OPINION AND ORDER

JOHN G. KOELTL, District Judge.

Zaire Lamarr-Arruz1 and Mominna Ansoralli (collectively, the "plaintiffs") on behalf of a purported class have brought claims against the defendant, CVS Pharmacy, Inc. ("CVS" or the "Company"), for hostile work environment pursuant to 42 U.S.C. § 1981 as amended by the Civil Rights Act of 1991; the New York Human Rights Law, New York State Executive Law § 296 et seq. (the "NYSHRL"); and the New York City Human Rights Law, Administrative Code of the City of New York § 8-101 et seq. (the "NYCHRL").

In a companion Opinion and Order (the "Summary Judgment Opinion"), this Court denied CVS's motions pursuant to Rule 56 of the Federal Rules of Civil Procedure for summary judgment dismissing the plaintiffs' individual claims. Familiarity with the Summary Judgment Opinion is presumed.

The plaintiffs have also moved for certification of a class based on their hostile work environment claims pursuant to Rule 23 of the Federal Rules of Civil Procedure.2 The plaintiffs have proposed the following class definition:

Pls.' Cert. Reply Mem. 1.3 The proposed class period is in excess of four years in length.4

CVS has filed a separate motion to exclude the proposed testimony of the plaintiffs' expert witness, David L. Crawford.

For the following reasons, the plaintiffs' motion for certification of a class is denied and CVS's motion to exclude the proposed Crawford testimony is granted.

I.

The factual background of this case is set forth in the Summary Judgment Opinion. The following factual background relevant to the present motion is undisputed, unless otherwise noted.5

CVS is a Rhode Island corporation that owns and operates retail drugstores throughout the United States, including in New York City. CVS employs undercover security guards in plainclothes known as "Market Investigators" (also known as "Store Detectives"). Market Investigators work in the "Loss Prevention Department" within CVS and are supervised by at least Regional Loss Prevention Managers ("RLPMs"). Each RLPM oversees a team of Market Investigators within a territory of CVS stores. The territory of an RLPM may change over time. See CVS's Cert. Memo. Op. Ex. D (Saliu Dep. dated July 12, 2016) at 26; CVS's Cert. Memo. Op. Ex. E (Salvatore Dep. dated July 20) at 21-24. Although a Market Investigator reports to one RLPM at a time, a Market Investigator may work under more than one RLPM over the course of the Market Investigator's employment. During the relevant period, there were approximately 142 CVS stores in the New York City Area and twelve RLPMs, who each supervised a team of Market Investigators. See Gottlieb Decl. Ex. ZZ; CVS's Cert. Memo. Op. Ex. A (R.J. Gaites Decl. II) ¶¶ 5-6.

CVS also employs Store Managers, each of whom is responsible for the operations of an assigned "home store."

Market investigators work at multiple CVS stores (including multiple stores within a given workweek) and rotate among multiple CVS stores within a geographic area. Because Market Investigators work at many different CVS stores, they also work with many different Store Managers. The degree and frequency with which any individual Market Investigator interacts with any individual Store Manager varies depending on a variety of factors, including respective work schedules; indeed, a Market Investigator may work at a CVS store without encountering a Store Manager. Market Investigators generally work independently (in other words, without any other Market Investigators present), although they may also occasionally work in pairs or in groups depending on need.

Named-plaintiff Lamarr-Arruz is an African-American who worked as a Market Investigator from around February 18, 2013 to November 14, 2013. Named-plaintiff Ansoralli, who is of Spanish, Portuguese, and West Indian descent, and whose mother is from Guyana, worked as a Market Investigator from October 2012 to February 2013.

The gravamen of the plaintiffs' claims is that the conduct of the RLPMs and Store Managers creates a hostile work environment for Hispanic and black employees. CVS denies the claims.

In connection with this motion, the plaintiffs have submitted documentary evidence in the form of declarations, depositions, and pleadings from fourteen other Market Investigators who currently have filed individual hostile work environment actions that are pending against CVS. Together with the named plaintiffs, the sixteen Market Investigators claim to have worked under eight of the twelve RLPMs and at around 111 (78%) of the approximately 142 CVS stores in the New York City area during the class period. The Market Investigators were employed for different periods and lengths of time. The plaintiffs estimate that CVS employed at least 80 black or Hispanic Market Investigators (and likely many more) during the class period. See Gottlieb Reply Decl. Ex. KKKK.

