Plaintiffs are 47 former longtime employees, mostly African American, of the District of Columbia (District) Child and Family Services Agency (Agency), many of whom successfully served the Agency for decades. They numbered among the employees terminated as part of a large-scale reduction in force at the Agency following budget cuts. Plaintiffs alleged that their firings were unlawfully discriminatory on the basis of age and race. They have abandoned their age-based claims, but appeal the summary judgment in the Agency's favor on the race discrimination claims.

We generally affirm the decision of the district court, but reverse and remand on one narrow question: whether the plaintiffs identified a "particular employment practice" susceptible to challenge for its adverse racial impact under Title VII. 42 U.S.C. § 2000e-2(k)(1)(A)(i). On this issue, the District prevailed below on the theory that a reduction in force, or "RIF," is not a particular employment practice. What is at issue here is not a RIF in the abstract, however, but the means by which the Agency implemented it. Plaintiffs challenge the practices of the Agency in selecting for elimination jobs and job categories disproportionately held by African American employees. Nothing in Title VII suggests that the practices an employer uses to effectuate the adverse employment action of layoffs, whether or not dubbed a RIF, are exempt from disparate-impact scrutiny. We accordingly reverse the "particular *1244practice" holding and the accompanying denial of class certification, and remand for further proceedings.

Having decided the case on that threshold question, the district court had yet to address whether plaintiffs' statistical evidence sufficed to make out a prima facie case of disparate impact, or whether the Agency had business justifications for the layoff criteria it used. We accordingly express no opinion on those issues. We affirm the district court's decisions with respect to plaintiffs' challenge to the college degree requirement the Agency added to one job category, and the applicability of estoppel to certain individual plaintiffs' claims.

I. Background

A. Factual Record

The District of Columbia Child and Family Services Agency provides critical support services to abused and neglected children and struggling families. The Agency's functions include investigating reports of child abuse and neglect, temporarily removing children from unsafe settings, and securing medical care for affected children and families. As of Fiscal Year (FY) 2009, the Agency employed nearly one thousand people in its six major components: Agency Programs, Community Services, Policy and Planning, Clinical Practice, Agency Management, and Financial Operations.

In the face of significant municipal revenue shortfalls, the District of Columbia City Council decreased the Agency's operating budget for fiscal years 2010 and 2011. Following the budget cuts, the Agency reduced the number of its full-time employees. Relevant here are the job cuts effected for the Agency's FY 2011 budget. The District represented, and plaintiffs did not dispute, that the District could make the needed spending cuts by reducing full-time positions by 52-from 892 to 840-although the Agency fired more than twice that many people and then hired several dozen new employees.

All told, the Agency let go 115 employees. Plaintiffs here challenge as racially discriminatory the procedures used to implement that reduction in force. At an agency that was 73.4 percent African American, 93 percent (107 out of 115) of the terminated employees were African American. The Agency has never claimed to have laid off the most expensive employees, nor did it set out to make proportional cuts to each department. And, according to the Agency's Director, the cuts were not performance based: the Director assured the fired employees that the layoffs "in no way reflect[ ] adversely on your performance of your official duties." Joint App'x (J.A.) 660.

Plaintiffs claim that the Agency instead chose to cut and cull the very job categories most densely occupied by African American employees. The Agency focused its cuts on the Agency Programs Office, the Office of Clinical Practice, and the Office of Community Services, with the Agency Programs Office bearing the brunt. There, the Agency eliminated wholesale two social-worker support positions: Social Worker Associate (SWA), which required a bachelor's degree, and Social Service Assistant (SSA), which did not. The Agency's decision to fire everyone in the SSA and SWA job categories resulted in the termination of approximately 70 employees, 67 of whom were African American. And the culling of positions elsewhere at the Agency resulted in layoffs of 45 employees, 40 of whom were African American.

The District claims that the Agency "did not utilize a single uniform criteria, test or requirement" in determining which positions *1245would be eliminated. Def.'s Statement of Undisputed Material Facts (SOF) ¶ 15, J.A. 235. Rather, the District represents that the choices of which jobs to eliminate came about through "realigning functions and implementing new service models," as well as "multiple individual decisions made by the Director working in close consultation with the Chief of Staff, the Deputy Directors in charge of CFSA's various divisions, and other senior level managers in the Agency's executive team." Id . ¶¶ 15, 17, J.A. 235.

Immediately following the layoffs, the Agency created a new posting to replace the SSW and SWA roles, Family Support Worker (FSW), which did similar work but required a bachelor's degree. The Agency sought to hire approximately three dozen people to fill the new FSW spots, and it considered applicants whom it had just discharged as well as outside candidates. According to the District, 44 of the 115 people who lost their jobs applied for a position as an FSW, but only 30 of those held the required bachelor's degree. The Agency ultimately hired back into the FSW role 18 of the employees whom it had fired.

B. Procedural History

Forty-seven former Agency employees who lost their jobs filed this case as a class action against the District of Columbia; they alleged race and age discrimination under Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq. , and the District of Columbia Human Rights Act, D.C. Code § 2-1402.11. Plaintiffs brought both disparate treatment and disparate impact challenges to (1) the Agency's choice to respond to budgetary constraints by eliminating two job categories in which African American employees were most concentrated, and by using a putatively individualized and at least partially subjective process to cull the remaining job categories; and (2) the Agency's imposition of a bachelor's degree requirement on the new FSW position, the duties of which were a close match with the work the SSAs had long performed successfully without a college degree. The district court granted defendant's motion to dismiss the disparate treatment claim against the firings, allowing the named plaintiffs to proceed with the companion claim of disparate racial impact, and both the impact and treatment claims against the degree requirement. See Davis v. Dist. of Columbia , 949 F. Supp. 2d 1, 14 (D.D.C. 2013).

