Davis v. Cintas Corporation

717 F.3d 476, 85 Fed. R. Serv. 3d 968, 2013 WL 2343302, 2013 U.S. App. LEXIS 10856, 96 Empl. Prac. Dec. (CCH) 44,839, 118 Fair Empl. Prac. Cas. (BNA) 903

• Court: Court of Appeals for the Sixth Circuit
• Decided: May 30, 2013
• Docket: 10-1662
• Status: Published

Opinion

OPINION

BOGGS, Circuit Judge.

Tanesha Davis sued Cintas Corporation, individually and on behalf of a class of female job applicants denied employment as entry-level sales representatives. She alleged that Cintas’s hiring practices led to gender discrimination, in violation of Title VII, and caused Cintas to reject her application for employment twice. The district court denied Davis’s motion for class certi fication, and ultimately granted summary judgment for Cintas on her individual claims. Davis appeals. We affirm the district court’s denial of class certification. We also affirm its grant of summary judgment on her individual disparate-treatment claim arising in 2004 and her disparate-impact claim. However, we reverse the district court’s grant of summary judgment for Cintas on Davis’s disparate-treatment claim arising in 2003, and remand for further proceedings consistent with this opinion.

I

According to Cintas’s promotional materials, service sales representatives are “[t]he # 1 link between our customers and our company operations. The Service Sales Representative is Cintas.” More specifically, the position involves delivering and selling Cintas’s wares—“corporate identity uniforms and value-added products”—and providing direct customer service.

Although Cintas is a large corporation with many locations nationwide, its corporate policy is “to run one business in many different cities instead of running many different businesses in different cities.” “[A]ll Cintas operations use the same terminology, use the same forms and paperwork, and ‘run their stores’ the same way.” Hiring for the service-sales-representative position is no exception. All Cintas locations use—and used at all times relevant to this claim—the “Meticulous Hiring System.” This policy provides: “No person will be hired in the company until [Cintas has] identified the traits and competencies needed to be successful in the position. The traits and competencies must be documented in a list of ‘must have’ and ‘preferred’ hiring standards for each position.”

For the service-sales-representative position, “Must Haves” include both objective and subjective elements. Objectively, applicants must achieve a minimum score on a standardized test given to all service-sales-representative applicants, have a driver’s license and a GED or high-school diploma, and be able to lift forty pounds. Subjectively, applicants must demonstrate customer orientation, sales orientation, integrity, dependability, achievement orientation, flexibility, stress tolerance, openness to differences, tenacity, initiative, persuasiveness, professionalism, compensation compatibility, and a stable employment history. There are also both objective and subjective “Preferreds.” Objectively, Cintas would like candidates to achieve a score between twenty-one and twenty-seven on the Wonderlic-WPT test (a type of intelligence test); subjectively, it seeks candidates who have successful sales experience, successful customer-service experience, and the ability to work without supervision.

Cintas uses a sixteen-step process to determine whether candidates meet these criteria. After Cintas decides that a location needs a service sales representative and posts the job, managers screen applications and resumes. Next, managers conduct screening interviews, using a guide that contains pre-scripted questions, and invite qualified applicants to visit a Cintas location and take pre-employment tests. Managers then collect the applicant’s application materials, and administer the Wonderlic-WPT test and the “ePredix SSR Test.” If an applicant scores well enough on these examinations, she receives a “1st In-depth Interview using [Cintas’s service-sales-representative] 1st In-depth Interview Guide.” After this first interview, qualified applicants go on a “route ride,” and Cintas “[c]olleet[s] paperwork,” including tax forms and driving records. If the applicant completes her “route ride” successfully, she receives a “2nd In-depth Interview using [Cintas’s service-sales-representative] 2nd In-depth Interview Guide.” The applicant’s last step is a final interview, often with a Location’s general manager. Management then confers with everyone involved in the interview process, performs a criminal background check, a drug screening, a driving-record and credit-record check, contacts the applicant’s references, and finally determines whether to offer the candidate a job.

Although this process has well-defined steps, set as a matter of corporate policy, individual managers at different locations ultimately make the hiring decisions, based on local needs. Cintas’s national hiring profile states that “[additional Preferreds can be added to accommodate the needs of the Division, Group, or Location.” Some locations, for instance, emphasized sales experience over customer-service experience because of intense market competition, while others preferred the inverse because difficult economic conditions made keeping existing customers crucial.

Service sales representatives were historically male. From June 1999 to October 2006, more than ninety percent of the managers charged with hiring service sales representatives were male. This overwhelmingly male group overwhelmingly hired males. After Cintas implemented the Meticulous Hiring System in 2003, however, female hiring rose dramatically. Between 1999 and 2002, the percentage of women hired for the service-sales-representative position never rose above seven percent. In 2003, the year corporate-level management instructed other managers to “put the myth that females cannot be SSRs out of [their] mind and hire more women SSRs,” and implemented the Meticulous Hiring System, that number rose to 7.8 percent. In 2004, it rose to 10.9 percent, and in 2005 it rose to 20.8 percent.

Anecdotal accounts support the data suggesting that Cintas managers saw the service-sales-representative position as a man’s position. According to a former manager, who was male, other managers at one Cintas location opined that women could not handle the responsibilities of a service sales representative. Sharon Punch-Johnson, a female applicant who was ultimately rejected, averred that a manager asked whether her husband would, be comfortable with his wife working predominantly with males. In her deposition, Kristi Clement Williams, another female applicant, claimed that a Cintas manager suggested that females might not be comfortable in the service-sales-representative position because the job required going into men’s locker rooms, and depended on interactions with customers in “a predominantly male environment.” Still another female applicant claimed in a declaration that a Cintas manager bluntly told her: “Cintas preferred to hire men in [the service-sales-representative] position ].”

Tanesha Davis, then a store manager for LensCrafters, first applied for a service-sales-representative position at Cintas’s Franklin, Wisconsin location, Location 447, in 2003. Human-Resources Manager Christine Richards conducted a screening interview and took notes. Although she does not now recall the interview, Richards’s contemporaneous notes suggest that she rejected Davis at the screening stage because Davis said that she disliked having to sell products that she considered overpriced, and that she wanted to continue working another job part-time. Richards, of course, is adamant that she did not reject Davis because of her race or gender. Rather, Richards claims, Davis was simply not the best-qualified applicant for Location 447’s available service-sales-representative position. Approximately three months before Davis applied, Location 447 had hired another female service sales representative. It hired one male service sales representative just before Davis applied, and another two while her application was pending.

Davis applied again one year later. This time, she advanced further into the hiring process. She passed her initial screening interview, even though she expressed concerns about working in bad weather, and achieved the second highest score recorded during 2003 and 2004 on one of Cintas’s standardized tests. Matt Presendofer, the manager who observed Davis’s route ride, reported that “she did a lot of things well out on the route.” Presendofer even opined that, “from a customer standpoint ... and from a sales standpoint ] [s]he had all the ... attributes we wanted for a SSR candidate.” He expressed concerns, however, about her level of physical energy and her efficiency. As in 2003, Cintas ultimately chose not to hire Davis. Eight days after Davis’s route ride in 2004, Location 447 hired another woman. While her 2004 application was pending Cintas hired three male service sales representatives, and hired a fourth man within two months after rejecting Davis’s application.

