OPINION AND ORDER
LYNCH, District Judge.Female employees of Novartis Pharmaceuticals Corporation (“NPC”) bring this gender discrimination suit under Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq. (“Title VII”). Defendant Novartis Corporation (“Corporation”), the corporate parent of NPC, moves for summary judgment, arguing that plaintiffs have failed to show that Corporation is subject to liability for its subsidiary’s actions. Plaintiffs also move for class certification pursuant to Rule 28 of the Federal Rules of Civil Procedure. Both motions will be granted.
BACKGROUND
Plaintiffs are nineteen women presently or formerly employed by NPC in sales-related positions.1 They claim that NPC discriminates against them in various ways, including in compensation, promotion and promotional opportunities, personnel evaluations, and by adverse treatment of women who take pregnancy leave. They seek injunctive relief, back pay and front pay, and compensatory and punitive damages.
NPC is a pharmaceutical company with about 6,000 sales representatives, headquartered in East Hanover, New Jersey, with operations in all 50 states. The corporation’s organizational structure divides the country into “territories,” each of which has one or more sales representatives responsible for marketing NPC’s products to local doctors. NPC also has “national field forces” of employees focusing on areas such as “mass markets,” institutional markets, and specialty physicians. NPC is wholly owned by defendant Corporation, a holding company with three employees, through another company, Novartis Financial. As discussed below, the parties dispute the extent and nature of the relationship between Corporation and NPC.
Defendant Thomas Ebeling is Chief Executive Officer (“CEO”) of Novartis Pharma AG, a related company that is not a party to this case. Ebeling is also a boardmember of NPC, and is alleged to have been actively involved with the management of NPC. (Compl. II44.)
After extensive discovery, plaintiffs move for certification of a class consisting of
[a]ll women who are currently holding, or have held, a sales-related job position with [NPC] during the time period July 15, 2002 through the present, including those who have held positions as Sales Representatives, Sales Consultants, Senior Sales Consultants, Executive Sales Consultants, Sales Associates, Sales Specialists, Senior Sales Specialists, and District Managers I.
(P. Class Cert. Mem. 1.) In support of the motion for class certification, plaintiffs offer evidence that includes the declarations of 87 women who are or were employees of NPC, as well as two expert reports.
Defendants oppose the request for class certification, primarily on the grounds that plaintiffs’ statistical and anecdotal evidence fails to show the existence of common questions of fact and law. Defendant Corporation moves for summary judgment on the grounds that plaintiffs have failed to show it that its operations are sufficiently integrated with NPC’s operations to give rise to liability for any acts of discrimination.
DISCUSSION
I. Novartis Corporation’s Motion for Summary Judgment
A. Summary Judgment Standard
Summary judgment is warranted where “the pleadings, depositions, answers to interrogatories and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to a judgment as a matter of law.” Fed.R.Civ.P. 56(c). The moving party has the burden to *250establish the absence of any material factual issues. Terry v. Ashcroft, 336 F.3d 128, 137 (2d Cir.2003), citing Celotex Corp. v. Catrett, 477 U.S. 317, 325, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986). “In determining whether there are genuine issues of material fact,” the Court must “resolve all ambiguities and draw all permissible factual inferences” in favor of the non-moving party. Id. (citation and internal quotation marks omitted).
B. Standards for Parent Company Liability Under Title VII
A parent company can be held liable for its subsidiary’s violations of Title VII under the “single or joint employer” test developed by the National Labor Relations Board and adopted by the Second Circuit in the Title VII context. Gulino v. N.Y.S. Educ. Dep’t, 460 F.3d 361, 378 (2d Cir.2006). Under this test,
[A] parent and subsidiary cannot be found to represent a single, integrated enterprise in the absence of evidence of (1) interrelation of operations, (2) centralized control of labor relations, (3) common management, and (4) common ownership or financial control.
Cook v. Arrowsmith Shelburne, Inc., 69 F.3d 1235, 1240 (2d Cir.1995).
The four-factor test may be satisfied “by a showing that there is an amount of participation that is sufficient and necessary to the total employment process, even absent total control or ultimate authority over hiring decisions.” Id. at 1241 (internal citations, alterations and quotation marks omitted). “We focus our inquiry ... on the second factor, centralized control of labor relations,” id. at 1241, a “crucial element of the inquiry.” Parker v. Columbia Pictures Indus., 204 F.3d 326, 341 (2d Cir.2000) (internal citations and quotation marks omitted). Because centralized control of labor relations is the focus of the analysis, it will be discussed first.
C. Centralized Control of Labor Relations
The most important element in the four-factor test is “whether the two enterprises exhibit centralized control of labor relations, including tasks such as handling job applications, approving personnel status reports, and exercising veto power over major employment decisions.” Parker, 204 F.3d at 341. “This particular criterion has been distilled to a critical question: what entity made the final decision regarding employment matters related to the person claiming discrimination?” Regan v. In the Heat of the Nite, Inc., 93 Civ. 862, 1995 WL 413249, at *3 (S.D.N.Y. July 12, 1995).
Corporation relies heavily on this Court’s decision in Salemi v. Boccador, Inc, No. 02 Civ. 06648, 2004 WL 943869 (S.D.N.Y. Apr. 29, 2004). In that case, the Court held that there was a genuine issue of material fact as to whether a corporation and its parent company should be treated as an integrated employer for Title VII purposes, even though the level of control exercised by the parent company was “less than that described in many of the cases that have allowed integration under Title VII.” Id. at *5. Acknowledging that centralized control of labor relations was the “key factor” in the inquiry, id. at *4, the Court analyzed that factor in terms of the specific allegations in the case, which concerned the conditions of the plaintiffs employment, not failure to promote or wrongful termination. Id. at *5. The Court held that
the relevant employment decisions for purposes of integrating [the two companies] are not simply the hiring or firing of plaintiff, but rather are those decisions that construct the conditions of employment for employees at the plaintiffs level, including not only the hiring and firing of employees at [her] level, but also the hiring and firing of the manager [of the subsidiary company] and the setting of overall policies for employee conduct and discipline.
Id. In this case, the relevant employment decisions for the most part also concern conditions of employment. Plaintiffs’ allegations include discriminatory promotion, disparate pay, differential treatment, hostile work environment, discrimination against women who take pregnancy leave, sexual harassment, and retaliation. (Compl.H 3.)
Nothing about the allegations concerning Corporation, however, suggests that *251it has any control over the personnel policies of NPC, much less the conditions of employment experienced by NPC employees. Unlike the situation in Salemi, where the officer who served as president of both corporations “controlled many of [the subsidiary’s] employment policies, even those as quotidian as dress code,” and “exercised this authority on behalf of [the parent corporation],” 2004 WL 943869 at *6, plaintiffs have pointed to no evidence that any Corporation officer has any control whatsoever over the conditions of employment at NPC. Uncontradicted testimony in the record indicates that Corporation never tells NPC’s HR personnel what policies to adopt or how to respond to an employee complaint, and never communicates with them to discuss the hiring or firing of NPC’s employees. (Reply Ex. 2 at 299.)
Plaintiffs contend that there is a “blending” of human resources functions between Corporation and NPC. They note that James Robinson, Vice President of Corporation’s HR department, is also Vice President of HR for two related companies, Novartis Services and Novartis Financial, and contend that Ronbinson “has also been responsible for the HR aspect of [NPC’s] acquisitions of sales personnel.” To support this claim, they note that on one occasion when Corporation planned to acquire certain operations from another company and merge those operations into NPC, Robinson performed due diligence on the personnel who would be transferred in support of those operations. (P.Ex. E.l at 52-53, 55-56.) No reasonable factfinder could find, however, that this due diligence was evidence of centralized control of labor relations.
In connection with the acquisition, Robinson wrote a report on the “HR piece of the due diligence,” but did not share it with anyone at NPC’s HR department. {Id. at 56, 60.) Although this service may have in some indirect sense involved personnel management at NPC (in that that was the subject of the report), nothing suggests that Robinson exercised any control over labor relations at NPC. The writing of the report was not a joint HR function, but a financial analysis of the prospects of a proposed acquisition by the holding company. Thus, Robinson’s analysis of the proposed deal is not evidence that Corporation exercised any control over the conditions of employment at NPC.
Other evidence submitted by plaintiffs is even weaker. Plaintiffs note that Corporation and NPC have submitted joint applications for Fortune Magazine’s “Best Places to Work” survey. Plaintiffs acknowledge, however, the rules of the contest required that corporations apply together with all affiliates, rather than individually.2 (P. Mem.3 11-12.) The only inference that can be drawn from the joint submission, therefore, is that the various Novartis entities wished to be known as good places to work. The law of corporate liability is not governed by the rules of a magazine contest.
Plaintiffs also rely on the Novartis companies’ use of Lisa DiPaolo, an NPC administrative employee. DiPaolo works for NPC and does separate work for Corporation, for which Corporation reimburses NPC. (Reply Ex. 4 at 135.) Plaintiffs do not contend that DiPaolo’s services for Robinson at Corporation pertain in any way to NPC’s personnel policies or practices. The fact that the NPC employee is paid for HR work at Corporation, and the fact that the employee’s work at Corporation is limited to Corporation’s HR needs, is evidence of separation of HR functions between the two corporations. Moreover, the fact that a single NPC employee is used by Corporation says nothing about whether Corporation controls conditions of employment at NPC.
*252Plaintiffs also note that Corporation’s Head of HR participates in HR-related meetings at NPC. The evidence suggests that Corporation sometimes adopts policies formulated by NPC. (D. Ex. E.3 at 229-34.) It is not surprising that Corporation, which has three employees, would use policies adopted by NPC, a much larger affiliated corporation, as a model. But nothing in the record indicates that it is in any way obligated to do so, or that Corporation has the ability to dictate the terms of any policies to NPC.
Finally, plaintiffs also argue that the two corporations share employee benefit plans. However, “the fact that the companies maintained the same benefits does not suggest centralized control of labor relations.” Balut v. Loral Elec. Sys., 988 F.Supp. 339, 347 (S.D.N.Y.1997). Rather, a “common benefits package speaks only to economies of scale ... and not to centralized control of labor relations.” Kellett v. Glaxo Enter., 91 Civ. 6237, 1994 WL 669975, at *5 (S.D.N.Y. Nov.30,1994).