The court bifurcated discovery and pretrial motions, limiting the first stage to the "existence and statistical validity of group-based disparities caused by" the practices challenged on disparate-impact grounds, as well as to several procedural matters including administrative exhaustion and class certification. See Scheduling Order, Davis v. District of Columbia , No. 10-1564 (D.D.C. Apr. 4, 2013) (Scheduling Order). The court held that plaintiffs met the administrative exhaustion requirement because two plaintiffs' timely-filed Equal Employment Opportunity Commission (EEOC) charges put the Agency on notice of the claims and vicariously satisfied the exhaustion requirement for the remaining plaintiffs. Davis v. District of Columbia , 246 F. Supp. 3d 367, 388-89 (D.D.C. 2017). Plaintiffs asserted an absence of evidence of business necessity to support the District's claims of "agency-wide realignment" and the decision to hire outsiders to the FSW positions, and requested an admission to that effect. J.A. 679. The Agency postponed responding on the ground that plaintiffs sought "information that is neither relevant nor reasonably calculated to lead to the discovery of admissible evidence relevant to an issue within the scope of the first phase of discovery" under the judge's scheduling order. Id . Discovery *1246into and motions testing the Agency's justifications for its layoff choices were for a later stage.

Within the constraints of the bifurcated discovery order, each side retained an expert as to the alleged disparate impact of the challenged practices. The experts framed the issues differently and reached contrary conclusions. Plaintiffs' expert, Dr. Paige Munro, found that the Agency's implementation of the layoffs resulted in a termination rate of 15.5% for African Americans, in contrast to a 5.6% rate for non-African Americans. The racial disparities were even more dramatic, according to Dr. Munro, once she analyzed a new dataset provided by the District, which included more focused and detailed demographic information about the Agency's workforce: The effective termination rate jumped from 277% greater for African Americans than non-African Americans to 444% greater for African Americans as compared to Caucasians.

The District's expert, Dr. Stephen Bronars, found no disproportionate adverse racial impact. He faulted Dr. Munro for calculating the racial disparities in termination rates across the entire agency; according to Dr. Bronars, Dr. Munro's data unreasonably assumed that all positions at the Agency were at risk of cuts. The Agency described itself as conducting an "agency-wide" reduction in force, Defendant's Answer to Third Amended Complaint ¶ 3, J.A. 159, but as Dr. Bronars saw it not all employees were at equal risk of losing their jobs because the District had informed him that it took into account "financial concerns, the reorganization concerns, the realignment of goals, different kind of service models. ..." Bronars Dep. 183-84, J.A. 820. Instead of assessing the impact of the entire package of layoffs, Dr. Bronars characterized the District's action as "7 different sets of layoff decisions," one for each job category that experienced cuts. J.A. 363. He then separately examined the racial impact of the terminations within each affected position.

Dr. Bronars concluded that the Agency's wholesale elimination of the SSA and SWA positions did "not contribute to the statistical significance calculation for adverse impact" because the District terminated every employee in those job categories. J.A. 412. As a consequence, he reasoned, those cuts involved no "excess" termination of African American employees. J.A. 367, 412. Dr. Bronars' statistical significance calculation also excluded any layoffs from job categories occupied exclusively by African American employees, again reasoning that there could be no "excess" termination of African Americans from those categories. J.A. 367. Setting aside all of those layoffs of African American employees, Dr. Bronars applied his job-category-specific methodology to find no statistically significant racial impact resulting from cuts within the remaining categories.

Following the close of phase I discovery, plaintiffs moved for class certification and the District moved for summary judgment. The district court granted summary judgment for the Agency on all issues.

Regarding plaintiffs' disparate impact challenge, the Agency contended that plaintiffs' expert evidence of statistical disparity was inadequate. See Def.'s Memo in Supp. of Motion for Summ. J. 2, 20-23, J.A. 207, 225-28. Alternatively, the Agency argued that the Agency's termination decisions were not subject to Title VII scrutiny for disparate racial impact: The Agency contended that its decisions were not actionable because they involved no objective "test or requirement," but only a series of subjective, contextual judgments made in "multiple individual decisions by the agency leadership" that it claimed cannot be challenged on a disparate-impact theory.

*1247Id . at 19-20, J.A. 224-25 (contending that "subjective decisions" are not practices subject to challenge for their disparate impact (quoting Leichihman v. Pickwick Int'l , 814 F.2d 1263, 1269 n.5 (8th Cir. 1987), abrogated by Watson v. Fort Worth Bank & Tr. , 487 U.S. 977, 989-91, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988) (plurality opinion), and citing Combs v. Grand Victoria Casino & Resort, No. 1:08-CV-00414-RLY-JMS, 2008 WL 4452460, at *3 (S.D. Ind. Sept. 30, 2008) )).

In addition to dwelling on the (erroneous) proposition that only objective employment criteria are subject to disparate impact scrutiny, see Watson , 487 U.S. at 989-91, 108 S.Ct. 2777, both parties got inexplicably sidetracked into arguing over whether a "facially neutral" employment policy had been identified, see J.A. 224-25 (Agency: "The RIF Was Not A Facially Neutral Employment Policy"); id . at 592-93 (Plaintiffs: "The RIF was a facially neutral policy"); see also Davis , 246 F. Supp. 3d at 395. The point of doctrinal references to "facially neutral employment practices" is not to make facial neutrality an element of proof in disparate-impact cases, but merely to make clear that-even though they may lack the overtly or intentionally discriminatory character of practices constituting disparate treatment-facially neutral practices, too, may be challenged under Title VII. Watson , 487 U.S. at 988, 108 S.Ct. 2777 ; Int'l Broth. of Teamsters v. United States , 431 U.S. 324, 349, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977) ("[T]he Court has repeatedly held that a prima facie Title VII violation may be established by policies or practices that are neutral on their face and in intent but that nonetheless discriminate in effect against a particular group."). Indeed, nothing prevents a plaintiff from challenging a practice as a Title VII violation because it is facially discriminatory and, alternatively, has a disparate impact.