This lawsuit, though, did not start with Davis. It began instead in January 2004 in the United States District Court for the Northern District of California, where a group of Cintas employees, which did not include Davis, filed a civil-rights class-action lawsuit under 42 U.S.C. § 1981 and the California Unfair Business Practices Act. The plaintiffs amended their complaint five months later, adding a Title VII claim, a claim under the California Fair Employment and Housing Act, and two additional plaintiffs. The parties then agreed to transfer the California action to the United States District Court for the Eastern District of Michigan, where another hiring-discrimination case was pending against Cintas, and to consolidate the two eases for pretrial proceedings.

■'Cintas next moved (1) for summary judgment on the individual claims of one of the California plaintiffs, and (2) to limit the Michigan case’s purportedly nationwide class to the State of Michigan. Counsel in the California case sought leave to add Davis and another woman, Cindy Patterson, as named plaintiffs.

WTiile these motions were, pending, plaintiffs in both cases made a joint motion to certify a class, consisting of “all females who unsuccessfully applied for the SSR job at any time on or after June 12, 1999.” They requested certification for claims involving declaratory and equitable relief under Federal Rule of Civil Procedure 23(b)(2) and certification for their class-wide punitive-damages claim under Rule 23(b)(3). Before resolving the class-certification motion, the district court granted leave to add Davis and Patterson as plaintiffs, granted summary judgment for Cintas on the California plaintiffs claim, and limited the Michigan claims to Michigan.

Ultimately, the district court denied class certification. Serrano v. Cintas Corp., Nos. 04-40132, 06-12311, 2009 WL 910702, at *1 (E.D.Mich. Mar. 31, 2009). It reasoned that the putative class did not satisfy Rules 23(a)(2), 23(a)(3), 23(a)(4), 23(b)(2), or 23(b)(3), because the case involved a “hiring process ... conducted by thousands of Cintas managers at hundreds of Cintas facilities,” id. at *5, there was a “conflict between the interests of the named and unnamed class members,” id. at *9, and. “the damages sought by Plaintiffs would require individualized determinations inappropriate for a [Rule 23(b)(2) ] class action.” Id. at *10. The court denied a Rule 23(f) petition for interlocutory appeal. All remaining plaintiffs, except for Davis, dismissed their individual claims and all class claims, except for the gender-discrimination claim before us. Davis maintained her individual employment-discrimination claim. Cintas, however, moved for—and was granted—summary judgment on that issue. Davis appeals, individually and on behalf of the gender-discrimination class.

II

A

Federal Rule of Civil Procedure 23 allows a plaintiff to pursue claims on behalf of a class of similarly situated individuals if she demonstrates that she is “part of the class and ‘possesses] the same interest and suffer[s] the same injury’ as the class members.” East Tex. Motor Freight Sys., Inc. v. Rodriguez, 431 U.S. 395, 403, 97 S.Ct. 1891, 52 L.Ed.2d 453 (1977). To justify this “exception to the usual rule that litigation is conducted by and on behalf of the individual named parties only,” Califano v. Yamasaki, 442 U.S. 682, 700-01, 99 S.Ct. 2545, 61 L.Ed.2d 176 (1979), the plaintiff must meet a series of conditions laid out in Rule 23 to ensure, first, that she is an appropriate representative for absent class members, and, second, that her claim is appropriate for classwide resolution. See Fed.R.Civ.P. 23.

Class-certification litigation is the process of determining whether a plaintiff can meet these conditions. The analysis proceeds in two steps. Rule 23(a) is “the starting gate.” Wal-Mart Stores, Inc. v. Dukes, — U.S.-, 131 S.Ct. 2541, 2562, 180 L.Ed.2d 374 (2011) (Ginsburg, J., concurring in part and dissenting in part). It establishes four requirements:

(1) the class [must be] 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)(l)-(4). 1 If the plaintiff does not satisfy each of these requirements, her class claim fails at the threshold. If, however, the plaintiff shows that she is an appropriate representative within the meaning of Rule 23(a), the focus shifts to the case itself.

Under Rule 23(b), four types of lawsuits may proceed as class actions. Specifically, class resolution is appropriate when:

(1) prosecuting separate actions by or against individual class members would create a risk of:

(A) inconsistent or varying adjudications with respect to individual class members that would establish incompatible standards of conduct for the party opposing the class; or

(B) adjudications with respect to individual class members that, as a practical matter, would be dispositive of the interests of the other members not parties to the individual adjudications or would substantially impair or impede their ability to protect their interests;

(2) the party opposing the class has acted or refused to act on grounds that apply generally to the class, so that final injunctive relief or corresponding deelar atory relief is appropriate respecting the class as a whole; or

(3) the court finds that the questions of law or fact common to class members predominate over any questions affecting only individual members, and that a class action is superior to other available methods for fairly and efficiently adjudicating the controversy.

Fed.R.Civ.P. 23(b)(l)-(3). If the plaintiffs claim does not fall into one of these categories, class certification is inappropriate, even if the plaintiff meets each of Rule 23(a)’s four threshold requirements.

In sum, then, a plaintiff must show that she meets all four Rule 23(a) criteria, and that her ease falls into at least one of the four Rule 23(b) categories. If she fails to satisfy any of these requirements, class certification is not appropriate.

The district court must conduct “a rigorous analysis,” Gen. Tel. Co. of the Sw. v. Falcon, 457 U.S. 147, 161, 102 S.Ct. 2364, 72 L.Ed.2d 740 (1982), at “an early practicable time ... [to] determine by order whether to certify the action as a class action.” Fed.R.Civ.P. 23(c)(1)(A). “Frequently that ‘rigorous analysis’ will entail some overlap with the merits of the plaintiffs underlying claim. That cannot be helped.” Dukes, 131 S.Ct. at 2551. Thus, in the class-certification context, courts are permitted to “probe behind the pleadings,” Falcon, 457 U.S. at 160, 102 S.Ct. 2364, and “touch[ ] aspects of the merits.” Dukes, 131 S.Ct. at 2552.

Because “[a] district court’s class-certification decision calls for an exercise of judgment,” our review is narrow. Pilgrim v. Universal Health Card, LLC, 660 F.3d 943, 946 (6th Cir.2011). We reverse only if the district court abused its discretion or applied an erroneous legal standard. Ibid.

B

“Title VII ... contains no special authorization for class suits maintained by private parties. [Therefore,] [a]n individual litigant seeking to maintain a class action under Title VII must meet the prerequisites of numerosity, commonality, typicality, and adequacy of representation specified in Rule 23(a).” Falcon, 457 U.S. at 156, 102 S.Ct. 2364. In 2011, the Supreme Court addressed this particular type of class certification in Dukes. Accordingly, our first task is to determine precisely what impact that case has on Davis’s bid for class certification.