Plaintiffs have failed, in short, to offer any meaningful evidence of centralized control of labor relations. There is no evidence that Corporation and NPC “appear to make joint hiring and firing decisions.” Lihli Fashions Corp. v. N.L.R.B., 80 F.3d 743, 747 (2d Cir. 1996). Nor is there evidence from which a reasonable factfinder could conclude that Corporation exerted control over the conditions of employment at NPC. As in a similar ease,
[i]t is uneontroverted that [the two companies] have separate human resources departments, and that [the subsidiary] establishes its own policies and makes its own decisions as to the hiring, discipline, and termination of its employees. It is likewise undisputed that plaintiffs worked for [the subsidiary] and that each was supervised by another [subsidiary] employee. It follows from these undisputed facts that [the subsidiary], and not [the parent], made the ‘final decisions’ regarding plaintiffs’ employment that are at issue here.
Duffy v. Drake Beam Morin, No. 96 Civ. 5606, 1998 WL 252063, at *4 (S.D.N.Y. May 19, 1998). Therefore, there is no genuine issue of material fact as to this critical prong of the inquiry.
D. Interrelation of Operations
When considering the “interrelation of operations” prong of the integration analysis, courts in this district have considered factors including:
(1) whether the parent was involved directly in the subsidiary’s daily decisions relating to production, distribution, marketing, and advertising; (2) whether the two entities shared employees, services, records, and equipment; (3) whether the entities commingled bank accounts, accounts receivable, inventories, and credit lines; (4) whether the parent maintained the subsidiary’s books; (5) whether the parent issued the subsidiary’s paychecks; and (6) whether the parent prepared and filed the subsidiary’s tax returns.
Herman v. Blockbuster Entm’t Group, 18 F.Supp.2d 304, 309 (S.D.N.Y.1998), aff'd 182 F.3d 899 (2d Cir.1999). Plaintiffs offer no evidence pertaining to the first, third, or fifth of these factors.
The evidence on which plaintiffs rely to show interrelation of operations is weak. There is no evidence that Corporation “establish[es] the operating practices and management practices” at NPC. Cook, 69 F.3d at 1241. Rather, plaintiffs’ evidence primarily establishes that Corporation, which is very small, shares some services with and purchases othei’s from NPC for reasons of efficiency.
Corporation and NPC share office space in Florham Park, New Jersey. A reasonable factfinder could conclude, based on testimony introduced by plaintiffs (P.Ex. C), that the office space is not clearly marked as separate, although there is contradictory evidence in the record. (D.Ex. 6.) Similarly, plaintiffs note that Corporation and NPC employees have access to the same phone directory and intranet services (P. Mem. 16), and that NPC provides IT services to Corporation. (P. Mem. 17.) Although these facts, if true, would indicate that Corporation and NPC are closely-affiliated companies, they do not indicate any interrelationship that is relevant *253to control over conditions of employment at NPC.
A lengthy list of other asserted connections offered by plaintiffs includes little of relevance. Plaintiffs contend that Corporation pays NPC to store its records at NPC’s warehouse, but offer no evidence that this was anything other than an arm’s-length transaction. Plaintiffs’ claim that all of the relevant Novartis companies use Pricewaterhousecoopers as external auditors is even less compelling. Any number of companies use that auditor’s services without being affiliated in any way. Plaintiffs’ contention that Novartis Services decides which NPC drugs need patent protection (P. Mem. 17) is irrelevant to whether Corporation exercises any control over NPC’s personnel functions. Plaintiffs also claim that Corporation’s deputy general counsel keeps custody of NPC’s minute books. (P. Mem. 17.) This fact establishes that the parent company is keeping track of the operations of its subsidiary, but not that it exercises any control over NPC’s operations. See Balut, 988 F.Supp. at 345-346 (“The fact that [the parent] Loral reviewed [the subsidiary’s] operations biannually, however, does not demonstrate an interrelationship, because a parent typically reviews a subsidiary’s progress on a periodic basis”). Plaintiffs note that Corporation provides some capital financing to NPC, but this is typical of a relationship between parents and subsidiaries. Nor does the allegation that Corporation has the power to approve or disapprove NPC’s budget suggest an interrelationship of operations; rather, this is a normal corollary of ownership.
Plaintiffs also rely on what they call a “lack of arm’s-length dealings” between the two corporations. (P. Mem. 9.) They note that Corporation does not have a written lease for the office space it rents from NPC, although the annual rent is over one million dollars. This is evidence of a close relationship, but it has nothing to do with labor relations. Nor does the informality of the arrangement raise a question about whether it was agreed to at arm’s length.
Plaintiffs claim that Corporation decides how much Novartis Services will bill NPC for audit, legal, and intellectual property-related services it performs for NPC (P. Mem. 9), but this allegation concerns the relationship between Corporation and Novartis Services, not Corporation and NPC. Similarly, plaintiffs’ claim that loans from Novartis Financial to NPC are approved by NPC’s board without any effort to get a better deal from non-Novartis companies. (P. Mem. 10.) Even if true, this allegation concerns the relationship between NPC and Novartis Financial, not NPC and Corporation.
Finally, plaintiffs note that employees of the various Novartis companies are told to think of the corporations as “one big family” (P. Mem. 18), and that NPC files its tax return through Corporation, Novartis Financial, and Novartis Services. Corporation files the consolidated tax return. (P. Mem. 17.) While these allegations do indicate some degree of interrelationship of operations, they do not suggest the degree of entanglement generally found to satisfy this prong of the analysis, even in combination with the other allegations discussed above. Cf. Regan, 1995 WL 413249, at *3 (finding interrelationship of operations where employees rotated informally between the relevant companies, and where employee records, payroll records, and bank deposits of each company were kept together); Parker, 204 F.3d at 341 (relying on pay stubs that listed an employee’s employer as paid “on behalf of’ the parent “through” the subsidiary); Linskey v. Heidelberg Eastern, Inc., 470 F.Supp. 1181, 1184 (E.D.N.Y.1979) (noting that the subsidiary could request employees from the parent, and the parent had the “absolute privilege” of appointing employees to the subsidiary, including its president). If such routine connections among corporate affiliates necessitated a finding of interrelated operations, most large corporate families would count as single enterprises for Title VII purposes. Even if the facts as alleged by plaintiffs are true, the evidence is insufficient for a reasonable factfinder to find for plaintiffs on this prong.
E. Common Management
Common management and common ownership, the last two prongs of the single employer test, “are less important as they *254represent ordinary aspects of the parent-subsidiary relationship.” Meng v. Ipanema Shoe Corp., 73 F.Supp.2d 392, 403 (S.D.N.Y. 1999). Thus, “the mere existence of common management and ownership are not sufficient to justify treating a parent corporation and its subsidiary as a single employer.” Id., quoting Lusk v. Foxmeyer Health Corp., 129 F.3d 773, 778 (5th Cir.1997).
As to common management, plaintiffs first point out that Paulo Costa, CEO of Corporation, was formerly the CEO of NPC, and that other officers have moved from one corporation to the other. This is not “common” management. As a general matter, “the fact that a parent assists a subsidiary’s employee in locating other employment within the company’s group does not indicate centralized control of labor relations where the parent typically provides such assistance.” Balut, 988 F.Supp. at 346 (internal citations omitted). “To hold otherwise would deter a parent from making referrals for valued employees, by transforming a common courtesy into a means of ‘control.’ ” Id. This is no less true when the employee in question is CEO.
As to actual overlap, the boards of directors of Corporation and NPC are separate and distinct, although it is undisputed that there is overlap in the membership of the boards. Three members of Corporation’s board are also members of NPC’s board, and a fourth member of Corporation’s board is an NPC officer. (See P. Mem. 6, D. Mem. 24.) There is no current overlap of corporate officers, although in the past there has been an overlap of up to three officers, none of whose functions pertained to HR or labor relations. (Robinson Deck, D. Ex. 6 119.) These facts must be considered “in the light of the well established principle that directors and officers holding positions with a parent and its subsidiary can and do change hats to represent the two corporations separately, despite their common ownership.” Herman, 18 F.Supp.2d at 312, quoting Lusk, 129 F.3d at 779 (citation omitted).4
Plaintiffs “need to prove more than an overlap in boards or management in order to use the single employer doctrine.” Dewey v. PTT Telecom Netherlands, U.S., Inc., No. 94 Civ. 5983, 1995 WL 425005, at *3 (S.D.N.Y. Jul.19, 1995). In Herman, the court found plaintiffs’ evidence of common management insufficient, despite the fact that the two companies shared a CEO and CFO, because the “two entities maintained distinct management structures,” 18 F.Supp.2d at 313, and, crucially, there was no evidence that the parent’s corporate officers had “participated in any respect in the employment decisions affecting Plaintiffs.” Id. The same is true here. Thus, plaintiffs’ evidence of common management is insufficient on this factor as well.
F. Common Ownership or Financial Control
It is undisputed that Corporation wholly owns Novartis Financial, which in turn wholly owns NPC, and that both Corporation and NPC are ultimately owned by Novartis AG, a non-party corporation based in Switzerland. “A parent corporation’s possession of a controlling interest in a subsidiary entitles the parent to the normal incidents of stock ownership, such as the right to select directors and set general policies, without forfeiting the protection of limited liability.” Meng, 73 F.Supp.2d at 403, quoting Lusk, 129 F.3d at 778. Thus, this factor alone is insufficient to support an integrated-enterprise theory of liability.
“[T]he law only treats the employees of a corporate entity as the employees of a related entity under extraordinary circumstances.” Murray v. Miner, 74 F.3d 402, 404 (2d Cir.1996). This is not that extraordinary case. Because plaintiffs have failed to show that Corporation and NPC function as a single enterprise, Corporation cannot be held liable for any violation of Title YII by NPC, *255and Corporation’s motion for summary judgment will be granted.5
II. Plaintiffs’ Motion for Class Certification
A. Exhaustion and the Scope of the Class Period
Defendants raise a threshold argument which, if meritorious, would affect the scope of the evidence relevant to class certification.
A plaintiff may bring an employment discrimination action under Title VII only after filing a timely charge with the EEOC or with “a State or local agency with authority to grant or seek relief from such practice.” 42 U.S.C. § 2000e-5(e). Defendants argue that plaintiffs did not file a charge with the Equal Opportunity Employment Commission (“EEOC”) satisfying the exhaustion requirement until January 25, 2005, and that the class period, which plaintiffs say opens July 15, 2002, therefore cannot open until March 31, 2004. (Ltr. from Richard H. Schnadig to the Court, dated June 13, 2007.)
