.D. 267, 274 (E.D.Tex.1999) (" Discrimination in hiring and advancement, even at the lower levels of the hierarchy, has a rippling effect upwards through that hierarchy." ) (citation omitted); Carpenter v. Stephen F. Austin State Univ., 706 F.2d 608 (5th Cir.1983).

In Butler I, 1996 WL 421436, the court granted class certification, finding that plaintiffs' statistical evidence created a common issue of law and fact. In a subsequent order denying defendant's motion for summary judgment, the court described plaintiffs' evidence as including expert testimony by Drs. Drogin and Bendick in which feeder pool analyses were used to measure gender disparities in promotion. Butler v. Home Depot, Inc., (Butler II), 1997 WL 605754, at *8 (N.D.Cal.1997). The court noted that Home Depot primarily promotes from within, and the court approved the use of the feeder pools, finding that they were of a kind and degree from which causation may reasonably be inferred. Id. See alsoPitre v. Western Elec. Co., 843 F.2d 1262 (10th Cir.1988) (approving reliance on feeder pools); Shidaker v. Tisch, 833 F.2d 627, 631 (7th Cir.1986) (same); Paxton v. Union National Bank, 688 F.2d 552, 563-65 (8th Cir.1982) (same); Smith v. Xerox Corp., 196 F.3d 358, 370-71 (2d Cir.1999) (approving feeder pool methodology but requiring very careful construction of pools that compare co-workers who competed directly against each other to receive a benefit).

In another case relied upon by Defendant, United States v. County of Fairfax, Va., 629 F.2d 932, 940 (4th Cir.1980), a decision regarding the merits of a discrimination action, the court held that actual applicant flow data could be used as a basis for extrapolation for years in which such data had been destroyed. However, in that case the court determined that the only alternative means of estimating applicant flow would be to use a zip code analysis (which is based on census information), and that the flaws in using that data outweighed the flaws inherent in the extrapolation method. Id. Indeed, the courts generally are highly suspect of census data for workplace comparison since applicant pools often are far narrower than the general population. Here, in contrast, Dr. Drogin's availability pools are drawn from internal data, and therefore do not suffer from the same problem as census data.

Finally, defendant again analogizes to Donaldson, 205 F.R.D. 558, but again the analogy fails. In Donaldson, the court found that " plaintiffs' own statistics belie the existence of [ ] a pattern [of discrimination]." Id. Indeed, the Donaldson plaintiffs' expert's initial study showed no pattern of discrimination in many significant areas. Id. at 567. Plaintiffs presented the court with an entirely new set of statistics in their reply materials, but they were based on the same raw data with different calculations. Id. Given these circumstances, it is no surprise that the court found that plaintiffs' statistical analysis deserved no weight. Id.

In a similar vein, Defendant argues that Dr. Bendick excluded approximately 96 percent of the workforce holding management jobs in the retail sales field. However, in the four percent he did include there are over 170,000 managers. Moreover, these are the most applicable comparators in that they are managers of large retail firms. In other words, Dr. Bendick's selection is consistent with the reasonable approach of comparing Wal-Mart to " other large chain[s] of large stores selling a large range of goods." Bendick Rebuttal Decl. ¶ 14.

In a formula case, however, the employer's total backpay liability is set as a lump sum. As such, the employer does not appear to have any significant, legitimate interest in which class members share in the formula award and which do not. Indeed, Defendant has failed to clearly articulate such an interest here. This general point was acknowledged by the Ninth Circuit in Hilao v. Estate of Marcos, 103 F.3d 767 (9th Cir.1996), which supports the general principle that a defendant does not necessarily have a protectable interest in contesting the validity of individual claims in a formula case, assuming the court has confidence that the formula reliably and accurately captures defendant's total liability to the class. Id. at 786. See alsoPettway V, 681 F.2d at 1266 (defendant had no basis for objecting to a formula approach on due process grounds so long as the " amount found due the class does not exceed the amount which all members of the class together would have been entitled to receive under a correct hypothesis, which we must assume the trial court would adopt" ). The Court also notes that neither O & G Spring, 38 F.3d 872, nor Hameed, 637 F.2d 506, indicate that the employer participated in eligibility determinations.