Without reaching the statistical evidence in the competing expert reports, the district court granted summary judgment to the Agency on the threshold ground that "plaintiffs have failed to identify a specific employment practice" actionable under a disparate-impact theory. Davis , 246 F. Supp. 3d at 394. As to plaintiffs' challenges to the FSW's bachelor's degree requirement, the district court granted judgment to the Agency because plaintiffs had failed to present evidence regarding the qualified labor pool. Id. at 399-401. Because plaintiffs chose to rest their disparate treatment claim exclusively on an inference of discriminatory purpose arising from statistical disparity, and no such disparity could be shown in the absence of evidence regarding the qualified labor pool, the court granted summary judgment on that claim as well. Id.

The court also ruled in favor of the District on issues pertaining only to certain plaintiffs. It held that two plaintiffs lacked standing to challenge the FSW's bachelor's degree requirement because they hold such degrees, id. at 387-88, and that two plaintiffs-one of those with a bachelor's degree, plus a third-were judicially estopped from participating in this lawsuit by their failures to disclose their discrimination claims among their assets in their personal bankruptcy cases, id. at 384-87. In the absence of any surviving claim, the district court denied as moot plaintiffs' motion for class certification. Id. at 401.

Plaintiffs appealed. They limit their appeal to the district court's grant of summary judgment on the race discrimination class claims, plus the individual standing and estoppel issues.

II. Analysis

Our review of the district court's grant of summary judgment is de novo .

*1248Aka v. Wash. Hosp. Ctr. , 156 F.3d 1284, 1288 (D.C. Cir. 1998) (en banc). Summary judgment is warranted where 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). That is, a movant is entitled to summary judgment when, drawing all inferences in favor of the non-movant, a reasonable jury could not return a verdict in the non-movant's favor. Celotex Corp. v. Catrett , 477 U.S. 317, 326-27, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986) ; Anderson v. Liberty Lobby, Inc. , 477 U.S. 242, 248, 251-52, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986).

Title VII of the Civil Rights Act makes it unlawful for an employer "to fail or refuse to hire or to discharge any individual ... because of such individual's race." 42 U.S.C. § 2000e-2(a)(1). The statute bars "intentional discrimination and artificial, arbitrary, or unnecessary barriers" that stand in the way of "equal opportunity" without regard to race. Segar v. Smith , 738 F.2d 1249, 1258 (D.C. Cir. 1984). Thus, a plaintiff may establish racial discrimination in violation of Title VII by proving either that the employer acted with a discriminatory motive (a "disparate treatment" claim), or that its action was the result of a process that, while apparently "fair in form," was "discriminatory in operation" (a "disparate impact" claim). Griggs v. Duke Power Co. , 401 U.S. 424, 431, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971) ; see Anderson v. Zubieta , 180 F.3d 329, 338 (D.C. Cir. 1999). As the parties recognized below, the District of Columbia Human Rights Act tracks Title VII in all respects relevant to this case. See Davis , 246 F. Supp. 3d at 393-94. We therefore treat plaintiffs' claims under District of Columbia law as coextensive with their federal claims.

We first address plaintiffs' claim that the particular practices by which the District carried out its reduction in force had a racially disparate impact in violation of Title VII. We then turn to their challenge to the degree requirement attached to the new FSW position. Finally, we address the district court's dismissal of three individual plaintiffs' claims.

A. Identification of Particular Practices Subject to Disparate Impact Analysis

The district court held that plaintiffs failed at the summary judgment stage to make out a prima facie case of disparate impact under Title VII on the ground that a RIF is not a "particular employment practice" under Title VII. 42 U.S.C. § 2000e-2(k)(1)(A)(i) ; see Davis , 246 F. Supp. 3d at 394-97. The court thought the plaintiffs identified only "an overall decisionmaking process," which did not meet the statutory requirement to identify the particular practices that caused them to lose their jobs. Davis , 246 F. Supp. 3d at 394-95. The court thus concluded that the practices by which the Agency fired plaintiffs could not be reviewed for adverse racial impact under Title VII. That decision was in error.

A disparate impact claim contends that an observed disparity caused by a particular employment practice cannot be justified as necessary to the employer's business. The purpose of disparate impact analysis under Title VII is to permit plaintiffs to challenge "practices, procedures, or tests" that may be "neutral on their face, and even neutral in terms of intent," but that disproportionately harm members of a protected class. Griggs , 401 U.S. at 430, 91 S.Ct. 849. Whereas disparate treatment requires a showing of discriminatory motive, disparate impact supports liability in the absence of proof of invidious intent, based on evidence that the challenged practices have a disproportionately adverse *1249effect on the plaintiffs that cannot be justified as necessary to an employer's business.

In calling on disparate-impact plaintiffs to identify the particular employment practices they challenge, the law "goes beyond the need to show that there are statistical disparities in the employer's work force." Watson , 487 U.S. at 994, 108 S.Ct. 2777 ; see Wal-Mart Stores, Inc. v. Dukes , 564 U.S. 338, 357, 131 S.Ct. 2541, 180 L.Ed.2d 374 (2011) (pointing to "overall sex-based disparity" in workforce is not enough). The requirement to identify the employment practice or practices responsible for the shortfall guards against holding employers "liable for 'the myriad of innocent causes that may lead to statistical imbalances' " in a given workforce. Smith v. City of Jackson , 544 U.S. at 241, 125 S.Ct. 1536 (quoting Wards Cove Packing Co. v. Atonio , 490 U.S. 642, 657, 109 S.Ct. 2115, 104 L.Ed.2d 733 (1989) ). As the Supreme Court explained:

Our disparate-impact cases have always focused on the impact of particular hiring practices on employment opportunities for minorities. Just as an employer cannot escape liability under Title VII by demonstrating that, "at the bottom line," his work force is racially balanced (where particular hiring practices may operate to deprive minorities of employment opportunities), see Connecticut v. Teal , 457 U.S. [440,] 450, 102 S.Ct. 2525, 73 L.Ed.2d 130 [ (1982) ], a Title VII plaintiff does not make out a case of disparate impact simply by showing that, "at the bottom line," there is racial imbalance in the work force.