A class representative can only represent those individuals who have either filed a timely EEOC charge or could have filed one at the time the class representative’s charge was filed. “[T]he timely filing of an administrative charge by a named plaintiff in a class action satisfies the charge filing obligation of all members of the class.” Tolliver v. Xerox Corp., 918 F.2d 1052, 1056 (2d Cir.1990). Such a charge must give “some indication that the grievance affects a group of individuals defined broadly enough to include those who seek to piggyback on the claim.” Id. at 1058 (discussing the issue under the ADEA). The plaintiffs’ claims must be “reasonably related” to the claims made in the timely EEOC charge, meaning that “the conduct complained of would fall within the scope of the EEOC investigation which can reasonably be expected to grow out of the charge of discrimination.” Holtz v. Rockefeller & Co., 258 F.3d 62, 83 (2d Cir. 2001).6 The Second Circuit has described this principle as “essentially an allowance of loose pleading.” Id. (internal citation and quotation marks omitted).
Plaintiffs argue that the EEOC charge of plaintiff Amy Velez, filed on July 15, 2003, and the November 10, 2004 charge filed by plaintiff Michelle Williams, satisfied the exhaustion requirement.7 Defendants contend that these charges do not raise the same claims raised by the class, and that the earliest relevant EEOC charge by a putative class representative is the January 25, 2005 charge filed by Jennifer Waxman-Recht. (Schnadig Ltr. at 3 & Ex. 1.) If Waxman-Recht’s EEOC charge is the first charge satisfying the exhaustion requirement, the class period could open no earlier than March 31, 2004, which is 300 days before Waxman-Recht’s charge was filed. See 42 U.S.C. § 2000e-5(e)(l) (establishing 300-day filing period for certain states).
Velez’s charge alleges, on behalf of herself and a putative class of women at NPC, that NPC discriminates against her as “part of a continuing pattern and practice of discrimination against female Novartis employees, including but not limited to subjecting female employees to a sexually hostile work *256environment, harassment, and denying female employees the full rights afforded to them under the FMLA.” (Schnadig Ltr. Ex. 2.) Another charge, filed by Michelle Williams on November 11, 2004, makes the same claims, on behalf of Williams and the class. It also claims that the “pattern and practice” of discrimination includes “denying female employees promotions and promotional opportunities in favor of less qualified male employees.” (Schnadig Ltr. Ex. 4.)
Defendants argue that claims of compensation and promotion discrimination are not reasonably related to claims of harassment and hostile work environment. In Holtz, however, the Second Circuit approvingly noted a district court’s conclusion that “it would have been reasonable to suspect that the EEOC, in investigating [a] complaint of failure to train because of age, would have assessed [the corporation’s] promotion and transfer policies.” 258 F.3d at 84 (quotation marks omitted). It would similarly have been reasonable here for the EEOC to investigation pay and promotion claims, because Velez’s complaint specifically references being denied a merit pay increase. (Schnadig Ltr., Ex. 2, at 3.) It also alleges that Velez’s incentive-based compensation is unfairly low because her manager failed to assign her a partner to help with her sales territory. (Id.) In short, Velez’s complaint adequately put NPC on notice of the nature of the class claims.8 Thus, defendants’ arguments are without merit, and there is no exhaustion-related reason why the class period cannot open on July 15, 2002.
B. Class Action Certification Standards
Rule 23 of the Federal Rules of Civil Procedure contains two sets of requirements for class certification. First, the party seeking class certification must show that the four prerequisites of Rule 23(a) have been met. Plaintiffs must show that: (1) the class is so numerous that joinder of all members is impracticable (“numerosity”); (2) there are questions of law or fact common to the class (“commonality”); (3) the claims or defenses of the representative parties are typical of the claims or defenses of the class (“typicality”); and (4) the representative parties will fairly and adequately protect the interests of the class (“adequacy”). Fed.R.Civ.P. 23(a). The court must be persuaded, “after a rigorous analysis, that the prerequisites of Rule 23(a) have been satisfied.” Gen. Tel. Co. of Southwest v. Falcon, 457 U.S. 147, 161, 102 S.Ct. 2364, 72 L.Ed.2d 740 (1982).
Second, the plaintiffs must also show that the proposed class action fits into one of the three categories of class actions listed in Rule 23(b). Those categories encompass class actions where: (1) prosecution of separate actions by individual parties either would create a risk of inconsistent adjudications or would be dispositive of the interests of those members not parties to the adjudication; (2) defendants have acted or refused to act on grounds generally applicable to the class; or (3) questions of law or fact common to members of the class predominate, and a class action is superior to other available methods for adjudication. Fed.R.Civ.P. 23(b).
The Second Circuit’s recent decision in In re Initial Public Offering Securities Litigation, 471 F.3d 24 (2d Cir.2006) (“In re IPO”) significantly clarified the standards for the adjudication of motions for class certification. The Circuit explained that
a district judge may certify a class only after making determinations that each of the Rule 23 requirements has been met; [and] such determinations can be made only if the judge resolves factual disputes relevant to each Rule 23 requirement and finds that whatever underlying facts are relevant to a particular Rule 23 requirement have been established and is persuaded to rule, based on the relevant facts and the applicable legal standard, that the requirement is met. *257Id. at 41. Thus, the court must make a legal determination as to whether class certification is merited, and this process may necessitate fact-finding by the court. “Definitive assessment” of each class certification requirement is required, even if those issues overlap with merits issues. Id. This Court has noted that “the holdings of In re IPO are both significant and narrow — a district judge must consider all of the relevant evidence in determining whether Rule 23 has been satisfied, but a district judge may not go beyond the boundaries of Rule 23 when making such a determination.” Hnot v. Willis Group Holdings Ltd., 241 F.R.D. 204, 209 (S.D.N.Y. 2007) {“Hnot II”)9
“[T]he district judge must receive enough evidence, by affidavits, documents, or testimony, to be satisfied that each Rule 23 requirement has been met.” In re IPO, 471 F.3d at 41. Defendants’ evidence, as well as plaintiffs’, will be considered. Plaintiffs have the burden of showing that the class certification requirements have been met. Id. at 40.
The proposed class consists of “all women who are currently holding, or have held, a sales-related job with Novartis during the time period July 15, 2002 through the present.” (P. Mem. 34.) Defendants do not challenge the numerosity of this class, which plaintiffs say could number in the thousands.10 The Court concludes that the class “is so numerous that the joinder of all members is impracticable.” Fed.R.Civ.P. 23(a)(1). Nor do defendants challenge the ascertainability of the class, which is clear. See Dunnigan v. Metro. Life Ins. Co., 214 F.R.D. 125, 135 (S.D.N.Y.2003) (discussing implicit threshold requirement of aseertainability). Three questions remain to be analyzed under Rule 23(a): commonality, typicality, and adequacy of representation.
C. Commonality Under Rule 23(a)
As various courts have noted, the requirements of commonality and typicality “tend to merge” because “[b]oth serve as guideposts for determining whether ... the named plaintiffs claim and the class claims are so inter-related 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. “The crux of both requirements is to ensure that maintenance of a class action is economical and that the named plaintiffs claims and the class claims are so interrelated that the interests of the class members will be fairly and adequately protected in their absence.” Marisol A. v. Guiliani, 126 F.3d 372, 376 (2d Cir. 1997) (internal citations, alterations and quotation marks omitted).
To determine commonality, it is not necessary to decide whether plaintiffs’ evidence is ultimately compelling. In re IPO makes clear that courts may resolve contested factual issues where necessary to decide on class certification, and when a claim cannot succeed as a matter of law, the Court should not certify a class on that issue. See In re IPO, 471 F.3d at 42 (denying class certification as to the issue of reliance because the presumption on which plaintiffs’ theory depended was inapplicable). However, “[cjommonality requires that plaintiffs present common questions of fact or law; plaintiffs’ ultimate success at trial on the merits requires an answer to that question, specifically that defendants actually did discriminate against plaintiffs.” Hnot II, 241 F.R.D. at 211 (emphasis in original). “For the Court to decide which expert report was *258more persuasive would be to decide whether the class was actually discriminated against by defendants. This the Court was not required to do, either before or after, In re IPO.”Hnot II, 241 F.R.D. at 211.
Thus, the Court may only examine expert reports or other evidentiary submissions “as far as they bear on the Rule 23 determination.” Id. at 210. For example, in order to analyze whether the class is large enough to satisfy Rule 23(a)’s numerosity requirement, it is necessary to know how many members are in the class. In re IPO, 471 F.3d at 40. A Court should not refrain from such fact-finding simply because it may overlap with a merits inquiry. However, “ ‘statistical dueling’ is not relevant to the certification stage unless such dueling presents a valid basis for denying class certification. ” Hnot II, 241 F.R.D. at 210 (internal citations and quotation marks omitted; emphasis in original).11 In short, the Court will consider all of the relevant statistical and anecdotal evidence to determine whether the requirements of Rule 23 have been met, making whatever subsidiary factual or legal findings are required as part of that analysis.
Plaintiffs assert that four questions of fact are common and appropriate for certification: (1) whether NPC’s employment policies are overly subjective and discriminatory; (2) whether NPC paid women less than similarly-situated male employees; (3) whether NPC denied or delayed promotion of women; and (4) whether NPC discriminates against pregnant employees.12
As to each of these questions, plaintiffs contend, each class member’s case will revolve around the same statistical and anecdotal evidence, personnel management analysis, deposition testimony, and documents. Each class member’s case, they contend, will present the same questions of law: whether a pattern or practice of disparate treatment exists at NPC, and whether NPC’s policies have a disparate impact on women. (P. Mem. 38-39.) In response, defendants essentially attack the merits of plaintiffs’ case, arguing that plaintiffs’ expert and anecdotal evidence fails to show bias or disparate impact. (D.Mem. 20-33.)
1. Structure of the NPC Personnel Evaluation and Management System
Plaintiffs contend that NPC’s personnel evaluation and management system is overly subjective, and that this subjectivity leads to discrimination.
In Caridad v. Metro-North Commuter Railroad, 191 F.3d 283 (2d Cir.1999), the Second Circuit held that the commonality requirement could, in some circumstances, be satisfied by a challenge to “the subjective components of company-wide employment practices,” based on a theory that “the grant of discretionary authority to supervisory employees either results in a pattern and practice of discrimination or affects one class of employees more harshly than others.” Id. at 291-92 (internal citation omitted). In other words, it is possible for class action plaintiffs to make out a discrimination claim on the basis of allegedly excessive subjectivity in the company’s employment practices. As this Court has previously held, In re IPO did not undermine or even address this holding, which remains good law. Hnot II, 241 F.R.D. at 210.