The claims of the sixteen Market Investigators with respect to CVS are similar: over the course of their respective employments some combination of RLPMs and/or Store Managers instructed them to racially profile black and Hispanic customers, and also used language that was racially degrading. See, e.g., Gottlieb Decl. Ex. CCC (Sheldon Thomas Declaration) ¶¶ 4, 10; Gottlieb Decl. Ex. DDD (Lacole Simpson Deposition) at 145, 227-28, 250. The plaintiffs have also adduced evidence of complaints from CVS customers that they were the subjects of racial profiling and racially derogatory language at various stores.

Of the twelve RLPMs, there is only evidence that four ___ Abdul Saliu, Anthony Salvatore, Marian Lanzilotti, and Ron Taylor ___ used racially opprobrious language and instructed Market Investigators to profile based on race. See Dkt. 230. The plaintiffs concede that there are no allegations of racial discrimination with respect to one RLPM who managed at least some of the sixteen Market Investigators. The plaintiffs also concede that they have adduced no evidence with respect to the conduct of four of the RLPMs. Dkt. 230 at 1. And none of the sixteen Market Investigators fault RLPMs John Marton, Eric Butler, or Joe Burgess for any racial remarks or racial profiling. Dkt. 230.

With respect to the Store Managers, the plaintiffs claim that they have adduced evidence that 59 Store Managers across the 142 CVS stores in the New York City area committed at least one act of racial discrimination over the course of the class period. Dkt. 230. The Market Investigators had different experiences with different Store Managers. With respect to the 59 Store Managers, the evidence ranges from isolated incidents to repeated occurrences. Some incidents were discussed at team meetings with an RLPM and the other Market Investigators on that RLPM's team, while other incidents were witnessed by a single Market Investigator and not repeated to any others. In some instances, multiple Market Investigators fault the same Store Manager. In other instances, there is only one accusation by a single Market Investigator against a single Store Manager.

Market Investigators testified that some Store Managers used unambiguously racist language or explicitly directed Market Investigators to racially profile customers. See, e.g., Gottlieb Decl. Ex. DDD at 261-63. Other Store Managers used what the plaintiffs characterize as coded language that had purportedly racist connotations. Some Store Managers used Hispanic or black in a descriptive sense (i.e., "do you see that black person with the bags, or did you see that Hispanic person"). Gottlieb Decl. Ex. MMM at 63-64. Still other Store Managers directed Market Investigators to follow customers who were black or Hispanic on a disproportionately high basis as compared to customers who were Caucasian. See, e.g., Gottlieb Decl., Ex A (Lamarr-Arruz Dep. dated Mar. 22, 2016) at 247.

Market Investigators also testified that they worked with Store Managers who never did anything discriminatory and that they worked in CVS Stores where they encountered no racial discrimination. See, e.g., Lamarr-Arruz R. 56.1 Counterst. ¶¶ 35-37; Ansoralli R. 56.1 Counterst. ¶¶ 37-38. And there is no evidence with respect to the working conditions at 22% of the CVS stores.

In connection with their motion for class certification, the plaintiffs have submitted an expert report by Crawford (the "Crawford Report"). Gottlieb Decl. Ex. WW (The Crawford Report). Crawford performed a Mantel-Haeznel statistical test, which is a method for assessing the association between two variables. CVS Exclusion Mem. Op. Ex. A (Crawford Dep. dated Aug. 30, 2016) at 102-03. Crawford's analysis purports to show that CVS disproportionately contacted the police when the Company detained a Hispanic or black shoplifting suspect, as compared to when it detained a suspected Caucasian shoplifter. The theory is that this serves as a proxy to show that CVS racially profiles on a company-wide basis.

CVS has submitted the report of Janet R. Thornton, Ph.D. (the "Thornton Report"), which criticizes Crawford's methodology. See Thornton Decl. Thornton concluded that there are no statistically significant differences based on a shoplifter's race with respect to how often CVS alerts police officers.

II.

A.