Dukes, like this case, involved allegations of gender discrimination. The plaintiffs alleged that Wal-Mart systematically discriminated against women in pay and promotion decisions. Dukes, 131 S.Ct. at 2547. Those “decisions at Wal-Mart [were] generally committed to local managers’ broad discretion,” although, particularly for promotion decisions, some objective requirements did apply. Ibid.

The plaintiffs brought a class-action lawsuit on behalf of 1.5 million women, under Title VII. Ibid.

Importantly for [the Court’s] purposes, [the plaintiffs] claim[ed] that the discrimination to which they [had] been subjected [was] common to all WalMart’s female employees. The basic theory of their case [was] that a strong and uniform “corporate culture” permitted] bias against women to infect, perhaps subconsciously, the discretionary decisionmaking of each one of WalMart’s thousands of managers—thereby making every woman at the company the victim of one common discriminatory practice. [The plaintiffs] therefore wish[ed] to litigate the Title VII claims of all female employees at Wal-Mart’s stores in a nationwide class action.

Id. at 2548. They sought “injunctive and declaratory relief, punitive damages, and backpay,” but not compensatory damages. Ibid.

For a 5-4 majority, Justice Scalia held that the case could not proceed as a class action because the plaintiffs could not establish commonality, within the meaning of Rule 23(a)(2). He began by noting that, to have suffered a violation of the same provision of law—Title VII, for instance—is not necessarily to have suffered the same injury. To satisfy commonality, Justice Scalia reasoned, a putative class representative’s

claims must depend upon a common contention—for example, the assertion of discriminatory bias on the part of the same supervisor. That common contention, moreover, must be of such a nature that it is capable of classwide resolution—which means that determination of its truth or falsity will resolve an issue that is central to the validity of each one of the claims in one stroke.

Id. at 2251.

Justice Scalia next discussed commonality in the Title VII context, where plaintiffs like Dukes “wish to sue about literally millions of employment decisions at once.” Id. at 2252. “Without some glue holding the alleged reasons for all those [employment] decisions together,” he wrote, “it will be impossible to say that examination of all the class members’ claims for relief will produce a common answer to the crucial question why was I disfavored.” Ibid.

To determine whether Dukes’s class had sufficient commonality, Justice Scalia went to a familiar source: Falcon. That case, he held, “describes how the commonality issue must be approached.” Id. at 2252-53. In Falcon, “an employee who claimed that he was deliberately denied a promotion on account of race obtained certification of a class comprising all employees wrongfully denied promotions and all applicants wrongfully denied jobs.” Id. at 2253. The Supreme Court reversed. It noted:

Conceptually, there is a wide gap between (a) an individual’s claim that he has been denied a promotion on discriminatory grounds, and his otherwise unsupported allegation that the company has a policy of discrimination, and (b) the existence of a class of persons who have suffered the same injury as that individual, such that the individual’s claim and the class claims will share common questions of law or fact and that the individual’s claim will be typical of the class claims.

Falcon, 457 U.S. at 157, 102 S.Ct. 2364. To bridge that conceptual gap, Justice Scalia reasoned, the party seeking class certification must show that the defendant “used a biased testing procedure,” Dukes, 131 S.Ct. at 2553 (citing Falcon, 457 U.S. at 159 n. 15, 102 S.Ct. 2364), or she must produce “[significant proof that an employer operated under a general policy of discrimination ... if the discrimination manifested itself in hiring and promotion practices in the same general fashion, such as through entirely subjective decision-making processes.” Ibid, (citing Falcon, 457 U.S. at 159 n. 15, 102 S.Ct. 2364). Dukes succeeded in neither of these tasks and thus did not meet Rule 23(a)(2)’s threshold commonality requirement.

Justice Scalia also held, this time for a unanimous court, that “claims for monetary relief ... may not [be certified under Rule 23(b)(2) ], at least where ... the monetary relief is not incidental to the injunctive or declaratory relief.” Id. at 2557. He reasoned:

at a minimum, claims for individualized relief (like the backpay at issue here) do not satisfy the Rule.... Rule 23(b)(2) applies only when a single injunction or declaratory judgment would provide re lief to each member of the class. It does not authorize class certification when each individual class member would be entitled to a different injunction or declaratory judgment against the defendant. Similarly, it does not authorize class certification when each class member would be entitled to an individualized award of monetary damages.

Ibid. The only possible exception, Justice Scalia continued, is “monetary relief that is incidental to requested injunctive or declaratory relief.” Id. at 2560 (internal quotation marks omitted). But the court did not decide the issue because the monetary relief Dukes sought, backpay, was not incidental to declaratory or injunctive relief. To the contrary, Wal-Mart had the right to raise affirmative defenses to each individual backpay determination. 2 “And because the necessity of that litigation will prevent backpay from being ‘incidental’ to the classwide injunction,” Rule 23(b)(2) certification was not appropriate. Id. at 2561.

Justice Ginsburg concurred in part and dissented in part. She agreed with Justice Scalia’s conclusion that Rule 23(b)(2) certification was not appropriate. Id. at 2561 (Ginsburg, J., concurring in part and dissenting in part). She took issue, however, with the majority’s interpretation of Rule 23(a)(2), arguing that the Opinion of the Court “imported] into the Rule 23(a) determination concerns properly addressed in a Rule 23(b)(3) assessment.” Id. at 2562. “The Court’s emphasis on differences between class members,” she argued, “mimics the Rule 23(b)(3) inquiry into whether common questions ‘predominate’ over individual issues. And by asking whether the individual differences ‘impede’ common adjudication, the Court duplicates 23(b)(3)’s question whether ‘a class action is superior’ to other modes of adjudication.” Id. at 2566. Therefore, while she agreed that Dukes’s claim failed under Rule 23(b)(2), she believed that “[a] putative class of this type may be certifiable under Rule 23(b)(3),” and suggested: “Whether the class the plaintiffs describe meets the specific requirements of Rule 23(b)(3) ... [should be] reserve[d] ... for consideration and decision on remand.” Id. at 2561.

Dukes, in many ways, is similar to this case. Each involves a challenge to a national corporation’s employment practices. In each, the allegedly discriminatory employment decisions are ascribed to a corporate culture allegedly unfavorable to women. In each, applicants had to meet a basic set of criteria, but managers retained significant discretion over the challenged employment decisions. And in each, the class representative sought to prove her discrimination claim with a combination of statistical and anecdotal evidence.