Of course, class certification is not appropriate unless plaintiffs can show “that the challenged practice is causally related to a pattern of disparate treatment or has a disparate impact.” Caridad, 191 F.3d at 292. “Where the decision-making process is difficult to review because of the role of subjec*259tive assessment, significant statistical disparities are relevant to determining whether the challenged employment practice has a class-wide impact.” Id. The court in Caridad found the statistical and anecdotal evidence presented in that case sufficient to show common questions of fact regarding the implementation of company-wide employment policies, regardless of whether the evidence would ultimately be persuasive on the issue of liability. Id.
At NPC, a substantial portion of employees’ evaluations are based on subjective factors, and the evaluations have a direct impact on employees’ compensation and chances for promotion. Plaintiffs’ expert David C. Martin analyzed “the performance management and related compensation system” used by NPC to determine whether it was “vulnerable to bias in decision making.” (Martin Report, P.Ex. 36 (“Martin Rep.”), at 1.) Martin’s report criticizes the performance evaluation system on a number of grounds. It argues that the ratings do not necessarily correspond to real performance, because they are subject to being modified by higher-level supervisors and because they are forced to fit a prescribed curve or “forced distribution” (id. at 8-9), and because appeals are directed to the same manager who had earlier approved the rating or to an HR employee with no authority to actually change the rating. (Id. at 9.) The report argues that each decision in the evaluation process is essentially subjective (id. at 11-12) without giving much detail; indeed, the entire report is only 14 pages long.
NPC argues that Martin’s report is flawed because, as Martin acknowledges, he “didn’t look at a single performance appraisal.” (Martin Dep., D. Ex. 36, at 21.) This argument is not persuasive. Martin’s report is offered to show the flaws in the system’s structure, not in its implementation. Martin did not purport to offer evidence that the system at NPC actually causes disparate treatment or has a disparate impact; he merely offered to show how the system makes discrimination possible. Whether his report and testimony do so successfully is ultimately a question for the jury. The report is sufficiently persuasive, however, to permit a conclusion, at this preliminary stage, that plaintiffs have raised a common question about whether NPC’s system is structured in a way that facilitates discrimination, and not merely a collection of individual claims of particular unfair evaluations.
NPC also asserts that Martin admits that NPC’s evaluation process includes many of the procedures and safeguards that he himself has recommended in published writings to cabin subjectivity. (D.Mem. 18.) This is an appropriate ground for impeachment of Martin’s conclusions, and may be offered to the factfinder as such, but it is hardly a basis for finding that Martin’s report raises no common question as to the plaintiffs in the putative class.13
Martin’s report is, by itself, insufficient to support class certification on any issue, because it illustrates only the potential for discrimination. The subjectivity of the NPC personnel management system presents a certifiable common question if, and only if, plaintiffs can show “that the challenged practice is causally related to a pattern of disparate treatment or has a disparate impact.” Caridad, 191 F.3d at 292. In order to show such an impact, plaintiffs offer statistical and anecdotal evidence of disparate impact in performance evaluation scores, compensation, and promotions, and of discrimination on the basis of pregnancy.
2. Disparity in Performance Evaluation Scores
Plaintiffs submit the report of Dr. Louis R. Lanier to show the discriminatory effects of *260NPC’s employment policies in the various areas in which discrimination allegedly occurred, including the scores on the performance evaluations themselves. As to the performance evaluations, Lanier finds that class members were systematically given lower ratings than male counterparts. (Lanier Report, D. Ex. 37 (“Lanier Rep”), at 4.)
NPC argues that Lanier’s analysis of different job performance scores should have controlled for job level, because higher-level staff, who were presumably promoted because of a tendency to achieve high scores on performance evaluations, will score disproportionately high. (D. Mem. 20, Welch Rep. 38.) According to Welch, after controlling for job level, the differences in high ratings given to women was statistically insignificant. (Welch Rep. 40.) Lanier responds that the “competencies” or skills required of higher-level employees are different than the competencies required of lower-level employees. (Lanier Deck, D. Ex. 39 (“Lanier Deck”), at 7.) That is, one can be a good sales representative but turn out to be a terrible manager.
Lanier also argues that controlling for job level in the analysis of disparate performance scores would introduce inaccuracies, because there are fewer females in higher-level jobs. (Lanier Rep. 14-15; see also Lanier Deck 6 (speculating that discrimination may cause women to depart earlier in their careers).) Defendants argue that there is no evidence that this unequal distribution across job levels is due to discrimination, and that therefore the control should be used anyway. While the parties disagree on the propriety of various controls, their arguments go to the merits of the expert analysis; they do not bear on whether the question of discrimination in performance evaluations is common across the class.
Other arguments also relate to the merits of Lanier’s approach, not to the question of commonality. Defendants contend that Lanier errs by counting each performance review as separate, even though some represent multiple reviews of the same individual. They argue that a higher-performing employee is likely to continue to be higher-performing. Plaintiffs respond that to control for multiple reviews of one employee would be to ignore the possibility of multiple acts of discrimination against a single employee. (P. Reply 13.) Defendants further argue that Lanier did not analyze whether ratings were given by female or male managers. (D.Mem. 22.) A culture of discrimination, however, may be such that female raters as well as male raters grade women unfairly; the Court cannot assume that women are incapable of discriminating against other women. Defendants’ objections to Lanier’s findings with respect to performance evaluation scores are unconvincing. The question presented — whether women are systematically given lower scores on performance evaluations — is common to the class, and it cannot be said that Lanier’s report is insufficient to present a serious question for the factfinder to resolve.
3. Discrimination in Pay
Lanier finds that class members were significantly under-compensated compared to their male counterparts. Compared to males in similar job positions, women were paid approximately $74.82 per month less. (Lanier Deck 11.) When all males and females are compared without regard to job position, women with similar levels of experience are paid approximately $220 less per month than men with comparable experience, which Lanier says shows both a pay inequity and a failure to promote women commensurate with their experience. (Id.)
Defendants argue that it is inappropriate to examine the pay disparity across all job groups. Higher-level employees make more money, they point out, so comparing pay without accounting for job level proves only that there are fewer higher-level women employees at NPC. (D.Mem. 29.) It is true that pay disparities not adjusted for job level are not a comparison of similarly-situated employees, and are therefore not necessarily evidence of discrimination in pay. The unadjusted comparison is relevant, however, to the question of whether women at NPC actually suffer any adverse impact as a result of bias in the allegedly subjective evaluation system. If it is true that bias reduces a woman’s chances of promotion, then it is useful to identify the resulting pay disparity *261as an additional consequence of the subjective and biased evaluation system.
Moreover, a significant disparity of $74.82 per month remains even after adjustment for job level. Defendants argue that $74.82 is only 1% less than the average male earnings of $7,463 per month. (Welch Rep. 12; see D. Mem. 24.) Plaintiffs respond, however, that the difference is significant to 5.4 standard deviations (see Lanier Decl. 9), and of course even discrimination that costs its victims only $897.84 per year is unlawful and compensable under Title VII.
Defendants argue that Lanier wrongly excludes from his pay calculation certain groups of employees, in particular employees hired into the sales force during the relevant pay period, and that the exclusions are “disproportionately female and those occupying entry level sales jobs.” (Welch Rep. 6.) If not for these exclusions, NPC’s expert argues, the correct pay differential would show that women were paid $19.55 more per month. (Id.) Lanier replies the exclusions are sensible because male and female employees in the excluded groups are not similarly situated — for example, men are less likely than women to work less than one year because they are on unpaid leave. (Lanier Decl. 4.) This is precisely the sort of “statistical dueling” that should be resolved by a factfinder.
Welch also argues that of the 12 types of job identified by Lanier, the difference in pay is statistically significant in only three. (Welch Rep. 8.) Lanier responds that the statistical significance of differences in individual job groups is immaterial to the larger question of whether there is a pattern of discrimination at NPC. (Lanier Decl. 4.) Moreover, Lanier responds that in 11 of the 12 groups, men were paid more than women, which is strong evidence of discrimination. (Id. at 12.) Welch also argues that Lanier’s analysis of “similarly-situated” women fails to control for work or educational experience prior to the employee’s hiring at NPC, but offers no evidence suggesting that controlling for these variables would have changed the outcome of Lanier’s analysis. (Welch Rep. 7.) These disagreements about statistical method are not appropriate for resolution at this stage.
Welch also argues that Lanier’s analysis is corrupted by his use of age as a proxy for experience (to determine whether women are similarly situated to their male counterparts). Women, Welch believes, are more likely to spend time out of the workforce, and so a woman and man of the same age are statistically unlikely to have comparable work experience. Lanier disputes the relevance of this proposition to the data (Lanier Decl. 5), but this, again, is a question for the factfinder. Lanier’s findings are sufficiently rooted in accepted statistical methodology to be received into evidence, and thus the accuracy of his conclusion identifying classwide discrimination presents a factual issue for trial common to the entire class.
Defendants argue that the difference in pay cannot be causally linked to discrimination because it includes types of pay that are not linked to subjective managerial judgments, including incentive-based pay. (D.Mem. 24.) Welch argues that Lanier should have used only base pay, not commissions, because commissions are formula driven, and not related to the performance evaluations. (Welch Rep. 7.) Had Lanier done so, according to Welch, he would have found the difference in pay to be a statistically insignificant five dollars. (Id.) Lanier argues that although commissions are formula-driven, the goals built into the formula and the assignments of the employees (which affect their ability to meet sales goals) are created through managerial discretion, and that incentive payments are therefore fairly considered in the pay disparity. (Lanier Decl. 3.)