Before certifying a class, the Court must determine that the party seeking certification has satisfied the four prerequisites of Rule 23(a): (1) numerosity, (2) commonality, (3) typicality, and (4) adequacy of representation. See Fed. R. Civ. P. 23(a); Teamsters Local 445 Freight Div. Pension Fund v. Bombardier, Inc., 546 F.3d 196, 201-02 (2d Cir. 2008); In re Initial Pub. Offerings Sec. Litig. ("In re IPO"), 471 F.3d 24, 32 (2d Cir. 2006). The Court must find, more specifically, that: "(1) the class is so numerous that joinder of all members is impracticable; (2) there are questions of law or fact common to the class; (3) the claims or defenses of the representative parties are typical of the claims or defenses of the class; and (4) the representative parties will fairly and adequately protect the interests of the class." Fed. R. Civ. P. 23(a). The Court must also find that the class qualifies under one of the three sets of criteria set forth in Rule 23(b). See Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 614 (1997); Teamsters, 546 F.3d at 203; In re IPO, 471 F.3d at 32.

Rule 23(a)(2) requires a showing that "there are questions of law or fact common to the class." Fed. R. Civ. P. 23(a)(2). To satisfy the commonality requirement, class members' claims must "depend upon a common contention," and the common contention "must be of such a nature that it is capable of classwide resolution." Wal-Mart Stores v. Dukes, 564 U.S. 338, 350 (2011). "Consideration of this requirement obligates a district court to determine whether plaintiffs have `suffered the same injury.'" Sykes v. Mel S. Harris & Assocs. LLC, 780 F.3d 70, 84 (2d Cir. 2015) (quoting Wal-Mart, 564 U.S. at 353)). "Commonality `does not mean that all issues must be identical as to each member, but it does require that plaintiffs identify some unifying thread among the members' claims that warrant[s] class treatment.'" Indergit v. Rite Aid Corp., 293 F.R.D. 632, 651 (S.D.N.Y. 2013) (quoting Damassia v. Duane Reade, Inc., 250 F.R.D. 152, 156 (S.D.N.Y. 2008)).

The plaintiffs here seek certification under Rule 23(b)(3), which provides for a class to be maintained where "the questions of law or fact common to the class members predominate over any questions affecting only individual members, and . . . a class action is superior to other available methods for fairly and efficiently adjudicating the controversy."6 Fed. R. Civ. P. 23(b)(3); see also Comcast Corp. v. Behrend, 133 S.Ct. 1426, 1432 (2013). If the requirements of 23(a) have been met, and the claims fall within the scope of Rule 23(b)(3), a court may, in its discretion, certify the class. See In re IPO, 471 F.3d at 41 ("[A] district judge may certify a class only after making determinations that each of the Rule 23 requirements has been met.").

"Rule 23 does not set forth a mere pleading standard. A party seeking class certification must affirmatively demonstrate his compliance with the Rule — that is, he must be prepared to prove that there are in fact sufficiently numerous parties, common questions of law or fact, etc." Wal-Mart, 564 U.S. at 350. Plaintiffs seeking class certification bear the burden of demonstrating by a preponderance of the evidence that the proposed class meets each of the requirements for class certification set forth in Rule 23. Teamsters, 546 F.3d at 202. When assessing whether plaintiffs have met this burden, courts must take into account "all of the relevant evidence admitted at the class certification stage." In re IPO, 471 F.3d at 42. A court may certify a class only after determining that "whatever underlying facts are relevant to a particular Rule 23 requirement have been established." Id. at 41. "[T]he obligation to make such determinations is not lessened by overlap between a Rule 23 requirement and a merits issue," although a court "should not assess any aspect of the merits unrelated to a Rule 23 requirement." Id.; see also Vincent v. Money Store, 304 F.R.D. 446, 452 (S.D.N.Y. 2015).

B.

This motion for class certification depends primarily on the ability of the class to satisfy Rule 23(a)(2)'s commonality requirement and Rule 23(b)(3)'s predominance and superiority requirements. Class certification in this case turns on the resolution of these issues. It is unnecessary to reach CVS's objections based on numerosity, typicality, adequacy, and ascertainability.7

The plaintiffs have moved to certify a class action for claims of hostile work environment on behalf of black and Hispanic Market Investigators pursuant to § 1981, NYSHRL, and NYCHRL. While there are some differences between what the plaintiffs must establish under § 1981 and the NYSHRL as compared to the NYCHRL, the plaintiffs must generally establish under all of the statutes, first, actionable conduct, and, second, a means of imputing liability to the Company. See, e.g., Mondesir v. N. Shore LIJ Health Sys., No. 14-cv-6496 (JGK), 2016 WL 6952254, at *2-3 (S.D.N.Y. Nov. 28, 2016) (discussing § 1981 and NYSHRL); Zakrzewska v. New Sch., 928 N.E.2d 1035, 1039 (N.Y. 2010) (discussing standard for imputing liability under the NYCHRL); Williams v. New York City Hous. Auth., 872 N.Y.S.2d 27, 39 (App. Div. 2009) (discussing the standard for proving actionable conduct under the NYCHRL).