C

The district court declined to certify Davis’s proposed class, based on Rules 23(a)(2)—(4), and 23(b)(2)-(3). Because the district court did not abuse its discretion as to Rule 23(a)(2) or Rule 23(b)(2), we affirm its decision denying class certification. We express no opinion on the district court’s treatment of Rules 23(a)(3) and 23(a)(4). 3

Rule 23(a)(2): Commonality

Under Rule 23(a)(2), a plaintiff must show that “there are questions of law or fact common to the class.” Fed. R.Civ.P. 23(a)(2). Dukes clarified the scope of this inquiry. To satisfy Rule 23(a)(2), a plaintiffs “claims must depend upon a common contention.... [which is] of such a nature that it is capable of class-wide resolution—which means that determination of its truth or falsity will resolve an issue that is central to the validity of each one of the claims in one stroke.” Dukes, 131 S.Ct. at 2551' The crucial inquiry, the court explained, is “the capacity of a classwide proceeding to generate common answers apt to drive the resolution of the litigation.” Ibid, (quoting Richard A. Nagareda, Class Certification in the Age of Aggregate Proof, 84 N.Y.U. L.Rev. 97, 132 (2009)). “In other words, Plaintiffs must have a common question that will connect many individual promotional decisions to their claim for class relief.” Ellis v. Costco Wholesale Corp., 657 F.3d 970, 981 (9th Cir.2011).

The district court held that Davis did not satisfy this requirement. It reasoned, first, that Cintas’s hiring process was not entirely subjective, and therefore might not “support a finding of a ‘general policy of discrimination.’ ” Serrano, 2009 WL 910702, at *5 (quoting Bacon v. Honda of Am. Mfg., 205 F.R.D. 466, 477 (S.D.Ohio 2001)). But “[m]ore importantly,” it continued, “the hiring process in this case is conducted by thousands of Cintas managers at hundreds of Cintas facilities. Hiring decisions are made for a diverse range of reasons and depend on widely differing circumstances at each facility.” Ibid. The district court also pointed out that Cintas’s hiring process had many different steps and, at some points, involved hiring managers who are women themselves. It reasoned, therefore: “Putative class members would have suffered the alleged discrimination in different ways at different stages of the hiring process, and depending on the different Cintas employees involved at each hiring stage.” Ibid. The court found Davis’s statistical evidence unpersuasive, holding that Cintas’s experts pointed to “discrepancies [that] undermine[d] a conclusion that the statistics are sufficient to demonstrate that there is a common, class-wide discriminatory impact against the putative class members.” Id. at *6. Equally unconvincing, it held, was Dr. Barbara Reskin’s expert opinion that Cintas had a white-male-dominated business culture, which replicated itself in hiring decisions. Finally, the court did “not find that [Davis’s anecdotal accounts were] ... compelling instances of discrimination against women.” Id. *7.

In response, Davis argues that her “statistical evidence establishes a pattern of underhiring across locations ... [that] is more than sufficient to satisfy commonality.” Appellant’s Br. 36. She also argues that the district court erred by holding that “the existence of the few minimal objective standards for the SSR job ... per se precluded certification,” ibid., and that “commonality was barred per se because the challenged decisionmaking is made by multiple managers at multiple locations.” Id. at 38. Finally, Davis claims that the existence of ■ individual questions, such as Cintas’s justification for an individual hiring decision or who Cintas hired instead of the class member, does not necessarily preclude commonality. Rather, she suggests, “these are issues for ... additional remedial proceedings.” Id. at 41.

Davis’s arguments do not show that the district court abused its discretion, particularly in light of Dukes. Under the Dukes framework, Davis would have to show that Cintas “used a biased testing procedure” or “operated under a general policy of discrimination.” Dukes, 131 S.Ct. at 2553. As in Dukes, the gravamen of Davis’s claim is not that the Meticulous Hiring System’s objective criteria led to an anti-female bias, but that subjective decisions made by some of Cintas’s managers favored males because of Cintas’s male-dominated corporate culture.

“[SJubjective or discretionary employment practices may be analyzed under the disparate impact approach in appropriate cases.” Watson v. Fort Worth Bank and Trust, 487 U.S. 977, 991, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988); accord Dukes, 131 S.Ct. at 2554. When plaintiffs challenge employment practices in a large, national corporation, however, “demonstrating the invalidity of one manager’s use of discretion will do nothing to demonstrate the invalidity of another’s.” Dukes, 131 S.Ct. at 2554. Unless a plaintiff can somehow show that the corporation’s managers all used “a common mode of exercising discretion that pervades the entire company,” Dukes explains, “[a] party seeking to certify a nationwide class will be unable to show that all the employees’ Title VII claims will in fact depend on the answers to common questions.” Id. at 2554-55.

The court in Dukes explained that the plaintiffs’ sociological, statistical, and anecdotal evidence—all of which was similar to the evidence offered here—was not sufficient to show a uniform, companywide practice of exercising discretion in a way that favored men over women. Applying the abuse-of-discretion standard, we affirm the district court’s determination that Davis’s statistical evidence, sociological analysis, and anecdotal accounts did not satisfy Rule 23(a)(2). As to each type of evidence, the district court weighed the parties’ competing arguments and found that Davis’s proffered evidence did not support a finding of companywide gender discrimination. We may not overturn this determination unless the district court “re-lie[d] on erroneous findings of fact, applie[d] the wrong legal standard, misapplied] the correct legal standard when reaching a conclusion, or ma[de] a clear error of judgment.” Pipefitters Local 636 Ins. Fund v. Blue Cross Blue Shield of Mich., 654 F.3d 618, 629 (6th Cir.2011).

Davis’s experts suggested that women and minorities were under-represented in service-sales-representative positions companywide. Serrano, 2009 WL 910702, at *6 Cintas’s experts countered by questioning Davis’s experts’ methods and models, and concluding that “although some Cintas locations under-hired women and racial minorities, other locations over-hired women and racial minorities during the same period.” Ibid. 4 , The district court found Cintas’s experts more persuasive.

Davis presented a report from Dr. Barbara Reskin, “purporting to show that a common white male business culture at Cintas tends to perpetuate its culture by discriminating against women and racial minorities.” Ibid. Cintas, of course, argued that Reskin’s conclusion was untenable. “After a close consideration of the parties’ respective positions, the Court [concluded] ... that Dr. Reskin’s report [was] not persuasive.” Ibid. It questioned Reskin’s methodology, the applicability of her general thesis to specific instances of discrimination, and noted that Cintas had made “sincere attempts to achieve greater diversity in its company.” Id. at *7.

Finally, Davis presented anecdotal evidence of Cintas managers telling women that the job involved heavy lifting, entering male locker rooms, and dealing with dirty laundry. The court found, first, that these statements “could be interpreted as instances of Cintas managers giving applicants full disclosure of the. demands and duties of the SSR position,” and second that “[e]ven assuming that these statements constituted particular instances of discrimination, commonality is not satisfied; on the contrary, these statements illustrate that the circumstances of discrimination are highly individualized and cannot be adequately treated in a generalized class action setting.” Ibid.

None of these evidentiary determinations was an abuse of discretion. Combined, they led the district court to the same conclusion that the Supreme Court reached in Dukes: the plaintiff did not satisfy Rule 23(a)(2) because she could not show that a number of women, who failed to obtain employment at many places, over a long time, under a largely subjective hiring system, shared a common question of law or fact. We affirm.