Wards Cove , 490 U.S. at 656-57, 109 S.Ct. 2115. Thus, when unidentified events or a "myriad of innocent causes" cumulate over time to result in racial, ethnic, religious, or gendered statistical imbalances in a workforce, those imbalances are not rendered susceptible to a Title VII challenge by the mere measurement of statistical shortfall. But where the object of the suit is an identifiable practice, criterion, or bundle of criteria behind a specified employment event like the rash of contemporaneous layoffs challenged here, disparate impact analysis applies.

To state a disparate impact claim, plaintiffs are "responsible for isolating and identifying the specific employment practices that are allegedly responsible for any observed statistical disparities." Wards Cove , 490 U.S. at 656, 109 S.Ct. 2115 (quoting Watson , 487 U.S. at 994, 108 S.Ct. 2777 ) (emphasis added). An actionable "specific employment practice" might be a set of "subjective criteria" such as hiring based on personal networks or firing based on a manager's subjective sense of who best to retain; or it might be comprised of "more rigid standardized rules or tests" like height, weight, length-of-service, or performance-based standards. Id. (quoting Watson , 487 U.S. at 994, 108 S.Ct. 2777 ). Disparate impact analysis is "no less applicable to subjective employment criteria than to objective or standardized tests." Watson , 487 U.S. at 990, 108 S.Ct. 2777 ; accord Wal-Mart Stores, Inc. v. Dukes , 564 U.S. 338, 355, 131 S.Ct. 2541, 180 L.Ed.2d 374 (2011) ("[A]n employer's undisciplined system of subjective decisionmaking [can have] precisely the same effects as a system pervaded by impermissible intentional discrimination." (quoting Watson , 487 U.S. at 990-91, 108 S.Ct. 2777 )). A combination of subjective and objective determinants, too, can count as a sufficiently specific employment practice. Watson , 487 U.S. at 990, 994, 108 S.Ct. 2777.

There is no mystery in this case as to the layoff practices plaintiffs challenge: the *1250Agency's choices to (a) target the SWA and SSA job categories for elimination; and (b) allow managers to make putatively individualized, discretionary and subjective choices of which positions to winnow from other units. See Third Am. Compl. ¶¶ 56, 58, 60 (alleging elimination of SWA and SSA jobs); Appellants' Br. 25-27 (contending that the Agency "left the livelihood of African American employees in the hands of multiple supervisors without requiring those supervisors use any uniform criteria or standardized guidance when making termination decisions"); Oral Arg. 15:14-20 (the layoffs were "carried out in an undisciplined and subjective way"). Indeed, although its stated reasons have yet to be tested through discovery, the District has acknowledged that the Agency employees to be fired were "not identified through the use of uniform criteria ... but rather through multiple individual decisions by the agency leadership." Def.'s Memo in Supp. of Summ. J., J.A. 225; see id. 224-25 (urging district court-erroneously-to reject plaintiffs' impact claim because "subjective decisions" are not practices subject to challenge for their disparate impact (citing Leichihman v. Pickwick Int'l , 814 F.2d 1263, 1270 n.4 (8th Cir. 1987)abrogated by Watson , 487 U.S. at 989-90, 108 S.Ct. 2777 )). As the Agency itself describes it, the procedures for culling jobs fit Watson 's description of "an employer's undisciplined system of subjective decisionmaking" as to which "it is difficult to see why Title VII's proscription against discriminatory actions should not apply." 487 U.S. at 990-91, 108 S.Ct. 2777.

This is the first time this court has been asked whether a RIF or, more precisely, the practices through which an employer implements a RIF are subject to disparate-impact review under Title VII, but we see no basis to exempt such practices from otherwise-applicable law. Our analysis in Aliotta v. Bair , 614 F.3d 556 (D.C. Cir. 2010), assumed without deciding that a targeted group of layoffs pursuant to a RIF could be reviewed for potential disparate impact. Although the district court in that case had held that the plaintiff failed to identify "a specific adverse employment practice within the 2005 downsizing" to support his claim of age-based disparate impact, Aliotta v. Bair , 576 F. Supp. 2d 113, 127 (D.D.C. 2008), we did not endorse that reasoning on appeal. We never questioned whether the employer's method of carrying out the RIF was a tenable subject of analysis. Rather, once we isolated layoffs representing "the independent effect of the 2005 RIF itself" from contemporaneous voluntary buyouts, plaintiff's claim failed because the statistics showed a disadvantage to younger employees rather than to the older group to which the plaintiff belonged. Aliotta , 614 F.3d at 569-70.

Courts that have applied Title VII in the context of RIFs have shown how to analyze the layoffs involved as a "particular employment practice." They go beyond the general concept of a "RIF" to identify actionable practices of "selecting only certain (predominantly female) departments," Shollenbarger v. Planes Moving & Storage , 297 Fed. App'x 483, 486 (6th Cir. 2008), or of focusing cuts on offices where "black employees are concentrated," Council 31, Am. Fed'n of State, Cty. & Mun. Emps. v. Ward , 978 F.2d 373, 375, 377-78 (7th Cir. 1992) ; see also Sengupta v. Morrison-Knudsen Co. , 804 F.2d 1072, 1073-74, 1076-77 (9th Cir. 1986) (analyzing a reduction in force for disparate impact). When an employer cuts back on its workforce by "targeting" demographically disproportionate departments for layoffs, that practice means that "the likelihood of selecting a[n individual in a protected class] increase[s]." Shollenbarger , 297 Fed. App'x at 486. Such a practice is what plaintiffs here identify, and is the kind of practice *1251the disparate impact theory of discrimination exists to scrutinize. It is consistent with precedent, and neither unwieldy nor unfair, to treat the processes by which the Agency identified plaintiffs' jobs for elimination as "particular employment practice[s]" under section 2000e-2(k)(1)(A)(i).