As explained in greater detail in the Summary Judgment Opinion denying the defendant's motion for summary judgment against Lamarr-Arruz and Ansoralli, a hostile work environment based on race requires the plaintiff to prove that the harassment in the individual plaintiff's employment was sufficiently severe or pervasive to create an abusive working environment. Meritor Sav. Bank, FSB v. Vinson, 477 U.S. 57, 67 (1986). A plaintiff must show that the plaintiff subjectively perceived the environment to be abusive, but also that the environment was objectively hostile and abusive. Littlejohn v. City of New York, 795 F.3d 297, 321 (2d Cir. 2015). Whether the working environment was hostile depends upon what each of the plaintiffs actually experienced in the workplace.

Courts have generally approved of hostile work environment class actions where the work environment is localized, for example, where the proposed class members worked under a single supervisor or at a single site (even a large building that has multiple departments). See, e.g., Brown v. Nucor Corp., 576 F.3d 149, 157-58 (4th Cir. 2009), as amended (Oct. 8, 2009); Brand v. Comcast Corp., Inc., 302 F.R.D. 201, 218-19 (N.D. Ill. 2014); Cokely v. N.Y. Convention Ctr. Operating Corp., No. 00-cv-4637 (CBM), 2004 WL 1152531, at *1, *4-5 (S.D.N.Y. May 21, 2004); Brown v. Yellow Transp., Inc., No. 08-cv-5908, 2011 WL 1838741, at *1, *5 (N.D. Ill. May 11, 2011); see also Wal-Mart, 564 U.S. at 350 ("Quite obviously, the mere claim by employees of the same company that they have suffered a Title VII injury, or even a disparate-impact Title VII injury, gives no cause to believe that all their claims can productively be litigated at once. Their claims must depend upon a common contention — for example, the assertion of discriminatory bias on the part of the same supervisor.").

Employees that work in close proximity hear, see, and thus experience the same discriminatory acts. Where they do not do so first-hand, they may do so second-hand. See Schwapp v. Town of Avon, 118 F.3d 106, 111 (2d Cir. 1997) ("[T]he fact that a plaintiff learns second-hand of a racially derogatory comment or joke by a fellow employee or supervisor also can impact the work environment."). Shared work environments are echo chambers; employees gossip. While some employees may hear a racial slur first-hand, those employees can repeat the racial slur to the absent employees in the same workplace. Thus, all employees experience the hostile effect of the same racial slur. See Yellow Transp., 2011 WL 1838741, at *5 (finding that commonality was satisfied because "the Plaintiffs allege that various racially hostile incidents were witnessed first-hand by multiple people, and discussed and shared with many others"). The link to a single supervisor or site reduces individualized issues that differentiate the environments experienced by class members, and provides the "glue" that holds the class together. Wal-Mart, 564 U.S. at 352.

By contrast, courts have viewed multi-site hostile work environment class actions with skepticism. See Bolden v. Walsh Const. Co., 688 F.3d 893, 895 (7th Cir. 2012); Boyd v. Interstate Brands Corp., 256 F.R.D. 340, 364-66 (E.D.N.Y. 2009) (collecting cases). Where multiple sites are implicated, the individual differences between the spatially different work sites tend to predominate — even if a class could establish commonality. The gap can be bridged by targeting "centralized personnel" or a company-wide policy (for example, a policy that encouraged supervisors to use racial slurs in disciplining employees). See Wright v. Stern, No. 01-cv-4437 (DC), 2003 WL 21543539, at *6 (S.D.N.Y. July 9, 2003). But the gap cannot be bridged by simply alleging that a company has a policy against racial discrimination that leaves compliance to the discretion of its employees, which results in some employees engaging in acts that create a hostile work environment notwithstanding the policy. Wal-Mart, 564 U.S. at 355; Bolden, 688 F.3d at 898.