Rule 23(b)(2): Injunctive or Declaratory Action

Davis moved for class certification under both Rule 23(b)(2) and Rule 23(b)(3). However, she only challenges the district court’s determination that class certification under Rule 23(b)(2) was inappropriate. Any appeal from denial of certification under Rule 23(b)(3) is, therefore, forfeited. See Miller v. Admin. Office of the Courts, 448 F.3d 887, 893 (6th Cir.2006) (holding that issues not raised in opening appellate briefs are considered forfeited).

Rule 23(b)(2) provides that class certification is permissible where a class representative meets all of Rule 23(a)’s requirements and “the party opposing the class has acted or refused to act on grounds that apply generally to the class, so that final injunctive relief or corresponding declaratory relief is appropriate respecting the class as a whole.” Fed.R.Civ.P. 23(b)(2).

The district court held that Davis did not meet this requirement because, in addition to seeking declaratory and injunctive relief, she sought front pay and back pay. Serrano, 2009 WL 910702, at *9-10. It reasoned that Cintas’s decentralized hiring policy meant that it did not act in a manner that applied generally to the class, and held: “the damages sought by Plaintiffs would require individualized determinations inappropriate for a [Rule 23(b)(2) ] class action.” Id. at *10. Front pay and back pay calculations, the court continued, would “necessarily predominate over requested declaratory or injunctive relief and the requested damages cannot be re covered pursuant to Rule 23(b)(2).” Ibid. (internal quotation marks omitted).

The Supreme Court unanimously spoke in a similar vein in Dukes. “[W]e think,” it wrote, “that, at a minimum, claims for individualized relief (like the backpay at issue here) do not satisfy the Rule.” Dukes, 131 S.Ct. at 2557. It continued:

Rule 23(b)(2) applies only when a single injunction or declaratory judgment would provide relief to each member of the class. It does not authorize class certification when each individual class member would be entitled to a different injunction or declaratory judgment against the defendant. Similarly, it does not authorize class certification when each class member would be entitled to an individualized award of monetary damages.

Ibid. The Court reasoned that there might be an exception to this rule if monetary relief were incidental to the declaratory or injunctive relief the plaintiffs sought. Id. at 2560. In the Title VII context, however, the Court held that defendants are entitled to “individualized determinations of each employee’s eligibility for backpay,” and thus “the necessity of [litigating individuals’ claims] will prevent backpay from being incidental to the class-wide injunction.” Id. at 2560-61.

Davis claims that Dukes does not bar certification of her Rule 23(b)(2) class action because her “shortfall-based model” is distinguishable from the “trial-by-formula” system the Supreme Court expressly rejected in Dukes. Appellant’s Supp. Br. 4-5. Under Davis’s system, the court would declare Cintas’s hiring practices discriminatory and issue an injunction ordering Cintas to hire class members “randomly selected in numbers equal to the proven shortfalls for each facility.” Id. at 5. “Davis will [then] calculate Cintas’ backpay liability for the class as a whole by multiplying the proven shortfall times lost wages. 5 This calculation will be the limit of Cintas’ liability for backpay, which will be distributed among eligible class members, pro rata.” Ibid.

The “trial-by-formula” system, which the Supreme Court rejected in Dukes, was similar. It provided that the district court would appoint a master to determine whether and how much backpay was due to a sample set of class members. The court would then multiply the total number of class members by the percentage of claims the special master determined were valid. Next, it would multiply that number by the average backpay award for sample claimants with a valid claim to determine the class’s recovery. Dukes, 131 S.Ct. at 2560-61. The Supreme Court rejected this system under the Rules Enabling Act, holding that it abridged or modified Wal-Mart’s statutory right to assert individual defenses to individual awards of backpay. Id. at 2561.

Davis’s proposed system suffers from a similar, but even more troubling, infirmity. Dukes made clear that Cintas has the right to present defenses before paying any person an award of backpay. Davis’s system deprives Cintas of that right. But worse, Davis’s “shortfall-based” model, unlike the “trial-by-formula” system, makes no effort to individualize damages at all. Under the “shortfall-based” model, a woman denied a job in 2004 would receive precisely the same recovery as a woman denied a job in 1999. At least in the “trial-by-formula” system, plaintiffs have statisti cal similarity. Davis’s system for determining class recovery, therefore, is worse than the system that the Supreme Court unanimously rejected in Dukes. Individualized monetary relief is therefore not incidental to the injunctive and declaratory relief Davis seeks, and the district court was correct to deny class certification under Rule 23(b)(2). Ibid.

Ill

That class certification is inappropriate in this case does not vitiate Davis’s individual claims. We therefore proceed to consider whether the district court properly granted summary judgment for Cintas on Davis’s individual disparate-treatment and disparate-impact claims.

Summary judgment is appropriate when “the movant shows that there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.” Fed.R.Civ.P. 56(a). We review the grant of a motion for summary judgment de novo, construing all evidence and drawing all inferences against the moving party. Hirsch v. CSX Transp., Inc., 656 F.3d 359, 362 (6th Cir.2011). However, “the mere existence of a scintilla of evidence in support of [the non-moving party’s] position will be insufficient; there must be evidence on which the jury could reasonably find for the [non-moving party].” Shropshire v. Laidlaw Transit, Inc., 550 F.3d 570, 576 (6th Cir.2008).

When a plaintiff alleges, based on circumstantial evidence, that she suffered disparate treatment in violation of Title VII, we analyze her claim using the familiar McDonnell Douglas burden-shifting analysis. See McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802-04, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973). In the McDonnell Douglas framework, Davis must make out a prima facie case of discrimination. Id. at 802, 93 S.Ct. 1817. If she meets this requirement, Cintas must offer some legitimate, nondiseriminatory explanation for its employment decision. Ibid. If the company produces such an explanation, Davis must point out “evidence from which a jury could reasonably reject [Cintas’s] explanation.” Chen v. Dow Chem. Co., 580 F.3d 394, 400 (6th Cir.2009). Davis bears this third burden, even when opposing a motion for summary judgment. Ibid. She must, therefore, point to evidence that, taken in a light most favorable to her, could lead a reasonable jury to reject Cintas’s proffered explanations.

While “a plaintiffs prima facie ease, combined with sufficient evidence to find that the employer’s asserted justification is false, may permit the trier of fact to conclude that the employer unlawfully discriminated,” there are also “instances where, although the plaintiff has established a prima facie case and set forth sufficient evidence to reject the defendant’s explanation, no rational factfinder could conclude that the action was discriminatory.” Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133, 148, 120 S.Ct. 2097,147 L.Ed.2d 105 (2000).

Davis alleges two separate instances of disparate treatment—one by Richards in 2003, and one by Cintas’s collective hiring personnel in 2004. Below, as here, Cintas conceded that Davis established a prima facie case of discrimination, and only litigated the issue of pretext.