Plaintiffs’ claims of pay discrimination would be more difficult to prove if, as Lanier’s report suggests, they rest not on a theory that women are paid less in the traditional sense, but on a theory that women earn less because of the goals set and assignments given to them by NPC management. However, Lanier’s report still raises a question that is common to the class: whether NPC management’s discretionary actions result in lower pay for women throughout the corporation. Plaintiffs offer significant anecdotal evidence from various women claiming that *262they were not paid consistently with then-performance (see P. Exs. 27-34), some of whom testified to the effects of their assignments on their pay. (See, e.g., P.Ex. 28, Bernice Dezelan Decl. H11.) This evidence supports their contention that the discretion allowed to managers by NPC’s personnel management system was abused. See Rossini v. Ogilvy & Mather, Inc., 798 F.2d 590, 604 (2d Cir.1986) (“In evaluating all of the evidence in a discrimination case, a district court may properly consider the quality of any anecdotal evidence or the absence of such evidence.”). Thus, even if plaintiffs’ claim is based not on direct base pay discrimination, but on pay disparities that resulted indirectly from assignment and other management actions, there is sufficient evidence to show that the question is common to the class.14
4. Discrimination in Promotions
Lanier’s report concludes that women at NPC had a lower probability of promotion, and that women are underrepresented in the Management Development Program (MDP), which is a prerequisite to promotion to management. (Lanier Rep. 5.) Moreover, plaintiffs allege that women at NPC are overrepresented at lower-level positions in the corporate hierarchy, but increasingly underrepresented at successively higher levels. (See P. Reply 15-16.)
Defendants argue that it is unfair to look at the actual representation of women at the various levels of the organization’s hierarchy, because employees reach their positions in that hierarchy through a number of events and decisions, some of which are not under the discretionary control of NPC management. (D.Mem. 28.) Defendants do not dispute, however, that promotion depends at least in substantial part on performance evaluations.
Defendants also argue that the snapshot approach employed by Lanier is not a useful methodology. “After specifying the employment practice allegedly responsible for excluding members of their protected class from a benefit, plaintiffs must identify the correct population for analysis. In the typical disparate impact case the proper population for analysis is the applicant pool or the eligible labor pool.” Smith v. Xerox Corp., 196 F.3d 358, 368 (2d Cir.1999), overruled on other grounds by Meacham v. Knolls Atomic Power Lab., 461 F.3d 134 (2d Cir.2006). NPC argues that Lanier’s findings with respect to representation in the hierarchy are invalid because they do not identify a relevant pool of women who were eligible and available for promotion, and that without identifying benchmarks based on the pool of available women, “no inference of discrimination can be drawn from the workforce profile.” (D.Mem. 29.)
In this case, it is true that Lanier’s snapshot of the corporate hierarchy does not identify an eligible pool of employees, and is therefore not directly useful in determining whether the decisions that caused the unequal distribution were infected by discrimination. Lanier does, however, offer an analysis of gender disparities in actual promotion decisions — that is, cases where employees were promoted within NPC — as does Welch. In both analyses, the eligible pool of employees is simply those employees at NPC who are eligible to be promoted to manager — that is, sales staff. Defendants do not contend that this is an inappropriate “eligible pool.”
Lanier’s analysis of promotions from sales employee to first-line manager concludes that male sales employees are 4.9 times more likely to be promoted to first-line manager than female sales employees. (Lanier Decl. 13.) As to this analysis, Welch faults Lanier for analyzing only one group of promotions, those from sales representative to first-level managerial positions, which only affects a small fraction of the class. Welch argues that Lanier’s analysis is based on too small a *263sample, but Welch does not dispute that within that sample the probability of male promotion is dramatically greater than the probability of female promotion; the disparity is significant to more than five standard deviations. (Lanier Decl. 13.)15
Lanier’s analysis also shows that women are underrepresented in the MDP, which is a prerequisite to promotion to manager. 15.2% of eligible males were selected for the MDP, compared to only 9.1% of eligible females, which is significant to 6.0 standard deviations. (Lanier Rep. 20.) Defendants argue that Lanier’s criteria for eligibility are based on his own guesses as to who might be eligible for such a program, rather than any official criteria of NPC’s. (D.Mem. 32.) This argument is specious. Defendants acknowledge that “there are no Company-wide requirements for nomination” (D.Mem. 14), so Lanier had no alternative but to rely on his own guesses as to who might be eligible. Defendants point to no inaccuracies or implausibilities in Lanier’s estimates of which personnel are “eligible” for participation in the MDP, and present no alternative definition of eligibility or statistical analysis based on an alternative definition.16
Lanier’s statistical findings are supported by anecdotal evidence that gives “texture” to the statistics. Robinson v. Metro-North Commuter R.R. Co., 267 F.3d 147, 168 (2d Cir.2001). Various declarants allege that they were passed over for promotions, or that they were kept out of training sessions to which their male counterparts were invited. (See, e.g., P.Ex. 29, Debra Benbow Decl. 1110; id., Deborah Davis Decl. 1111). Taking all the evidence into account, “[wjhether or not plaintiffs’ statistical evidence is ultimately sufficient to establish that plaintiffs and the members of their proposed class have actually suffered discrimination, it is certainly adequate to establish that whether or not [NPC’s] promotion and compensation policies subject class members to discrimination is an issue common to all class members.” Hnot I, 228 F.R.D. at 483.
5. Discrimination on Account of Pregnancy
The Pregnancy Discrimination Act of 1978(PDA) provides that Title VII’s prohibition on gender discrimination includes discrimination on the basis of:
pregnancy, childbirth, or related medical conditions; and women affected by pregnancy, childbirth, or related medical conditions shall be treated the same for all employment-related purposes, including receipt of benefits under fringe benefit programs, as other persons not so affected but similar in their ability or inability to work.
42 U.S.C. § 2000e(k).
Lanier’s report concludes that women who take FMLA leave earn an average of $210.20 less per month for the first six months after returning than women not returning from leave, which is statistically significant to 6.4 *264standard deviations, and that they experience a slightly lower disparity for the following six months.17 (Lanier Rep. 24.)
Lanier’s analysis of the effects of pregnancy leave includes both base salary and incentive-based earnings (i.e., pay that is determined on the basis of sales in a representative’s territory). Defendants argue that there is nothing discriminatory in failing to pay women (and men) incentive bonuses that are not earned. When the effects of incentive pay are excluded, according to Welch, the base pay of women who take leave is unaffected. (Welch Rep. 44.)
Plaintiffs do not dispute that pregnancy is treated the same as any other legitimate reason for leave by NPC’s compensation policies. But women at NPC, Lanier notes, take 76 times more FMLA leave than do men. (Lanier Decl. 8.) Plaintiffs argue that the NPC’s payment structure, which gives employees who return after taking leave for pregnancy the same partially incentive-based compensation as all other employees, has a larger and more detrimental impact on women than on men, even though it treats pregnancy the same as any other legitimate reason for leave. They argue that NPC “could implement, but has not [implemented], a commission payment system that accounts for time away from the territory during FMLA leave.” (P. Reply 22.)
“It has been repeatedly affirmed that the PDA does not require the creation of special programs for pregnant women; nor does it mandate any special treatment. To the contrary, the statute specifically requires that pregnant women be treated the same as all other employees with similar disabilities.” Dimino v. N.Y.C. Transit Auth., 64 F.Supp.2d 136, 157 (E.D.N.Y.1999). See Urbano v. Cont’l Airlines, 138 F.3d 204, 206 (5th Cir.1998) (affirming summary judgment where a defendant treated the plaintiff “in exactly the same manner as it would have treated any other worker who was injured off the job”); Gratton v. JetBlue Airways, No. 04 Civ. 7561, 2006 WL 2037912, at *6 (S.D.N.Y. Jul. 21, 2006) (holding that plaintiff “has not identified any accommodation given to other temporarily disabled employees that was withheld from her”); Minott v. Port Authority of N.Y. and N.J., 116 F.Supp.2d 513, 521 (S.D.N.Y.2000) (“Title VII and the Pregnancy Discrimination Act do not protect a pregnant employee from being discharged for absenteeism even if her absence was due to pregnancy or complications of pregnancy, unless other employees are not held to the same attendance standards.”).
In this ease, the compensation system does not differentiate between employees who take leave for pregnancy and employees who take leave for other reasons. If sales figures in a territory drop while a representative is on leave, that representatives’ incentive payments will fall, regardless of the reason for the leave. “A policy may discriminate between those employees who take off long periods of time in order to raise children and those who either do not have children or are able to raise them without an appreciable career interruption. That is not inherently sex specific and does not give rise to a claim under Title VII.” Fisher v. Vassar Coll., 70 F.3d 1420, 1448 (2d Cir.1995) reheard en banc on other grounds, 114 F.3d 1332 (2d Cir.1997). The failure to adjust incentive payments is not unlawful because “[t]he Pregnancy Discrimination Act requires the employer to ignore an employee’s pregnancy, *265but ... not her absence from work, unless the employer overlooks the comparable absences of nonpregnant employees ... in which event it would not be ignoring pregnancy after all.” Troupe v. May Dep’t Stores Co., 20 F.3d 734, 738 (7th Cir.1994).
There appears to be no factual dispute over whether NPC’s compensation policies systematically discriminate against women except insofar as they fail to adjust incentive payments that drop when an employee goes on leave. Under the PDA, “it is discriminatory to treat pregnancy-related conditions less favorably than other medical conditions.” Newport News Shipbuilding and Dry Dock Co. v. E.E.O.C., 462 U.S. 669, 684, 103 S.Ct. 2622, 77 L.Ed.2d 89 (1983). It is not discriminatory to treat pregnancy-related leave the same as other forms of leave, and plaintiffs have offered no evidence that NPC’s compensation policies do otherwise.18 Accordingly, plaintiffs’ motion to certify a class is denied as to their claims of discrimination in pay on the basis of pregnancy.
There are, however, other allegations of pregnancy discrimination. The complaint alleges that women returning from pregnancy leave are subjected to denial of promotions and promotional opportunities, “stricter scrutiny,” hostile comments, unreasonable discipline, and “adverse employment actions” upon return. (4th Am.Compl. U 68.) In support of these allegations, plaintiffs rely on the anecdotal evidence of the twenty-eight declarants and three plaintiffs who claim to have been the victims of pregnancy discrimination. (See P. Mem. 43-44.)
The commonality and typicality requirements can be satisfied by “affidavits, statistical evidence, or both.” Attenborough v. Const. and General Bldg. Laborers’ Local 79, 238 F.R.D. 82, 95 (S.D.N.Y.2006). Anecdotal testimony is generally used as a supplement to statistical evidence, to bring “the cold numbers convincingly to life.” Int’l Bhd. of Teamsters v. United States, 431 U.S. 324, 339, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). At the liability phase, anecdotal evidence will be usually be introduced not “to establish that the particular instances of discrimination actually occurred nor that the particular employees were in fact victims of discrimination,” but simply to provide “texture” to the statistical evidence. Robinson, 267 F.3d at 168. Courts have discretion to rely on anecdotal evidence, “but they commonly do so where the statistical evidence is weak.” Hnot I, 228 F.R.D. at 484.19 When *266statistical evidence does not exist, however, anecdotal evidence can suffice if “plaintiffs ... compile sworn statements from a large enough sample of plaintiffs to demonstrate that common issues exist.” Cokely v. N.Y. Convention Ctr. Operating Corp., No. 00 Civ. 4637, 2003 WL 1751738, at *3 (S.D.N.Y. Apr. 2, 2003).