The decision by the Court of Appeal for the Seventh Circuit in Bolden is instructive. In that case, twelve African-American construction worker-plaintiffs sought to certify a hostile work environment class action and a disparate impact class action against a construction company on behalf of the company's African-American construction workers in the Chicago area that worked at each of the company's 262 sites. Id. at 895. The hostile work environment claim was predicated on racial epithets and graffiti at the worksites while the disparate impact claim was predicated on the failure to assign overtime at a proportionate rate. Id. at 895-96. Due to the nature of the construction industry, the plaintiffs worked at multiple sites for different lengths and periods of time. Many sites had different superintendents whose practices (and tolerance for the racism of others) differed and different sites had materially different working conditions. Id. A review of the record showed that the plaintiffs' allegations were concentrated against "a handful of superintendents and foremen. . . ." Id. at 896.

The Court of Appeals reversed the district court's grant of class certification. The Court of Appeals discussed the commonality and predominance problems that tend to arise in multi-site hostile work environment class action litigations: "The large number of sites, and the fact that plaintiffs' experiences differ, raise the question whether the classes satisfy Rule 23(a)(2). . . . To evaluate plaintiffs' grievances about [the defendant-company], however, a court would need site-specific, perhaps worker-specific, details, and then the individual questions would dominate the common questions (if, indeed, there turned out to be any common questions)." Id. at 896. The court found that the proposed hostile work environment class failed the test of commonality: "The 12 plaintiffs did not experience the working conditions at all 262 sites either individually or collectively, and a given plaintiff's bad experience with one of the five supervisors we have named does not present any question about the conduct of [the company's] many other superintendents and foremen." Id. at 898. Moreover, the court found that the class would not be manageable. Id.

The court also observed that certification of the hostile work environment claims posed greater challenges than certification of the disparate impact claims (which likewise could not be certified) because it was more difficult to establish a classwide policy that created a shared hostile work environment across multiple sites through anecdotal evidence or the type of statistical evidence that is usually used to establish disparate impact claims on a classwide basis. See id. at 896-98.

In this case, as in Bolden, the plaintiffs have failed to establish that there are common questions of law or fact that can be answered on a class-wide basis that cover all of the members of the purported class who worked for all of the RLPMs over an extended period of time. This is particularly true when they have only adduced evidence of racially hostile remarks by four of the twelve RLPMs and the interactions between potential class members and those RLPMs necessarily varied.

C.

The plaintiffs have also failed to establish the predominance and superiority necessary to certify a class under Rule 23(b)(3).

(i)

To satisfy the predominance requirement of Rule 23(b)(3), the issues subject to generalized proof and applicable to the class as a whole must predominate over, and be more substantial than, the issues that are subject to individualized proof. See Mazzei v. Money Store, 829 F.3d 260, 272 (2d Cir. 2016). To determine whether the plaintiffs have met their burden on predominance, a court "must assess (1) the `elements of the claims and defenses to be litigated'; and (2) `whether generalized evidence could be offered to prove those elements on a class-wide basis or whether individualized proof will be needed to establish each class member's entitlement to relief." Johnson v. Nextel Commc'ns Inc., 780 F.3d 128, 138 (2d Cir. 2015) (citation omitted). "Predominance requires a further inquiry, however, into whether the common issues can profitably be tried on a classwide basis, or whether they will be overwhelmed by individual issues." Id. Rule 23(b)(3) imposes a "far more demanding" criterion than Rule 23(a) (2). Amchem Prods., 521 U.S. at 624.

In this case, common issues do not predominate over individual issues with respect to the first element of the hostile work environment claim. The determination of actionable conduct is not subject to generalizable proof. Here, Bolden is on point.

The plaintiffs argue that a class action is appropriate — even though the Market Investigators worked independently at various CVS locations with different Store Managers throughout New York City for different periods and lengths of time under the supervision of different RLPMs — because all of the class members shared a common hostile work environment created by the RLPMs. But that is not true. The Market Investigators reported to twelve different RLPMs within the Loss Prevention Department. The main thrust of this action was that the Market Investigators all received the same instruction from their RLPMs that they were obliged to follow if they wanted to keep their jobs: target black and Hispanic customers. The RLPMs allegedly accompanied those instructions with opprobrious epithets. That was allegedly actionable conduct, which was the centrifugal force for this class action.

But the evidence shows the majority of the RLPMs did not instruct any Market Investigators to profile based on race or use any racially degrading language. There is no evidence that eight (75%) of the RLPMs instructed Market Investigators to racially profile and almost no evidence to suggest that those same eight RLPMs engaged in any actionable conduct.