A plaintiff may show pretext by demonstrating: “(1) that the proffered reasons had no basis in fact, (2) that the proffered reasons did not actually motivate [the adverse employment action], or (3) that they were insufficient to motivate [the adverse employment action].” Hedrick v. Western Reserve Care Sys., 355 F.3d 444, 460 (6th Cir.2004); Chen, 580 F.3d at 400. This test, however, is not rigid, and “it is important to avoid formalism in its application, lest one lose the forest for the trees. Pretext is a commonsense inquiry: did the employer fire [or, as here, refuse to hire] the employee for the stated reason or not?” Chen, 580 F.3d at 400 n. 4. “[Because a prima facie case [of discrimination] and sufficient evidence to reject the employer’s explanation may permit a finding of liability ... [Davis need not] introduce additional, independent evidence of discrimination” to survive summary judgment. Reeves, 530 U.S. at 149, 120 S.Ct. 2097.

The district court concluded that Richards’s decision not to hire Davis in 2003 was based on legitimate, nondiscriminatory reasons. Specifically, it held that Davis did not provide evidence that Richards’s proffered reasons for not hiring her— Davis’s stated dislike for “up-selling,” her desire to remain employed part-time at LensCrafters, and the fact that she was applying for other jobs—were a pretext for gender discrimination. Avalos v. Cintas Corp., No. 06-12311, 2010 WL 1417804, at *7 (E.D.Mich. Apr. 5, 2010). Davis responds that: (1) Richards’s reasons were mere speculation, since she could not recall the specifics of her screening interview with Davis, and thus had to rely on notes she made during the interview process; and (2) Richards advanced a number of men past the screening stage who were not as well qualified as Davis.

Davis’s first argument ignores her ultimate burden to prove pretext. To survive summary judgment, Davis must provide evidence that could lead a reasonable jury to find that Cintas’s proffered reasons for declining to hire her were pretextual. See Chen, 580 F.3d at 400. While Davis questions the reliability of Richards’s “contemporaneous ... business records,” Appellee’s Br. 60, she offers no evidence that Richards’s notes were misleading, unreliable, or anything other than what Richards claimed they were: a record of the red flags that motivated her not to advance Davis past the screening stage.

Davis’s second argument—that Richards’s advancing less qualified men to later stages of the screening process is evidence of pretext—fares better. “Whether qualifications evidence will be sufficient to raise a question of fact as to pretext will depend on whether a plaintiff presents other evidence of discrimination.” Bender v. Hecht’s Dept. Stores, 455 F.3d 612, 626 (6th Cir.2006). If a plaintiff presents such evidence, “that evidence, taken together with evidence that the plaintiff was as qualified as or better qualified than the successful applicant, might well result in the plaintiffs claim surviving summary judgment.” Id. at 626-27. If, however, “there is little or no other probative evidence of discrimination, to survive summary judgment the rejected applicant’s qualifications must be so significantly better than the successful applicant’s qualifications that no reasonable employer would have chosen the latter applicant over the former.” Id. at 627.

Whether Davis produces “other evidence of discrimination,” id. at 626, depends on the outcome of a battle of experts. Davis’s expert, Dr. Thomas DiPrete, analyzed hiring data for Location 447. He stated: “Between 1999 and 2004, 78 men were hired into SSR positions [at Location 447], but only 2 women were hired into these positions. All of the hires in the period from 1999-2002 were men. During 2003 and 2004, 32 hires occurred; 30 of these hires were men and 2 were women.” This was so, even though women accounted for between 26% and 27% of service-sales- representative applicants and 30% to 38% of the external labor market during that time period. “No women at all were hired into SSR jobs in Cintas location 447 for the four years prior to 2003.” From July 1, 2003 to December 31, 2004, “there were 27 hires into SSR positions. Only one of these hires (3.7% of the hires) was a woman even though more than 25% of these applicants were women.”

To calculate the number of women available in the labor market, Dr. DiPrete used an amalgam of ninety-three different census codes. He “analyzed in detail the last occupation of a sample of approximately 300 applicants who were hired into the SSR-Uniform or SSR-FS jobs,” and “defined labor market availability by weighting the occupations from which Cintas hired in proportion to their presence in [his] sample.”

Cintas’ expert witnesses, Drs.' Mary Baker and Janet Thornton, claimed that Dr. DiPrete’s analysis was fundamentally flawed. Of importance to this discussion, 6 Drs. Baker and Thornton criticized DiPrete for “measur[ing] availability [of applicants in the labor market] using the demographic composition of workers in many occupations,” instead of “using hiring benchmarks based on occupation 913 alone.” 7

As Drs. Baker and Thornton point out, Dr. DiPrete’s availability measure hinges on the assumption that “the men and women ... who work in the origin jobs are similarly interested and qualified for the SSR job.” However, as Dr. DiPrete points out, “census code 913 includes many jobs which are not ‘highly similar’ to the Cintas SSR job.” Which view to accept is ultimately an issue for the fact finder. But, in the current posture of this case, we must draw all inferences in Davis’s favor. We therefore assume that Davis would win the battle of experts, and treat DiPrete’s statistical analysis as evidence of discrimination. Accordingly, we now consider whether a reasonable jury could conclude that Davis was “as qualified as or better qualified than the successful applicant^].” Bender, 455 F.3d at 626-27.

Location 447 hired three men soon after interviewing Davis. At least on paper, their credentials appear equal to, or slightly less impressive than, Davis’s. 8 One of the men, Tim Koelbl, was a onetime route sales driver for an ice company, who had a master’s degree in education. Another, Andrew Hansen, had participated in a rental car agency’s management program, which “stressed ... [c]ustomer service, sales, marketing and management.” The third, Damian Vertz, was a former college football player, who had worked as a parts technician and machine operator.

None of the three men had real-world experience in management; none had extensive experience in sales. Davis, by contrast, had worked as a manager for three different companies. She had significantly more customer-service and sales experience than any of the three male candidates, even if she disliked up-selling prod ucts and planned to continue working for LensCrafters part-time. She was, in other words, “as qualified or better qualified than [any of] the successful applicant[s].” Bender, 455 F.3d at 626-27. This, together with Dr. DiPrete’s statistical analysis, is enough at this stage. The district court should not have granted summary judgment to Cintas on Davis’s 2003 disparate-treatment claim.

The district court also rejected Davis’s 2004 disparate-treatment claim. It reasoned that Cintas chose not to hire Davis because of her poor performance on the route ride, not because she was female. Avalos, 2010 WL 1417804, at *7. It also dismissed as speculation Davis’s claim that Cintas delayed another female applicant’s route ride so that it could compare the two and hire one. Id. at *8.

On appeal, Davis opposes the first of these conclusions vigorously. She suggests that the district court erred by holding that she did not perform well on her route ride, since, in fact, the manager who conducted the route ride recommended her for a route that involved less physical exertion. Appellant’s Br. 61. But, as Cintas notes, such a route did not exist when Davis applied for her position. Id. at 63. The two male candidates Cintas hired soon after Davis interviewed in 2004, unlike Davis, performed well on their route rides, demonstrating an adequate level of physical energy for the positions Cintas then sought to fill. Davis has not produced evidence suggesting that Cintas’s hiring decisions were based on these candidates’ gender, rather than their ability to perform necessary job functions.