As to the extent of the anecdotal evidence required when statistical evidence is unpersuasive, plaintiffs argue that they “are not required to produce a statistically significant amount of declarations.” (P. Reply 14.) Cf. Ross v. Nikko Secs. Co. Int’l, 133 F.R.D. 96, 97 (S.D.N.Y.1990) (holding that in the absence of statistical proof of commonality, “[t]he testimonial proof must identify a statistically significant number of aggrieved persons in the putative class in relation to the size of the relevant work force.”). The concept of statistical significance is not useful here. A number is statistically significant when it differs from the number that could be expected from a random distribution, that is, when it is not within the margin of error. To determine what number of declarations are “statistically significant,” it would be necessary to determine how many declarations would be filed due to the operation of randomness — that is, how many declarations would be filed in the absence of any class-wide discrimination. It is difficult to imagine any meaningful way to identify such a number.
Moreover, anecdotal evidence is not statistical evidence. The declarations are offered not primarily for their quantity, but for their quality. The testimony in the declarations is valuable insofar as it could persuade a reasonable factfinder that a pattern or practice of discrimination exists at NPC. The factfinder would examine the content of the testimony to determine whether the incidences of discrimination described actually occurred, and, if so, whether they are isolated incidents or symptomatic of a deeper pattern. That is, the factfinder would examine the declarations not merely to see how many have been produced, but to see what they say.
Of course, the anecdotal evidence must encompass “a large enough sample of plaintiffs to demonstrate that common issues exist.” Cokely, 2003 WL 1751738, at *3. But there is no minimum number of statements that must be compiled in relation to the total number of similarly-situated employees. Rather, the question is whether the statements submitted, in light of their persuasiveness and whether the incidents they describe appear to be isolated or generalized, “show that sex discrimination ‘was the company’s standard operating procedure — the regular rather than the unusual practice.’ ” Carter v. Newsday, Inc., 528 F.Supp. 1187, 1197 (E.D.N.Y.1981), quoting Teamsters, 431 U.S. at 336, 97 S.Ct. 1843. “In the case ... of a showing of nonexistent statistical discrimination, anecdotal testimony must by itself support an inference of sex-based [employment] decisions.” Carter, 528 F.Supp. at 1197.
Three plaintiffs (Earl, Compl. H 302, Durkin, id. 11323-337, and Deyne, id. UH 389-90) and 28 declarants have submitted declarations claiming pregnancy discrimination. (See P.Ex. 35.) The declarants claiming pregnancy discrimination worked for NPC in fourteen different states. (Id.) They point to a variety of ill-treatment suffered by pregnant women and mothers at NPC, from arbitrary discipline to verbal harassment to denial of promotion and termination. All but one give specific examples from personal experience of differential treatment on the basis of pregnancy or motherhood.20 Many attest to *267specific comments by managers indicating hostility to pregnancy, and are therefore directly relevant to the question of whether a nationwide pattern or practice of discrimination exists. One declarant reports being told by her manager that she “did not qualify for a pay increase because [she] had not been in her territory during [her] Maternity Leave.” (P.Ex. 32, Ramona Pouncy Decl. H12.) Another claims that her manager told her that he preferred not to hire young females, explaining, “First comes love, then comes marriage, then comes flex time and a baby carriage.” (P.Ex. 30, Jennifer Ryan Tselikis Decl. 1113.) Another declarant says that after her leave, she was disciplined for low sales numbers and her teammate was not, although the two employees were “listed together” for purposes of sales rankings. (P.Ex. 33, Renee Tittle Decl. H 8.) Another manager allegedly encouraged a declarant employee to get an abortion. (P.Ex. 28, Christine Macarelli Decl. 117.) Still another declarant alleges that employees were urged during a training session to avoid getting pregnant. The declarant, five months pregnant at the time, drew the eye of the trainer, who said, “Oops, too late.” (P.Ex. 29, Ivette Flower Decl. H 6.)
It is of course possible that these declarants will be found not credible. At this stage, however, it is only necessary to determine whether a common question has been presented. It has. See Selzer v. Bd. of Educ., 112 F.R.D. 176, 180 (S.D.N.Y.1986) (although the statistics were inconclusive, affidavits from named plaintiffs and five proposed class members were sufficient to establish the existence of an aggrieved class); Donaldson v. Pillsbury Co., 554 F.2d 825, 830-32 (8th Cir.1977) (aggrieved class established where plaintiff produced six affidavits alleging discrimination and identified eighteen other individuals who claimed to be victims of discriminatory policies). A “court must be wary of a claim that the true color of a forest is better revealed by reptiles hidden in the weeds than by the foliage of countless free-standing trees,” NAACP v. Claiborne Hardware Co., 458 U.S. 886, 934, 102 S.Ct. 3409, 73 L.Ed.2d 1215 (1982), but in this case plaintiffs have produced enough foliage to raise a question about the forest’s color. Whether or not the declarations are ultimately convincing to a factfinder, they are numerous enough and detailed enough to establish that a common question exists.
Moreover, in deciding whether the anecdotal evidence specifically directed at pregnancy discrimination is adequate to establish a common question, it would be inappropriate to ignore the statistical evidence adduced in other contexts in this case. Pregnancy discrimination, after all, is a form of discrimination against women, and so the fact that plaintiffs have offered significant statistical evidence of other forms of gender discrimination sheds light on their anecdotal evidence of pregnancy discrimination. The Court is required to consider all the evidence, and in context it is clear that a common question is presented. Accordingly, class certification will be granted on the question of pregnancy discrimination, except with respect to the issue of incentive-based compensation discussed above.
D. Typicality Under Rule 23(a)
“Typicality ... requires that the claims of the class representatives be typical of those of the class, and is satisfied when each class member’s claim arises from the same course of events, and each class member makes similar legal arguments to prove the defendant’s liability.” Marisol A., 126 F.3d at 376 (internal quotation marks and citations omitted). Typicality “does not require that the factual background of each named plaintiffs case be identical to that of all class members; rather it requires that the disputed issue of law or fact occupy essentially the same degree of centrality to the named plaintiffs claim as to that of other members of the proposed class.” Caridad, 191 F.3d at 293 (internal quotations and citations omitted).
To show that plaintiffs are not typical of the class, defendants offer an analysis by Dr. *268Welch showing that the named plaintiffs’ claims of promotion and compensation discrimination are not reflective of other women hired for similar jobs at the same time as plaintiffs. Welch concludes each named plaintiffs was, with one exception, the only-woman in her “hiring cohort” — the group of women hired for similar jobs at the same time — -who experienced the disparate treatment of which she complains. (D. Mem. 35; Welch Rep. 11-12.)
This is a potentially significant finding, which could be interpreted to suggest, as defendants contend, that plaintiffs’ treatment “was attributable to something specific to [each] Plaintiff.” (D.Mem. 35.) Plaintiffs respond that this method effectively carves up each sample into a group too small to have any statistical meaning — also a potentially convincing point. (P. Reply 5.) These questions are for the factfinder. “As long as plaintiffs assert, as they do here, that defendants committed the same wrongful acts in the same manner, against all members of the class, they establish [the] necessary typicality.” In re Towers Fin. Corp. Noteholders Litig., 177 F.R.D. 167, 170 (S.D.N.Y.1997) (internal quotation marks and citation omitted). Moreover, in a ease alleging that subjectivity in evaluation systems resulted in disparate treatment, it is entirely foreseeable that incidences of discrimination will be distributed throughout the corporation, rather than clustered in any particular “hiring cohort.”
Defendants also attempt to show that each named plaintiff is unusual in some respect. For example, they contend that plaintiff Velez performed poorly and had attendance problems, and that plaintiff Lopes reported damage to her company car and missed work while in substance abuse programs. (D.Mem. 36.)21 Defendants attack the typicality of various named plaintiffs’ claims, essentially arguing that those claims are “unique” because the particular factual circumstances of each plaintiffs situation — not discrimination — explains their treatment. (D.Mem. 37-46.) This is, of course, always the defendant’s contention in class action discrimination claims: that the plaintiffs suffered no discrimination, or at least that any discrimination that occurred was isolated rather than systematic.
Defendants cannot rebut typicality by claiming that something other than discrimination explains the named plaintiffs’ experience. The question presented by each plaintiffs claim is undoubtedly typical of the class, whether or not defendants are eventually able to prove that the answer to that question is unique to each plaintiff. “The primary criterion for determining typicality is the forthrightness and vigor with which the representative party can be expected to assert the interests of the members of the class.” Latino Officers Ass’n v. City of New York, 209 F.R.D. 79, 89-90 (S.D.N.Y.2002) (internal citation and quotation marks omitted). Defendants have offered no persuasive basis on which to question plaintiffs’ typicality. The Court finds that the named plaintiffs’ claims are entirely typical of those presented on behalf of the class.
E. Adequacy of Representation Under Rule 23(a)
The last of the Rule 23(a) requirements is adequacy of representation. A party seeking class certification must show that “the representative parties will fairly and adequately protect the interests of the class.” Fed. R.Civ.P. 23(a)(4).
There are two separate inquiries under Rule 23(a)(4). One requires a showing that class counsel is experienced, qualified, and able to conduct litigation. “In determining the adequacy of counsel, the court looks beyond reputation built upon past practice and examines counsel’s competence displayed by present performance.” Towers Fin., 177 F.R.D. at 171 (internal quotation marks and citation omitted). Defendants do not contest counsel’s experience, qualifications, or ability *269to conduct the litigation, and the Court finds that counsel is suitable.