The plaintiffs have adduced no evidence with respect to four of the twelve RLPMs, and concede that they have no evidence of actionable conduct against a fifth. None of the sixteen Market Investigators who provided evidence fault RLPMs Marton, Butler, or Burgess for any acts of racial discrimination — the best the plaintiffs can do against these three RLPMs is to argue that (in the plaintiffs' view) they each testified that they did not adequately investigate isolated complaints of racism and, in addition, that Burgess occasionally used coarse language that was unrelated to race. The argument that these three RLPMs contributed to a hostile work environment is not plausible.

That leaves evidence of direct acts of racial discrimination by four RLPMs: Lanzilotti, Taylor, Saliu, and Salvatore. And the evidence of actionable conduct against those four RLPMs differs with respect to severity and pervasiveness.

The remaining allegations of actionable conduct are store-specific. There were 142 CVS stores. The plaintiffs claim that the sixteen Market Investigators worked at about 112 stores. There was little temporal or spatial overlap between Market Investigators. There are no complaints of racial discrimination with respect to many of the stores and Store Managers. For the stores where there are complaints, the content and nature of the evidence of actionable conduct would differ for each class member. The Market Investigators complained about different Store Managers because they worked with different Store Managers at different stores at different times. The frequency with which a Market Investigator interacted with a Store Manager varied. Because Market Investigators did not generally work together, their accounts of the alleged discriminatory acts by even the same Store Managers differed. Their experiences did not overlap because their work environments were different.

In many instances, the plaintiffs claim that a reasonable jury could conclude the coded language by Store Managers contributed to a hostile work environment. That might be true for each individual class member, but would require contextual analysis that is not susceptible to proof on a classwide basis.

The ultimate result is individualized inquiries to determine liability, and then to determine damages. The Market Investigators that worked under Saliu, Salvatore, Lanzilotti and Taylor, respectively, would present four cases about the actionable conduct of each respective RLPM. The remainder of the class who worked under the other eight RLPMs would, in practical terms, have to establish that their interactions, principally with their Store Managers, created a hostile work environment. The individual issues with respect to proving actionable conduct overwhelm the common issues.8

Moreover, there are other individual issues with respect to imputing liability to the Company that contribute to the conclusion that the common questions do not predominate over the individual questions. For example, if a jury concluded that the Store Managers are not supervisors, the jury may have to make individualized findings with respect to CVS's negligence with respect to each Store Manager.

(ii)

To satisfy Rule 23(b)(3), the plaintiffs must also show "that a class action is superior to other available methods for fairly and efficiently adjudicating the controversy." A court must consider four nonexclusive factors: (1) the interest of the class members in maintaining separate actions; (2) the extent and nature of any litigation concerning the controversy already commenced by or against members of the class; (3) the desirability or undesirability of concentrating the litigation of the claims in the particular forum; and (4) the difficulties likely to be encountered in the management of a class action. Fed. R. Civ. P. 23(b) (3).

It is apparent that a class action is not superior to individual actions.

A class action would be unmanageable. The Court of Appeals for the Seventh Circuit observed in Bolden that the hostile work environment class in that case "would require at least one trial per site (to ascertain site-specific conditions) and perhaps one trial per week or month per site (because construction crews are constantly changing, and workers on site while concrete was being poured may have encountered working conditions different from those that prevailed when cabinet work was being installed)." Bolden, 688 F.3d at 898-99. The work schedules of the Market Investigators and Store Managers present analogous concerns in this case. The Market Investigators moved among different stores for different periods of time and experienced different Store Managers under different RLPMs, and there is no evidence that most of the RLPMs exhibited any racial discrimination. The requirement of individualized inquiries as to the conditions that each Market Investigator experienced at each store during different periods of time defeats the salutary benefits of the class action mechanism. And the plaintiffs have not proposed a way to make a class action manageable. See id. at 899 (observing that class definition alternatives may be conceivable but could encounter other problems under Rule 23(a) and Rule 23(b)).

The other Rule 23(b)(3) factors also weigh against a class action. The fourteen additional Market Investigators — one of whom could not meet the proposed class definition because he is not black or Hispanic — have commenced individual actions against CVS arising out of similar allegations. The actions are currently pending in various state, federal, and arbitral fora. The actions are at various stages, and many have been pending for over a year. Plainly, "these are not cases where it makes no economic sense for an individual to pursue his own claim." Oakley v. Verizon Commc'ns Inc., No. 09-cv-9175 (CM), 2012 WL 335657, at *18 (S.D.N.Y. Feb. 1, 2012); see also In re Bayou Hedge Fund Inv. Litig., 248 F.R.D. 404, 405 (S.D.N.Y. 2008) ("While judicial economy will generally result from the avoidance of multiple actions, in this case we already have multiple actions — eight of them, to be exact. As those arbitrations have been proceeding for well over a year — some may even have gone to hearing since this motion was filed — certifying a class is unlikely to produce judicial economy by reducing the number of pending matters."). The proposed class is not so numerous that a class action would be superior to individual actions, especially given that a class action in this case would simply devolve into individual trials on the merits for each class member.