Davis leaves mostly untouched the district court’s ruling that dismissed as speculation her argument that Location 447 had determined to hire only one female applicant in 2004. She discusses the factual basis for this claim, id. at 21, but only mentions it in passing in her summary-judgment argument, claiming that Location 447’s general manager wanted to hire only one woman, and deferred to a lower-level manager’s preference for a woman other than Davis. Such a perfunctory, unsubstantiated statement does not preserve the issue for appeal. Davis has not carried her burden to create an issue of material fact on pretext. The district court was correct to dismiss her 2004 disparate-treatment claim.

“By enacting § 2000e-2(k)(l)(A)(i), Congress allowed claims to be brought against an employer who uses a practice that causes disparate impact, whatever the employer’s motives.” Lewis v. City of Chicago, 560 U.S. 205, 130 S.Ct. 2191, 2200, 176 L.Ed.2d 967 (2010). To establish a prima facie disparate-impact ease, a plaintiff must: (1) identify a specific employment practice; and (2) present data indicating that the specific practice had an adverse impact on a protected group. Grant v. Metro. Gov’t of Nashville and Davidson Cnty., 446 Fed.Appx. 737, 740 (6th Cir.2011); see also Kovacevich v. Kent State Univ., 224 F.3d 806, 830 (6th Cir.2000). “[I]f the complaining party can demonstrate to the court that the elements of a respondent’s decisionmaking process are not capable of separation for analysis, the decisionmaking process may be analyzed as one employment practice.” 42 U.S.C. § 2000e-2(k)(l)(B)(i).

Once the plaintiff succeeds in making a prima facie disparate-impact case, the defendant may avoid liability by showing “that the protocol in question has ‘a manifest relationship to the employment.’ ” Dunlap v. Tenn. Valley Auth., 519 F.3d 626, 629 (6th Cir.2008) (citing Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971)). If the defendant makes such a showing, the plaintiffs disparate-impact claim will succeed only if she demonstrates “that other tests or selection protocols would serve the employer’s interest without creating the undesirable discriminatory effect.” Ibid, (citing Albemarle Paper Co. v. Moody, 422 U.S. 405, 425, 432, 95 S.Ct. 2362, 45 L.Ed.2d 280 (1975)).

The district court held that Davis failed to establish her prima facie case because she did not identify a specific employment practice. Citing our decision in Phillips v. Cohen, 400 F.3d 388, 398 (6th Cir.2005), it explained that, while an entire “decisionmaking process may be analyzed as one practice,” before such analysis is appropriate, “the plaintiff [must] demonstrate[] that the elements of a respondent’s decisionmaking process are not capable of separation for analysis.” The district court found that Davis did not meet this burden. “[T]he fact that [she] advanced to different points,” it reasoned, “demonstrates that the hiring process is capable of separation.” Avalos, 2010 WL 1417804, at *10. Thus, the district court rejected her claim at the threshold.

Davis argues that the district court erred. She suggests that she did “identify the system’s subjective elements as the cause of the challenged underhiring: (a) the managers’ subjective assessment of the applications and interviews; and (b) their unconstrained discretion to weigh the mix of negatives and positives for each applicant.” Appellant’s Br. 63-64.

Davis’s briefs, below and in this court, do discuss the possibility of considering an entire hiring system as one employment practice when its discrete parts cannot be separated for analysis. The gravamen of her claim, though, is somewhat different. At bottom, Davis argues that the Meticulous Hiring System’s subjective elements, together, caused a disadvantage to women in Service-Sales-Representative hiring.

Even though it does not address the issue specifically, the district court’s opinion suggests that, in a multi-step system, a “particular employment practice” must be one specific test, or one specific manager’s exercise of discretion. Davis, by contrast, necessarily claims that “particular employment practice” is broad enough to comprise many different steps of a multi-step process, as long as those steps share a common characteristic: subjectivity.

To determine which of these interpretations is correct, we must give content to the phrase “particular employment practice” in 42 U.S.C. § 2000e-2(k)(l)(B)(i). The language of the statute, our starting point, is somewhat enigmatic. Title VII provides:

It shall be an unlawful employment practice for an employer—

(1) to fail or refuse to hire or to discharge any individual, or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual’s race, color, religion, sex, or national origin; or

(2) to limit, segregate, or classify his employees or applicants for employment in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee, because of such individual’s race, color, religion, sex, or national origin.

Id. at § 2000e-2(a)(l)-(2). Section 2000e-2(k)(l)(A), however, uses the term “employment practice” differently. Under that section, an employer faces liability if:

a complaining party demonstrates that a respondent uses a particular employment practice that causes a disparate impact on the basis of race, color, religion, sex, or national origin and the respondent fails to demonstrate that the challenged practice is job related for the position in question and consistent with business necessity....

Id. at § 2000e-2(k)(l)(A)(i). “Employment practice” here cannot mean the same thing that it means earlier in the statute. Such a construction would make § 2000e-2(k) hopelessly question-begging: it would ban discrimination based on a prohibited classification that causes discrimination based on a prohibited classification. Instead, the two provisions must use the term in distinct ways: § 2000e-2(a) defines an act of discrimination, a specific “employment practice” in which a person must not engage. Conversely, § 2000e-2(k) does not define a specifically forbidden act—it prohibits an unacceptable outcome arising from some undefined “employment practice.”

However, the starting point of § 2000e-2(k) analysis is still some discrete “employment practice,” that is, something that the employer does. That “something” cannot be the hiring system itself, since § 2000e-2(k)(l)(B)(i) distinguishes an “employment practice” from “a respondent’s decision-making process.” Id. But whether the “something” must be one isolated element of a multi-step hiring procedure, or whether it can include all of the elements of such a procedure that share a common characteristic (subjectivity, for example) is not entirely clear.

The text of § 2000e-2(k)(l)(B)(i), though, sheds some light on the issue. That section softens the Supreme Court’s decision in Wards Cove Packing Co., Inc. v. Atonio, 490 U.S. 642, 109 S.Ct. 2115, 104 L.Ed.2d 733 (1989), superseded by statute, 42 U.S.C. § 2000e-2(k), by allowing plaintiffs to identify a decision-making process as the “particular employment practice” in a disparate-impact case if that process is incapable of separation for analysis. Section 2000e-2(k)(l)(B)(i) requires that, as a general rule, “the complaining party [must] ... demonstrate that each particular challenged employment practice causes a disparate impact.” Two features of this language are suggestive. First, the language that deals with the “employment practice” is entirely in the singular, not the plural. This syntax would be strange if a plaintiff could bundle a number of discrete steps of a multi-phase hiring process together, based on a common characteristic. Second, the words “each” and “particular” suggest specificity, not amalgamation. The text of § 2000e2(k)(l)(B)(i), then, suggests that Davis must identify one specific step of the Meticulous Hiring Process as a particular employment practice, rather than pointing to a group of steps that share a common characteristic.