We are mindful that, as important as it is, the requirement that plaintiffs identify a particular employment practice does not alone do all the work of shielding employers from liability for mere racial (or other protected-class) imbalance in a workforce: Plaintiffs also have the burden to show "caus[ation]," 42 U.S.C. § 2000e-2(k)(1)(B)(ii), and, pursuant to Title VII's familiar burden-shifting framework, the defendant has the opportunity to demonstrate "business necessity." 42 U.S.C. § 2000e-2(k)(1)(A)(i) ; see Wards Cove , 490 U.S. at 656, 109 S.Ct. 2115 ; Watson , 487 U.S. at 994-98, 108 S.Ct. 2777. The Supreme Court in Texas Department of Housing & Community Affairs v. Inclusive Communities Project, Inc. , --- U.S. ----, 135 S. Ct. 2507, 192 L.Ed.2d 514 (2015), emphasized the important constraint the causation requirement imposes on disparate impact claims. In holding that the Fair Housing Act provides for disparate impact liability, the Court affirmed the impact theory against charges that it is unreasonably expansive, largely by stressing that a "robust causality requirement ... protects defendants from being held liable for racial disparities they did not create." Id. at 2523. And, ultimately, "the 'touchstone' for disparate-impact liability is the lack of 'business necessity' " supporting the challenged practice. Ricci v. DeStefano , 557 U.S. 557, 578, 129 S.Ct. 2658, 174 L.Ed.2d 490 (2009). Thus, "[h]ow far this prima facie showing will carry the plaintiff toward its ultimate burden of persuasion depends on both the strength of the plaintiffs' evidence and the nature of the defendant's response." Segar , 738 F.2d at 1267.

The district court appeared to grasp the clear racial effect of the Agency's method of implementing its reduction in force. Plaintiffs' expert, Dr. Munro, presented statistics showing dramatic over-representation of African American employees in the positions chosen for elimination. The "targeted positions/divisions" were disproportionately occupied by African Americans. J.A. 380. And the termination rate was 444% higher for the African American employees than Caucasians. See J.A. 377. The court accurately observed that record evidence showed that the Agency's job "cuts were not equally distributed"; the layoff "left some positions or divisions relatively unscathed, while it completely eliminated other positions." Davis , 246 F. Supp. 3d at 395. "In fact," the district court expounded, "cuts to just two positions," held almost entirely by African American employees, "constituted the majority of terminations in the RIF." Id. (citing Def.'s SOF ¶ 20). The district court declined to address the statistics, however, because it thought plaintiffs had failed to identify a "specific" employment practice. The court cautioned that "simply pointing to a RIF generally is not sufficient," id . at 395, and drew on out-of-circuit, primarily district-court decisions to support its rule that a RIF is not a particular practice subject to disparate-impact challenge under Title VII. See Davis , 246 F. Supp. 3d at 395 (citing Powell v. Dallas Morning News , 776 F. Supp. 2d 240, 258 (N.D. Tex. 2011) ; Zawacki v. Realogy Corp. , 628 F. Supp. 2d 274 (D. Conn. 2009) ; Mustelier v. Equifax, Inc. , No. Civ. 08-1008, 2009 WL 890468, at *6 (D.P.R. Mar. 25, 2009) ; Kourofsky v. Genencor Int'l, Inc. , 459 F. Supp. 2d 206, 215 (W.D.N.Y. 2006) ); see also id. (citing Leichihman , 814 F.2d at 1269 n.5 ).

We need not generally decide whether a RIF as such might ever be a "particular employment practice" under *1252section 2000e-2(k)(1)(A)(i). Cf. Dissenting Op. at 1259. Terminating a large group of employees in a compressed timeframe is clearly an adverse employment action within the meaning of Title VII, and an employer's assertion that the firings were a "RIF" required by budget cuts does not somehow immunize them from Title VII scrutiny. To the extent that a completed RIF is an identified event comprising selection and termination of a rash of employees, it is a far cry from the challenges to bottom-line "racial imbalance in the work force" that precedent and our colleague eschew. Dissenting Op. at 1259 (quoting Wards Cove , 490 U.S. at 657, 109 S.Ct. 2115 ). In our view, however, it is more confusing than clarifying to ask whether RIFs in general are "particular employment practices" under Title VII. For one thing, "RIF" is not a legally defined term. It is often used as a shorthand for downsizing a workforce, and can refer to the general anticipation ("We'll need to RIF ten percent of our employees") or completion thereof ("I was RIF'd last month"). And, because the term is often used in the context of economic exigency, it unhelpfully invites prejudgment of the question whether the procedures used to determine which employees to let go were supported by business necessity.

This case does not present that question because plaintiffs' claim is not that the RIF in the abstract was unlawful. Nor, for that matter, do they take issue with the decision that laying off 52 employees would enable the Agency to comply with the FY 2011 budget cut. What calls for identification and scrutiny, and what plaintiffs challenge here, is not the Agency's decision to reduce its workforce, but the process the Agency used to select positions for the chopping block. See Third Am. Compl. ¶¶ 56, 58, 60 (alleging elimination of SWA and SSA jobs); Appellants' Br. 25-27 (contending that the Agency disproportionately fired African American employees without requiring supervisors to "use any uniform criteria or standardized guidance when making termination decisions"); Oral Arg. 15:14-20 (the layoffs were "carried out in an undisciplined and subjective way"). And those processes are susceptible of challenge under disparate impact precedents. See Wards Cove , 490 U.S. at 656, 109 S.Ct. 2115 ; Watson , 487 U.S. at 990-91, 108 S.Ct. 2777.