Accordingly, superiority is not met.

Because the plaintiffs have failed to satisfy the commonality requirement of Rule 23(a)(2) and the requirements of Rule 23(b)(3), the motion for class action certification is denied.

III.

CVS has also moved to exclude the proposed Crawford testimony. That testimony does not affect the resolution of this class action motion and, in any event, is unreliable and should be excluded.

The Crawford testimony does not obviate the problems with commonality, predominance, and superiority discussed above. The testimony does not exclude the need for individualized assessments of the alleged racially hostile environment experienced by the individual members of the class. Evidence that CVS disproportionately contacts the police for black and Hispanic detainees at the aggregate level across all CVS stores would not change the need for individualized inquiries to evaluate each class member's experience and environment at the store level.

The Crawford testimony purported to show that a disproportionate number of blacks and Hispanics who were detained at CVS stores were reported to the police. But that testimony is removed from the issues in this case with respect to whether class members experienced a racially hostile environment. Whether a disproportionate number of blacks or Hispanics were reported to the police after they were detained does not even disclose whether a disproportionate number of blacks or Hispanics were initially detained by Market Investigators — the gist of the plaintiffs' complaint of racial profiling — nor does it disclose whether Market Investigators were subject to a racially hostile environment.

In any event, Crawford's proposed testimony should also be excluded because it is unreliable. Federal Rule of Evidence 702 requires a trial judge to ensure that all expert testimony is both relevant and reliable. See Daubert v. Merrel Dow Pharm., Inc., 509 U.S. 579, 589 (1993). In order for an expert's opinion to be reliable, the trial court must ensure that (1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the expert has reliably applied the principles and methods to the facts of the case. Fed. R. Evid. 702; see also Macaluso v. Herman Miller, Inc., No. 01-cv-11496 (JGK), 2005 WL 563169, at *8 (S.D.N.Y. Mar. 10, 2005).

A party may raise a Daubert challenge to proposed expert opinion testimony at the class certification stage. See, e.g., Scott v. Chipotle Mexican Grill, Inc., 315 F.R.D. 33, 55 (S.D.N.Y. 2016); see also Wal-Mart, 564 U.S. at 354 ("[T]he district court concluded that Daubert did not apply to expert testimony at the certification stage of class-action proceedings. We doubt that is so. . . ." (citation omitted)). "When a motion to exclude expert testimony is made at the class certification stage, the Daubert standard applies, but the inquiry is limited to whether or not the expert reports are admissible to establish the requirements of Rule 23." Fort Worth Employees' Ret. Fund v. J.P. Morgan Chase & Co., 301 F.R.D. 116, 126 (S.D.N.Y. 2014) (quoting Ge Dandong v. Pinnacle Performance Ltd., No. 10-cv-8086 (JMF), 2013 WL 5658790, at *13 (S.D.N.Y. Oct. 17, 2013)).

Crawford based his analysis on two CVS internal records that contained external case theft incident report data on shoplifting at CVS stores in the New York City Area. Crawford Report ¶ 12. The incident reports contained data related to the identification of the shoplifter; the shoplifter's race; the date of the alleged offense; the store; whether the police were contacted; whether the incident involved a weapon or violence; whether the shoplifter was a juvenile; whether Organized Retail Crime was involved; the category of the product(s) involved; the dollar amount of the items targeted; and a description of the incident.

Crawford removed all instances where a shoplifter was not detained (for example, because the shoplifter escaped before detention) from the data universe, and then performed a Mantel-Haenszel statistical test based on the reported race of the suspect and whether the police were notified, while controlling for the date of the offense and the CVS store where the detention occurred. On this basis, Crawford found that police contacts were disproportionately high for alleged black and Hispanic shoplifters across all CVS stores. Crawford also found that police contacts were disproportionately high for detentions involving alleged black and Hispanic shoplifters regardless of whether the value of the products targeted was above or below $50.