The second requirement is that “[p]laintiffs must ... demonstrate that there is no conflict of interest between the named plaintiffs and other members of the class.” Marisol A., 126 F.3d at 378 (citation omitted). See Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 625-26, 117 S.Ct. 2231, 138 L.Ed.2d 689 (1997) (“The adequacy inquiry under Fed. R.Civ.P. 23(a)(4) serves to uncover conflicts of interest between named parties and the class they seek to represent. A class representative must be part of the class and possess the same interest and suffer the same injury as the class members.”); In re Visa Check/MasterMoney Antitrust Litig., 280 F.3d 124, 142 (2d Cir.2001) (requiring courts to “ask whether plaintiffs interests are antagonistic to the interest of other members of the class”) (quotation marks and citation omitted). “[N]ot every potential disagreement between a representative and the class members will stand in the way of a class suit. The conflict that will prevent a plaintiff from meeting the Rule 23(a)(4) prerequisite must be fundamental, and speculative conflict should be disregarded at the class certification stage.” Id. at 145.
NPC argues that there are three problems with the named plaintiffs’ representation of the class. First, NPC contends that the inclusion of pregnancy discrimination claims creates conflict between class members. As previously discussed, sales representatives at NPC are paid partly on an incentive basis, with pay determined by sales figures in the territory to which the given team of representatives is assigned. NPC argues that this creates “inherent tensions” between a representative who takes leave and the teammates who are still working in the territory: if one representative leaves, the other representative’s pay will suffer if sales decline due to their absence. (D.Mem. 48.) It is possible to imagine tension over lost incentive pay being a factor in some relationships between pregnant women and their teammates, but it is difficult to imagine this tension playing any role in the current litigation, and indeed NPC does not suggest any particular impact that the putative tension might have on these proceedings.
Second, NPC argues that the inclusion of managers in the class creates conflict. It notes that at least eleven declarants allege that their female managers harassed them on the basis of their pregnancies and resulting leaves. While this shows that female managers may be subject to discrimination claims in this litigation, it does not show that the plaintiff manager (there is only one) will herself be subject to discrimination claims. “Even if one female officer supervised another, it is still possible, as plaintiffs allege, that they all suffered from gender discrimination by the key decisionmakers.” Hnot I, 228 F.R.D. at 485. In a similar case involving alleged discrimination within the New York Police Department, a district court rejected this argument for reasons that apply equally well here:
The Court does not see more than a hypothetical conflict. Commanding-officer class members allegedly are subject to the disparate impact of the disciplinary system themselves. There is no reason for the Court to think that the commanding officer class members have any less of a desire to end any discrimination within the NYPD than the police officer members. Their interests on this point are aligned. If an actual conflict develops, the Court is prepared to revisit this question and consider certifying a separate subclass for each rank of uniformed officer.
Latino Officers Ass’n, 209 F.R.D. at 90. This Court rejected a similar contention of conflict in Hnot I, noting that “[i]f supervisory employees and supervisees all are subject to discrimination, all have an equal interest in remedying the discrimination, and the named plaintiffs can still be expected to litigate the case with ardor. A potential for conflict need not defeat class certification.” Hnot I, 228 F.R.D. at 485-86.
Finally, defendants argue that there is no named plaintiff who can adequately represent the interests of the female managers in the class, because only one named plaintiff, Kelly Corbett, is a manager. {See Compl. 1111279, 288.) Corbett, according to NPC, is subject to a defense that will not apply to the *270other plaintiffs, because she allegedly shared confidential NPC sales reports with a friend who was not employed by NPC. (D.Mem. 49.)
“[C]lass certification is inappropriate where a putative class representative is subject to unique defenses which threaten to become the focus of the litigation.” Gary Plastic Packaging Corp. v. Merrill Lynch, Pierce, Fenner & Smith, Inc., 903 F.2d 176, 180 (2d Cir.1990). “Regardless of whether the issue is framed in terms of the typicality of the representative’s claims, or the adequacy of its representation, there is a danger that absent class members will suffer if their representative is preoccupied with defenses unique to it.” Id. (citations omitted). However, “any allegations concerning the representative’s adequacy must be relevant to the claims in the litigation, such that the problems could become the focus of cross-examination and unique defenses at trial, to the detriment of the class.” German v. Fed. Home Loan Mortgage Corp., 168 F.R.D. 145, 154 (S.D.N.Y.1996) (internal citations omitted).
Defendants have not suggested any way in which their pursuit of the defense to which Corbett is allegedly subject — that she improperly disclosed corporate information to a friend in violation of NPC policy — could affect the interests of the other class members. “Plaintiffs’ testimony or credibility that is subject to attack must be on an issue critical to one of their causes of action.” Id. Improper information-sharing, one supposes, could provide defendants with a legitimate reason for adverse employment action, but of course defendants claim to have had a legitimate reason for all of the relevant adverse employment actions, so this can hardly be a basis for a finding of inadequate representation. Defendants do not suggest that Corbett’s credibility is adversely affected by their allegations.
If an actual conflict should arise, the Court is prepared to revisit the question, but at this point defendants have not even presented a hypothetical conflict for the Court to address. Accordingly, the Court finds that the named plaintiffs adequately represent the class, and that the prerequisites for class certification under Rule 23(a) have been satisfied.
F. Rule 23(b) Requirements
When the prerequisites of Rule 23(a) have been met, a class may be certified if it fits within one of the three categories in Rule 23(b). Fed.R.Civ.P. 23(b). “The requirement of ‘rigorous analysis’ to ensure ‘actual, not presumed conformance’ with Rule 23(a) applies with ‘equal force to all Rule 23 requirements, including those set forth in Rule 23(b)(3).’” In re Vivendi Universal, S.A., 242 F.R.D. 76 (S.D.N.Y.2007), quoting In re IPO, 471 F.3d at 33 & n. 3. Plaintiffs contend that the class should be certified under Rule 23(b)(2), which applies to classes in which “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,” or Rule 23(b)(3), which applies when “questions of law or fact common to the members of the class predominate over any questions affecting only individual members, and that a class action is superior to other available methods for the fair and efficient adjudication of the controversy.”
Certification under Rule 23(b)(2) is appropriate where “(1) the positive weight or value to the plaintiffs of the injunctive or declaratory relief sought is predominant even though compensatory or punitive damages are also claimed, and (2) class treatment would be efficient and manageable, thereby achieving an appreciable measure of judicial economy.” Robinson, 267 F.3d at 164 (internal alteration, citation and quotation marks omitted). When a case involves claims for both injunctive relief and non-incidental monetary damages, courts must assess “the relative importance of the remedies sought, given all of the facts and circumstances of the case.” Id., quoting Hoffman, 191 F.R.D. at 536.
Although the assessment of whether injunctive or declaratory relief predominates will require an ad hoc balancing that will vary from case to case, before allowing (b)(2) certification a district court should, at a minimum, satisfy itself of the following: (1) even in the absence of a possible *271monetary recovery, reasonable plaintiffs would bring the suit to obtain the injunctive or declaratory relief sought; and (2) the injunctive or declaratory relief sought would be both reasonably necessary and appropriate were the plaintiffs to succeed on the merits. Insignificant or sham requests for injunctive relief should not provide cover for (b)(2) certification of claims that are brought essentially for monetary recovery.
Id22
In this case, there can be little question that reasonable plaintiffs would sue to obtain the injunctive relief sought. The central goal of this lawsuit is to alter practices at NPC that plaintiffs believe are discriminatory. If plaintiffs prevail on the merits, that injunctive relief will be appropriate and reasonably necessary, because it would serve little purpose to award money damages for discrimination without addressing the institutional structure that perpetuates it. Defendants are alleged to have acted on grounds generally applicable to the class, and “[plaintiffs seek to reform defendants’ practices to provide for equitable employment opportunities and compensation for women.” Hnot I, 228 F.R.D. at 486.
Defendants’ argument against class certification is that plaintiffs have “fail[ed] to show the existence of any class-wide discriminatory practice in need of injunctive relief.” (D. Mem. 50.) Plaintiff have not yet been asked to prove any such thing. This is a class certification motion, not a trial, and if plaintiffs fail to prove the existence of class-wide discriminatory practices, no injunctive relief will be awarded. For now, the court need only determine whether such relief would be appropriate and necessary “were the plaintiffs to succeed on the merits,” Robinson, 267 F.3d at 164, and it clearly would be.
Thus, the requirements of Rule 23(b)(2) are satisfied and the class will be certified under that provision. Because the Court finds that plaintiffs meet the requirements of Rule 23(b)(2), it is unnecessary at this point to determine whether class certification under Rule 23(b)(3) would also be warranted. See Hnot I at 486. Additionally, Title VII civil rights cases may be divided into liability and remedial phases. Id. Therefore, the issue of whether plaintiffs meet the requirements of Rule 23(b)(3) is deferred until after the liability phase of this action.
CONCLUSION
Defendant Novartis Corporation’s motion for summary judgment (Doc. # 7) is granted.
As to the remaining defendants, plaintiffs have demonstrated that their proposed class and its representatives satisfy the requirements of Federal Rule of Civil Procedure 23(a) and 23(b). Accordingly, plaintiffs’ motion for class certification (Doc. #89) is granted. Amy Velez, Penni Zelinkoff, Minel Hider Tobertga, Michael Williams, Jennifer Waxman-Recht, Karen Liggins, Lori Horton, Holly Waters, Wendy Pinson, Roberta Vonlintel, Catherine White, Kelly Corbett, Jamie Holland, Joan Durkin, Simona Lopes, Maryanne Jacoby, and Marta Deyne are appointed as class representatives. The firm of Sanford, Wittels & Heisler, LLP, is appointed as lead counsel for the class.
SO ORDERED.
. Two of the nineteen named plaintiffs, Ashley Narmour and Sue Earl, sue only in their individual capacities, and not as putative class representatives. (P. Class Cert. Mem. 3 n. 1.)
. The joint-application policy makes eminent sense, given that many major corporations have a surfeit of affiliated subsidiaries. If the contest were structured otherwise, the entire "Best Places to Work” list could theoretically be filled with one corporation and its subsidiaries.
. This Opinion and Order adjudicates two motions, which were separately briefed. For the sake of brevity, the memoranda of law will be cited throughout the “Discussion” section of this opinion as “D. Mem.," "P. Mem.,” "D. Reply,” and so forth. It should be understood that such citations refer to the parties’ summary judgment memoranda when they appear in the discussion of summary judgment, and that they refer to the class certification memoranda when they appear in the discussion of class certification.
. Plaintiffs' contention that NPC has no "independent” directors adds nothing to this analysis. The testimony on which they rely establishes merely that the directors in question are not "independent directors” in the normal sense of that term, that is, directors not otherwise associated with the company they serve. (Merkelson Dep., Reply. Ex. 3, at 154.)