The dissent posits that plaintiffs only challenge the RIF "writ large," and not the Agency's particular means of implementing it. Dissenting Op. at 1261; see id . at 1261-63. We disagree: the record makes clear that plaintiffs' challenge to "the RIF" is shorthand for its attack on the specific processes the Agency used in order to cut positions to meet its budget shortfall.1 To be sure, neither party's briefing has been entirely clear, and in the district court each party on occasion advanced legally erroneous propositions and focused on points of no apparent relevance. Perhaps not surprisingly, plaintiffs brought this claim into better focus on appeal. See, e.g. , Appellants' Br. 12, 21 (discussing the Agency's "implementation of the RIF"). In any event, we take the Agency at its word that, to carry out its workforce reduction, it chose to target the SWA and SSA job categories for elimination, and cut jobs from other categories according to "multiple *1253individual decisions" by management. J.A. 235. Because those processes-which plaintiffs identify on appeal and which the Agency itself says it used-are "properly before the court," we are "not limited to the particular legal theories advanced by the parties, but rather retain[ ] the independent power to identify and apply the proper construction of governing law." Kamen v. Kemper Fin. Servs. Inc. , 500 U.S. 90, 99, 111 S.Ct. 1711, 114 L.Ed.2d 152 (1991) ; see also Lebron v. Nat'l R.R. Passenger Corp ., 513 U.S. 374, 379, 115 S.Ct. 961, 130 L.Ed.2d 902 (1995) ("Our traditional rule is that once a federal claim is properly presented ... parties are not limited to the precise arguments they made below.") (internal quotations omitted).

The dissent is right that the framing of the specific employment practice matters, see Dissenting Op. at 1262-62, but not exactly as to how. The distinction between challenging the RIF simpliciter and challenging the processes by which it was implemented could have litigation consequences down the road. But on the disparate-impact question, the latter framing does not "render[ ] irrelevant the agency-wide statistics." Id . at 1263. Either way, the agency-wide statistics speak to the threshold issue the district court raised and the parties addressed: Whether, in shrinking an agency of 892 people by at least 52 employees, the identified mechanisms by which defendants did so had a statistically significant racial impact.

We take no stance here on other potential hurdles the dissent identifies with challenges to subjective employment decisions. Id . Because the district court bifurcated discovery to first tee up whether there was a statistical disparity, see Scheduling Order, and postponed inquiry into other questions such as business justification and commonality that the dissent highlights, Dissenting Op. at 1263, neither that court nor this one is yet in a position to pass on those points. The Agency's reasons for its actions evince a fact-intensive inquiry into business justification. The dissent's doubts that the defendants' practices tie the class together more closely than the claimed diffuse, nationwide practice at issue in Wal-Mart could be relevant to a motion for class certification. Those matters will be explored only if, on remand, the district court finds that the practices by which the Agency implemented the layoffs had a statistically significant racial impact. On the matter now before us, the dissent does not dispute that an identified practice used to implement a RIF may be challenged for its disparate impact, but would hold that plaintiffs forfeited that claim. See Dissenting Op. at 1261-62.

As the district court chose to manage it, this case remains at the first step: plaintiffs' prima facie case. But the Agency's expert, Dr. Bronars, presumed operational justification for (and therefore excluded from analysis) the Agency's selection of certain offices for downsizing and its wholesale elimination of the SSA and SWA jobs. The premise of Dr. Bronars's statistics, the Agency conceded at oral argument, put the "cart before the horse" by assuming the very facts that a successful statistical showing by plaintiffs would next require the Agency to show: that the reason the Agency targeted certain positions for elimination was justified by business necessity. Oral Arg. 36:54-37:29; see id. at 22:23-39 (plaintiffs' counsel explaining same). If on remand plaintiffs clear the statistical hurdle, the parties will have an opportunity after appropriate discovery to address whether the Agency's execution of the reduction in force was justified by business necessity. Justification supporting elimination or downsizing of certain offices *1254might at that point be seen to respond to the relevant statistical showing.

Because plaintiffs have leveled their disparate impact challenge against the particular target of the Agency's process for cutting and culling job categories, we reverse the district court's decision to the contrary, and the accompanying denial on mootness grounds of the motion for class certification, and remand for further proceedings consistent with this opinion.

B. Challenges to the Bachelor's Degree Requirement

The district court also granted summary judgment in the Agency's favor on plaintiffs' disparate impact and disparate treatment challenges to the bachelor's degree requirement associated with the Family Support Worker position. The court correctly concluded that plaintiffs failed to raise a triable issue.

1. Disparate Impact

The district court held that plaintiffs failed to meet their burden to identify a race-based statistical disparity potentially caused by the challenged degree requirement. Davis , 246 F. Supp. 3d at 398-99. This is a defect that plaintiffs as much as admitted on appeal by asking this court to take judicial notice of extra-record census data showing general racial disparities among degree holders in the District of Columbia. See Appellants' Reply 18-20 & n.5. The burden rests with plaintiffs to show the racial disparity, see Wards Cove , 490 U.S. at 651, 109 S.Ct. 2115, and we are not prepared on appeal to take judicial notice of census data not brought to the district court's attention. More fundamentally, that data could not by its own terms fill the gap in this record. To show an adverse racial impact attributable to imposition of the challenged degree requirement, plaintiffs would have to identify the qualified applicant pool. But there is no evidence that these jobs are positions for which all District of Columbia resident applicants would necessarily be qualified. Because plaintiffs failed to make the relevant showing, we affirm the district court's grant of summary judgment in the Agency's favor on this ground.

The district court also doubted the plaintiffs' race-based challenges to the degree requirement for the FSW position on the ground that all of the Agency's FSW hires were African American. Davis , 246 F. Supp. 3d at 399. The Agency has since correctly conceded that these plaintiffs' disparate impact claim cannot be defeated by evidence that the Agency ultimately hired other African Americans to fill the FSW jobs. See Appellee's Br. 43-44; Oral Arg. 49:39-50:06. The Supreme Court in Connecticut v. Teal held that, even where plaintiffs are replaced by persons in their protected class, such a " 'bottom line' does not preclude [plaintiff] employees from establishing a prima facie case, nor does it provide [a defendant] employer with a defense to such a case." 457 U.S. at 442, 102 S.Ct. 2525. Despite the surface appeal of "measur[ing] [disparate impact] ... at the bottom line," the Court recognized that doing so "ignores the fact that Title VII guarantees these individual black respondents the opportunity to compete equally with white workers on the basis of job-related criteria." Id. at 451, 102 S.Ct. 2525 (emphasis in original). To the extent that the "bottom line" was alluded to here, however, it did not detract from the correctness of the district court's holding that plaintiffs lacked evidentiary support for this claim.