To be reliable, a data analysis must account for major variables, including confounding variables. See Wills v. Amerada Hess Corp., 379 F.3d 32, 50 (2d Cir. 2004) (Sotomayor, J.); Bickerstaff v. Vassar Coll., 196 F.3d 435, 449-50 (2d Cir. 1999), as amended on denial of reh'g (Dec. 22, 1999); Reed Const. Data Inc. v. McGraw-Hill Companies, Inc., 49 F.Supp.3d 385, 403 (S.D.N.Y. 2014) ("Omitted-variable problems — as the name suggests — arise when important control variables are left out of the model."), aff'd, 638 F. App'x 43 (2d Cir. 2016).

While courts have found that the Mantel-Haenszel test can be an appropriate way to analyze data in discrimination cases (typically disparate impact cases, which this case is not) if applied correctly, see, e.g., Oliver v. VSoft Corp., No. 1:09-cv-0185 (CAP) (WEJ), 2010 WL 11505776, at *10-11 & n.25 (N.D. Ga. Feb. 2, 2010), report and recommendation adopted, No. 1:09-cv-185 (CAP) (WEJ), 2010 WL 11506873 (N.D. Ga. Feb. 23, 2010), there are numerous problems with the way Crawford applied that methodology here that make his proposed testimony unreliable. Most significantly, Crawford did not attempt to control for variables other than race that could explain alternative reasons for the police contacts. See Bolden, 688 F.3d at 896.

Crawford failed to analyze the effect of potential confounding variables that were contained in the external case theft incident reports, including the use of weapons or violence in the shoplifting; whether the detainee was a juvenile; whether Organized Retail Crime was involved; and the category of the product(s) involved.

Crawford did not offer a sound reason for excluding these data points, which were major variables. Crawford testified that he simply did not investigate the relationship of these data points to race or police contacts. See, e.g., CVS Exclusion Mem. Op. Ex. A (Crawford Dep. dated Aug. 30, 2016) at 106-07, 135, 139-40, 173. While the plaintiffs contend that these factors were not material, "it is impossible for the Court to agree with [the] assumption that these factors are `unimportant' to an analysis." Bonton v. City of New York, No. 03-cv-2833 (SAS), 2004 WL 2453603, at *3 (S.D.N.Y. Nov. 3, 2004). Even Crawford conceded that "[i]t's certainly reasonable to expect that the police would be notified" if a shoplifter used a weapon or violence. CVS Exclusion Mem. Op. Ex. A (Crawford Dep. dated Aug. 30, 2016) at 142. Nevertheless, he testified that he did not investigate whether the use of weapons or violence had any correlation with police contacts. CVS Exclusion Mem. Op. Ex. A (Crawford Dep. dated Aug. 30, 2016) at 142.

As Crawford conceded with respect to weapons and violence, it is plain that other data points could have relevance to the decision to contact the police. Moreover, there were good reasons based on the record to believe that they were important factors the relevance of which needed to be at least investigated. CVS policy required its personnel to "contact law enforcement for assistance" in the event a juvenile could not be released to a legal guardian because of liability issues. Gaites Decl. Ex. 1 (CVS Market Investigator Training Presentation dated Mar. 2011) at CVS 00002917. Some CVS stores had policies that they would notify the police whenever certain products, such as cosmetics, were targeted by a shoplifter. CVS Exclusion Mem. Op. Ex. E (David Gonzalez Dep. dated July 21, 2016) at 161-62. Likewise, CVS personnel testified that police were consistently contacted in the event of Organized Retail Crime. CVS Exclusion Mem. Op. Ex. F (Eric Butler Dep. dated July 7, 2016) at 89; CVS Exclusion Mem. Op. Ex. G (Marian Lanzilotti Dep. dated July 7, 2016) at 69-72.

Crawford conceded at his deposition that he could have performed an analysis to control for these other variables. Crawford Dep. at 104-05. Crawford's failure to investigate the impact of these major variables, let alone control for them, renders his proposed testimony unreliable.9

Accordingly, the motion to exclude the proposed Crawford testimony is granted.

CONCLUSION

The Court has considered all of the arguments raised by the parties. To the extent not specifically addressed above, the arguments are either moot or without merit. The plaintiffs' motion for certification of a class is denied. CVS's motion to exclude the proposed Crawford testimony is granted. The Clerk is directed to close all pending motions.

SO ORDERED.