Precedent from the Supreme Court, our circuit, and our sister circuits supports this conclusion. True, “[i]f an employer’s undisciplined system of subjective decisionmaking has precisely the same effects as a system pervaded by impermissible intentional discrimination ... [the employer’s] subjective or discretionary employment practices may be analyzed under the disparate impact approach.” Watson, 487 U.S. at 990-91, 108 S.Ct. 2777. 9 But even after Watson, “a plaintiff must [normally] demonstrate that it is the application of a specific or particular employment practice that has created the disparate impact under attack.” Wards Cove Packing Co., 490 U.S. at 657, 109 S.Ct. 2115. It is simply not enough to “point out that the [hiring practices] at issue [are] relatively less generous” to some workers than to others. Smith v. City of Jackson, 544 U.S. 228, 241, 125 S.Ct. 1536, 161 L.Ed.2d 410 (2005).

We too have acknowledged that, even if an employment practice “involve[s] subjectivity, [it] may nonetheless constitute [an] ‘employment practice[]’ for purposes of the disparate impact analysis.” Phillips, 400 F.3d at 398. At the same time, though, we have emphasized “that a plaintiff is responsible for isolating and identifying the specific employment practices that are allegedly responsible for any observed statistical disparities.” Grant, 446 Fed.Appx. at 740 (quoting Watson, 487 U.S. at 994, 108 S.Ct. 2777). In keeping with this obligation, we have required that “Plaintiffs make [some] effort to isolate any ... [allegedly adverse] practices [and] to examine their individual effects on the promotions process.” Ibid. Our sister circuits have done the same. See, e.g., McClain v. Lufkin Indus., Inc., 519 F.3d 264, 275-78 (5th Cir.2008) (noting that plaintiff must “demonstrate that each particular challenged employment practice causes a disparate impact,” and upholding use of subjective decision-making as specific employment practice); E.E.O.C. v. Joe’s Stone Crab, Inc., 220 F.3d 1263, 1279 (11th Cir.2000) (holding that subjective interview could trigger Title VII disparate-impact liability, but holding that liability was not appropriate because no showing of causation).

Thus, Davis did not identify a “particular employment practice” within the meaning of Title VII by pointing to all of the subjective elements in the Meticulous Hiring System. She could still survive summary judgment, however, if she showed that the Meticulous Hiring System’s many steps were so intertwined that they were not capable of separation for analysis. As the district court noted, though, Davis did not explain why the well-defined, discrete elements of the Meticulous Hiring System are “not capable of separation for analysis.” 42 U.S.C. § 2000e-2(k)(l)(B)(i). Indeed, not all of the system’s subjective elements are the same. Each different interview, for instance, has a specific interview guide, and different managers conduct interviews at different stages of the process.

Of course, as Davis pointed out at oral argument, the same small group of managers does conduct all of Location 447’s interviews. And surely, Davis might urge, if the statistical data indicate that those managers have gender biases, their exercising discretion at different steps of the Meticulous Hiring Process should not insulate Cintas from disparate-impact liability.

This may be so, but the simple fact remains: Davis did not isolate the specific practices that caused the disparate impact she alleges; nor did she show that the managers’ various exercises of discretion in the Meticulous Hiring System were incapable of separation for analysis. Davis’s disparate impact claims for both 2003 and 2004, therefore, fail.

IV

This case presents a number of complex legal and factual issues. In sum, we (1) AFFIRM the denial of class certification; (2) REVERSE the district court’s grant of summary judgment on Davis’s 2003 disparate-treatment claim; and (3) AFFIRM the district court’s grant of summary judgment on Davis’s 2004 disparate-treatment claim and both of her disparate-impact claims. We remand to the district court for proceedings consistent with this opinion.

1 . Some courts also impose two "implicit requirements,” definiteness and membership in the class, in addition to Rules 23(a)(1)—(4). See 5 Herbert B. Newberg & Alba Conte, New-berg on Class Actions § 3:1 (2011). These supplemental—perhaps superfluous—prerequisites, however, are not at issue here.

2 . As a solution to this problem, Dukes proposed a "Trial by Formula” process. Under this system, the district court would appoint a master to determine whether and how much backpay was due to a sample set of class members. The court would then multiply the total number of class members by the percentage of claims the special master determined were valid. Next, it would multiply that number by the average backpay award for sample claimants with a valid claim to determine the class's recovery. The Court did not make clear whether Dukes proposed that the class's recovery would be distributed pro rata, whether there would be some sort of claims procedure, based on the particular applicant's date of non-promotion, or whether class counsel would dispose of the money through a cy pres distribution. Regardless, the Court held that the Trial by Formula approach would violate the Rules Enabling Act because it would abridge or modify WalMart’s right to present affirmative defenses to individual backpay determinations. Dukes, 131 S.Ct. at 2561.

3 . Davis does not appeal the district court’s determination that her suit does not satisfy Rule 23(b)(3).

4 . The district court, for the purposes of argument, accepted Davis's data on this point, meaning that it measured under- or over-hiring in relation to employment rates in certain classes of jobs, based on one or more census codes. Census codes are numbers as signed to particular jobs (not necessarily the same as Cintas’s jobs) when tallying responses to a census.

5 . Davis’s expert calculated the shortfall as approximately 900 jobs. Appellant's Supp. Br. 5 n. 8. Davis does not, however, explain how it would calculate when the shortfall occurred. For instance, a hypothetical woman not hired in 1999 would be due substantially more than a hypothetical woman not hired in 2003.

6 . Drs. Baker and Thornton also criticized DiPrete's aggregating data from different Cintas locations nationwide, and some of his methods of analyzing the aggregated data. These points, however, are relevant only to Davis’s class claim, since her individual claim implicates only Location 447.

7 . As discussed above, census codes are numbers assigned to particular jobs when tallying responses to a census. Census code 913 comprises “driver/sales” jobs, including “truck driver, light or delivery services ... [and] truck driver, heavy and tractor-trailer.”

8 .Since, in 2003, Davis did not advance past the application screening stage, the applicants’ qualifications on paper are the focus of the inquiry.

9 . In Dulces, Justice Scalia stated: "Other than the bare existence of delegated discretion, respondents have identified no 'specific employment practice—much less one that ties all their 1.5 million claims together.” Dulces, 131 S.Ct. at 2555-56. It is possible to argue that this language undermines Watson’s core holding by requiring that plaintiffs do more than allege that managers' discretion led to discrimination. Justice Scalia, however, gave no indication that he intended such a drastic consequence, or even that he intended significantly to dilute Watson's import. Indeed, he called Watson “the landmark case of ours which held that giving discretion to lower-level supervisors can be the basis of Title VII liability under a disparate-impact theory.” Id. at 2555. The better reading of this language takes into account that Justice Scalia discussed Watson in the class-certification context. His point was that, on a nationwide scale, bare discretion was too tenuous to support class certification, absent some showing that all managers exercised their discretion in the same way.