2. Disparate Treatment

The district court also granted summary judgment in the Agency's favor on the disparate treatment challenge to *1255the FSW's degree requirement. That disparate treatment claim relied solely on statistical disparities. See Davis , 246 F. Supp. 3d at 400-01 ; see also Oral Arg. 23:10-20; id. at 24:15-30, 25:54-26:26 (confirming that plaintiffs "are relying entirely on statistical evidence" of disparate impact to establish disparate treatment). Although it is possible to make a prima facie case of disparate treatment based solely on statistics, see Teamsters , 431 U.S. at 339, 97 S.Ct. 1843, in order to do so plaintiffs must present a "significant" pattern of discrimination unexplainable on grounds other than race, see Aliotta , 614 F.3d at 562. And in the absence of information about the composition of the qualified applicant pool, the district court correctly found no cognizable disparate impact, let alone a significant one. Davis , 246 F. Supp. 3d at 400-01.

Plaintiffs could have sought to prove their disparate treatment claim with other circumstantial or direct evidence that the Agency implemented the degree requirement for racial reasons. And because the district court bifurcated discovery, limiting the first phase to statistical evidence of disparate impact, plaintiffs had not yet had a chance to develop record support regarding the relevant decision makers' motivation for creating the FSW post and its attendant degree requirement. See Scheduling Order. But, by conceding that their claim was based solely on statistical evidence and failing to oppose summary judgment on this claim with a Rule 56(d) declaration asserting their entitlement to discovery to support it, Plaintiffs forfeited any claim to investigate the motive behind the FSW's degree requirement. As a result, we affirm the district court's grant of summary judgment in the Agency's favor as to the disparate treatment claim against the degree requirement.

C. Estoppel by Bankruptcy Filings

Finally, we consider the district court's dismissal of the claims of some individual plaintiffs. Because no claims regarding the FSW's degree requirement remain, we need not address the district court's holding that two plaintiffs who have bachelor's degrees, Darius Morris and Zaccheus Ajakaiye, lacked standing to challenge that requirement. We therefore turn to the court's decision that two plaintiffs' Title VII claims are barred by estoppel.

The court held that Ajakaiye and another plaintiff, Stephanie Alston, were judicially estopped from proceeding because they had failed to disclose their Title VII claims in their personal bankruptcy proceedings. See Davis , 246 F. Supp. 3d at 385-87. We review only for abuse of discretion the district court's decision to invoke judicial estoppel. Marshall v. Honeywell Tech. Sys. Inc. , 828 F.3d 923, 928 (D.C. Cir. 2016).

Judicial estoppel "prevents a party from asserting a claim in a legal proceeding that is inconsistent with a claim taken by that party in a previous proceeding." Moses v. Howard Univ. Hosp. , 606 F.3d 789, 798 (D.C. Cir. 2010) (quoting New Hampshire v. Maine , 532 U.S. 742, 749, 121 S.Ct. 1808, 149 L.Ed.2d 968 (2001) ). In a bankruptcy petition, a debtor must disclose "all potential claims." Id. at 793. "This means that a debtor is under a duty both to disclose the existence of pending lawsuits when he files a petition in bankruptcy and to amend his petition if circumstances change during the course of the bankruptcy." Id. That duty comports with the bankruptcy estate's authority to control the debtor's assets for the benefit of creditors. The disclosure obligation extends to administrative complaints, including those before the EEOC. Marshall , 828 F.3d at 924-25. A debtor's failure to comply with that duty can trigger *1256judicial estoppel to prevent him from pocketing proceeds of a previously pending but undisclosed suit, which proceeds should have been distributed to creditors in the bankruptcy.

There is no dispute that plaintiffs failed to disclose their potential Title VII claims in their bankruptcy petitions. Ajakaiye filed his bankruptcy petition on July 2, 2010. Two weeks later, on July 16, 2010, he filed the EEOC charge that, on September 16, 2010, germinated into the complaint in this case. The bankruptcy court discharged Ajakaiye's debts on October 14, 2010, without Ajakaiye's ever having disclosed the EEOC proceeding or the ensuing suit. Alston, for her part, filed her bankruptcy petition on May 21, 2013, three years after plaintiffs filed suit in this case; she also failed to disclose the circumstances of her potential claim before her debts were discharged in bankruptcy.

Plaintiffs challenge the application of judicial estoppel to their claims on two grounds. First, they argue that judicial estoppel is an affirmative defense that the District did not raise in its Answer so forfeited. But the doctrine is not only a defense; because it also protects the integrity of the judicial process, a court may invoke judicial estoppel "at its discretion." See New Hampshire , 532 U.S. at 750, 121 S.Ct. 1808 (quoting Russell v. Rolfs , 893 F.2d 1033, 1037 (9th Cir. 1990) ); see also Allen v. C & H Distribs., LLC , 813 F.3d 566, 571 n.4 (5th Cir. 2015) (explaining that the doctrine may be invoked " 'sua sponte' and therefore 'the court is not bound to accept a party's apparent waiver of the doctrine' " (quoting 18 Moore's Federal Practice § 134.34 (3d ed. 2015))).

Plaintiffs also contend that the application of judicial estoppel is inappropriate in cases of an inadvertent or mistaken failure to disclose claims or potential claims. See Marshall , 828 F.3d at 930 (citing New Hampshire , 532 U.S. at 753, 121 S.Ct. 1808 ). It is easy enough to imagine that many people-perhaps these plaintiffs included-do not see potential, or even filed and pending, legal or administrative claims as assets that must be disclosed in bankruptcy; they therefore might innocently fail to list such claims. After all, plaintiffs seeking justice may not be thinking of their yet-to-be-vindicated claims as "assets," especially at the outset of a long and winding litigation road with an uncertain end. But once the estoppel question was raised, plaintiffs (who are here represented by counsel) failed to introduce even their own sworn declarations to support the assertions in the legal briefs that their failures to disclose were inadvertent. See id. at 930-31 (describing a plaintiff's affidavit "stating that when she filed her bankruptcy petition and schedules ... 'I had no knowledge that I was required to list my discrimination administrative proceedings on my bankruptcy petition schedules or on any financial statements' ").