. Because Corporation’s motion for summary judgment is granted, there is no need to address its arguments against class certification, which primarily concern the lack of evidence implicating Corporation in any discrimination. For the remainder of the opinion, therefore, references to "defendants” should be understood to mean defendants NPC and Ebeling, who submitted joint briefs on the class certification question.
. There are two other kinds of claims that may be considered “reasonably related”: those alleging "retaliation by an employer against an employee for filing an EEOC charge," and those alleging "further incidents of discrimination carried out in precisely the same manner alleged in the EEOC charge.” Butts v. City of N.Y. Dep’t of Hous. Pres. & Dev., 990 F.2d 1397, 1401 (2d Cir.1993). Neither is at issue here.
. Defendants argue that Velez’s charge cannot support class claims because it was "never perfected” — meaning that the EEOC dismissed the charge on August 24, 2004 for failure to file a timely charge. (Schnadig Ltr. Ex. 3.) The EEOC, however, withdrew this determination upon Velez’s submission of evidence that the charge had in fact been timely filed, and the reason for dismissal was "altered to reflect an administrative dismissal of the charge.” (Sanford Ltr. Ex. B.) Thus, there is no basis in the record on which to conclude that Velez’s charge was not timely filed.
. Because defendants’ argument that the class period cannot start earlier than 2004 is without merit, it is unnecessary to address defendants' argument that Ledbetter v. Goodyear Tire & Rubber Co., - U.S. -, 127 S.Ct. 2162, 167 L.Ed.2d 982 (2007), precludes consideration of statistical data pre-dating the putative 2004 beginning of the class period (Schnadig Ltr. at 4-5).
. In re IPO discussed at length circumstances under which a class certification motion requires analysis of the merits of a case. Previously, the Supreme Court had held that "nothing in either the language or history of Rule 23 ... 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 v. Carlisle and Jacquelin, 417 U.S. 156, 177, 94 S.Ct. 2140, 40 L.Ed.2d 732 (1974). Noting that Eisen concerned a preliminary inquiry into the merits for purposes of assigning costs of notice, the Second Circuit in In re IPO held that this language does not apply to "cases where a merits inquiry either concerns a Rule 23 requirement or overlaps with such a requirement.” 471 F.3d at 34.
. Indeed, the number of declarations by potential class members complaining of disparate pay and promotion exceeds the 40-member level at which numerosity is presumed. See Consol. Rail Corp. v. Town of Hyde Park, 47 F.3d 473, 483 (2d Cir.1995).
. It is particularly unwise for the Court to become too deeply involved in an analysis of the mathematical and statistical merits of an expert report, rather than leaving such an analysis to the factfinder, because "[l]awyers and judges working with statistical evidence generally have only a partial understanding of the selection processes they seek to model, they often have incomplete or erroneous data, and are laboring in an alien and unfamiliar terrain.” Waisome v. Port Auth. of N.Y. & N.J., 948 F.2d 1370, 1372 (2d Cir.1991).
. As defendants note, plaintiffs do not appear to seek class certification on their claims of sexual harassment or discrimination in the administration of discipline. (D.Mem. 34.) Plaintiffs’ reply does not contest this assertion. Accordingly, there is no need to address those claims in the class certification context.
. Plaintiffs also argue that Watson v. Fort Worth Bank and Trust, 487 U.S. 977, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988), compels a finding that NPC “employs subjective practices which discriminate against women.” (P. Reply 10.) Plaintiffs appear to interpret Watson as holding that any evaluation system that combines subjective and objective elements must be treated as subjective, and therefore inherently impermissible. (Id.) This is a misreading of Watson, which held that evaluation systems based on subjective criteria were subject to disparate impact analysis, just as were evaluation systems based on objective criteria. 487 U.S. at 989-90, 108 S.Ct. 2111. The language on which plaintiffs rely addressed the question of whether subjective evaluation systems should be exempted from disparate impact law, not whether Title VII liability exists whenever an evaluation system is partially subjective. Id.
. The experts also appear to disagree about the standard for showing commonality. Welch criticizes Lanier for failing to show that the “estimated $79 difference is common across the women in the proposed class.” (Welch Rep. 13.) To show commonality, however, it is not necessary to find the same common difference in each group. In other words, plaintiffs need not show that they each suffer the same degree of pay disparity. The asserted common question is whether there was discrimination; the degree of damage presumably differs in most class-action discrimination cases.
. Welch's report examines the total number of promotions, broken down by gender, and finds that women are slightly more likely to receive a promotion. (Welch Rep. 32.) To this plaintiffs respond rather lamely that analyzing all promotions is unnecessary, because the unequal distribution across the corporate hierarchy is sufficient evidence of discrimination, and that "Dr. Lanier need not provide statistical evidence for every allegation in order to compel a finding of commonality.” (P. Reply 17.) While it is possible to imagine reasons why plaintiffs' sample might be a better one despite its small size — for example, a contention that NPC managers are more likely to interfere with a woman’s elevation from the ranks of the sales staff to a managerial post than they are to interfere with her promotion within the ranks — it is unnecessary to consider these questions, which pertain to the merits of the expert reports, at the class certification stage.
. Defendants also argue that Lanier's analysis fails to account for women's unwillingness to relocate, which is a prerequisite for management positions in many if not all cases. (D.Mem. 31.) Dr. Lanier examined questionnaires given to applicants for positions as sales representatives, in which 68.8 percent of men were willing to relocate, but only 43.5 percent of women. (Welch Rep. 37.) Plaintiffs respond that Welch’s analysis of willingness to relocate is flawed because his data comes from sales representatives, not managers. The question to which that data relates, however, is whether women in the pool of employees eligible for promotion are interested in management positions, and so it is not clear that a survey of managers would be more useful. Welch's alternative explanation is a matter to be considered by the factfinder.
. Defendants argue that "[n]othing” in this conclusion "provides any supporting evidence that NPC discriminates against women, as a class in compensation or on the basis of pregnancy” because Lanier did not examine "whether both men and women who take FMLA leave are similarly treated or whether women who take FMLA leave for pregnancy reasons are less favorably treated.” (D.Mem. 33.) Defendants conspicuously fail to offer such an analysis themselves; they offer no evidence that the comparison Welch seems to advocate — between men and women who take leave, and men and women who do not — would be any less suggestive of discrimination. In fact, Welch’s report appears to compare women to women when it concludes that base pay does not fall after leave. (Welch Rep. 44.) In any event, Lanier argues dial to include men in the analysis would make it impossible to determine whether FMLA leave — as opposed to gender differences themselves — was the cause of any disparity. (Lanier Rep. 23.) This methodological dispute is appropriately addressed to the factfinder; for present purposes it is clear tiiat Lanier’s findings raise a significant question as to whether leave causes a decline in women's pay.
. At least one court has held that an employee may make out a claim for sex discrimination by showing that a facially neutral evaluation system has a disparate impact on women who take leave. In Vosdingh v. Qwest Dex, Inc., No. Civ. 03-4284, 2005 WL 914732 (D.Minn. Apr. 21, 2005), a district court held that the plaintiff employees had shown that an evaluation system had a disparate impact on women who took leave, and accordingly applied the three-step framework of 42 U.S.C. § 2000e-2(k), shifting the burden to the employer to demonstrate a legitimate, business-related justification for the evaluation system, which the plaintiff could defeat by showing that an alternative approach — such as one normalizing the evaluation scores to compensate for absence — was comparably effective. Id. at *14-*16. See Smith, 196 F.3d at 365 (discussing burden-shifting framework for disparate impact cases).
In Vosdingh, however, employees were given a score of zero on certain categories in their performance evaluations for the time they were on leave — the same score they would have received had they been on the jobs but failing miserably. Effectively, their evaluations appeared to report poor performance for the time they were on leave, which much more closely resembles an affirmative sanction for taking leave. Here, the employee is paid the full and normal base salary after return from leave. Incentive-based payments are simply not adjusted to account for decreases in sales due to a representative's absence from the job. In effect, incentive pay is not awarded when it is not earned.
Nothing in the PDA requires NPC to adjust incentive payments in light of the nature of the leave that caused their decline. To hold otherwise would effectively imply that paying employees on an incentive basis is unlawful wherever women are more likely than men to take leaves that might adversely affect their productivity when they return, for reasons having nothing to do with their employer's actions.
. At least one court in this Circuit has stated, in the context of a summary judgment motion dealing with a challenges to an affirmative action program, that "[ajnecdotal evidence alone will not suffice to prove a systematic pattern of discrimination." Id. at *4. North Shore Concrete and Assoc., Inc. v. City of New York, No. 94 Civ. 4017, 1998 WL 273027, at *4 (E.D.N.Y. Apr. 12, 1998). That decision cited Coral Const. Co. v. King County, 941 F.2d 910, 919 (9th Cir.1991), which held that "[wjhile anecdotal evidence may suffice to prove individual claims of discrimina*266tion, rarely, if ever, can such evidence show a systemic pattern of discrimination necessary for the adoption of an affirmative action." (emphasis added). While anecdotal evidence alone may be insufficient to justify a government body's adoption of race-conscious policies, this does not mean it is insufficient to justify class certification, which merely requires that a common question be presented.
. The one declaration that appears not to present specific examples from personal experience is that of Ann Hogan. (P.Ex. 27.) Hogan alleges that “Novartis management” disciplined two unidentified female employees and the declarant, but not the other (also unidentified) employee in the group, who was male. The other two female employees, but not the declarant, had taken maternity leave. The manager is not identified, and it appears that the declarant is not entirely sure of the relative performance of the employees ("his sales performance was the same or worse than our sales performance”). Moreover, it appears that of the employees who did not take *267maternity leave, one was disciplined anyway. This declaration is notably unpersuasive on the question of pregnancy discrimination, compared to the other twenty-seven, but its unpersuasiveness is anomalous.
. Defendants also point out that one of the women claiming pregnancy discrimination, Minel Tobertga, adopted her child, rather than giving birth naturally. (D.Mem. 36.) The Court need not decide whether adoption is a permissible basis on which to make a claim of gender discrimination based on parental status under the Pregnancy Discrimination Act, because a finding of typicality would be appropriate even without Tobertga’s evidence.
. Although In re IPO modified other aspects of class certification analysis under Rule 23, "In re IPO is entirely inapposite to the Court’s determination under Rule 23(b)(2).” Hnot II, 241 F.R.D. at 211.