To be sure, the district court could have exercised its discretion differently, given that estoppel "looks toward cold manipulation and not unthinking or confused blunder." Konstantinidis v. Chen , 626 F.2d 933, 939 (D.C. Cir. 1980) (quoting Johnson Serv. Co. v. Transamerica Ins. Co. , 485 F.2d 164, 175 (5th Cir. 1973) ). Ajakaiye had not yet filed even his administrative claim when he sought bankruptcy, and we see no signs of manipulation by either plaintiff. However, given the absence of any evidentiary submissions to support the assertion of mistaken nondisclosure, the district court did not abuse its broad discretion in holding that, on this record, plaintiffs failed to create a genuine dispute of material fact defeating estoppel. See Marshall , 828 F.3d at 932 ; see also id. at 933-34 (Griffith, J., dissenting). We therefore affirm the district court's dismissal of *1257these plaintiffs' claims on estoppel grounds.

* * *

We reverse the district court's entry of summary judgment for the District as to plaintiffs' disparate impact challenge to the firings under both Title VII and the District of Columbia Human Rights Act, and the associated denial of the motion for class certification, and remand for further proceedings consistent with this opinion. We affirm the district court's grant of summary judgment on the remaining Title VII and District of Columbia Human Rights Act claims. We also affirm the district court's dismissal of all claims by plaintiffs Alston and Ajakaiye on judicial estoppel grounds.

So ordered.

This case arises out of a reduction in force (RIF) conducted by the District of Columbia Child and Family Services Agency. The plaintiffs challenged the RIF as the source of an alleged disparate impact on black employees. On summary judgment, the district court held that the RIF itself-a series of layoffs-was not a particular employment practice subject to disparate-impact challenge under Title VII.

My colleagues remand for the district court to consider other challenges to more specific practices through which the RIF might have been implemented-decisions to eliminate certain job categories and to permit individual supervisors to use subjective criteria in making layoff decisions. Because the plaintiffs disavowed those challenges, and because the challenge that they made lacks merit, I would affirm the summary judgment in its entirety.

I

Title VII of the Civil Rights Act of 1964 makes it unlawful for an employer "to limit, segregate, or classify" employees in any way that would "adversely affect" an individual's "status as an employee, because of ... race." 42 U.S.C. § 2000e-2(a)(2). In Griggs v. Duke Power Co. , 401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971), the Supreme Court construed this provision to prohibit employment practices that, without business justification, produce adverse impacts correlated to race. The Court elaborated on the scope of disparate-impact liability under Title VII in Watson v. Fort Worth Bank & Trust , 487 U.S. 977, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988), and Wards Cove Packing Co. v. Atonio , 490 U.S. 642, 109 S.Ct. 2115, 104 L.Ed.2d 733 (1989).

Congress responded to these decisions with the Civil Rights Act of 1991, which codified disparate-impact liability and specified its parameters. Pub. L. No. 102-166, § 105, 105 Stat. 1071, 1074-75. As a result, Title VII now provides that "[a]n unlawful employment practice based on disparate impact is established ... only if," as relevant here, the plaintiff "demonstrates" that the employer "uses a particular employment practice that causes a disparate impact on the basis of race." 42 U.S.C. § 2000e-2(k)(1)(A) & (A)(i). Moreover, when challenging multiple practices, the plaintiff "shall demonstrate that each particular challenged employment practice causes a disparate impact," unless the plaintiff "can demonstrate ... that the elements of [an employer's] decisionmaking process are not capable of separation for analysis." Id. § 2000e-2(k)(1)(B)(i).

The D.C. Child and Family Services Agency provides services to abused and neglected children. In 2009 and 2010, the City Council reduced the Agency's budget by $ 37.4 million and lowered its cap on *1258employees from 940 to 840. As a result, the Agency undertook a RIF in which 115 employees lost their jobs. About 70 of these employees were Social Worker Associates (SWAs) or Social Service Assistants (SSAs). The others worked in various divisions throughout the Agency.

The plaintiffs are former employees terminated during the RIF. They contend that the RIF writ large produced an unlawful disparate impact on black employees, and they seek to represent a putative class of all employees terminated during the RIF, including a subclass of all terminated black employees. Between 2010 and 2015, the plaintiffs filed four different complaints, conducted discovery on the alleged disparate impact, and produced two expert reports addressing it. The plaintiffs' expert sought to measure the effect of the RIF as a whole, by comparing the racial composition of the terminated employees to that of the overall Agency workforce. See Expert Report of Dr. Paige Munro, Davis v. District of Columbia , No. 10-cv-1564 (D.D.C.), ECF Doc. 146-3, Ex. I at 1-4; Rebuttal Report of Dr. Paige Munro, ECF Doc. 146-4, Ex. K at 2-3.

After all of this, the District moved for summary judgment. As relevant here, it argued that the plaintiffs had neither identified a "particular employment practice" subject to disparate-impact scrutiny nor produced evidence of any statistical disparity caused by such a practice. See Def.'s Mem. Supp. Mot. Summ. J., ECF Doc. 146 at 18-23. In response, the plaintiffs argued that "the RIF" was the challenged "particular employment practice," which produced a disparate impact because the Agency terminated 15.5% of its black employees but only 5.6% of its other employees. See Pls.' Mem. Opp'n Def.'s Mot. Summ. J., ECF Doc. 148 at 37-41.

The district court granted summary judgment for the District. It ruled that the plaintiffs had failed to establish a prima facie case because "the RIF" was not a "particular employment practice" subject to challenge under Title VII's disparate-impact provisions. Davis v. District of Columbia , 246 F. Supp. 3d 367, 393-97 (D.D.C. 2017).

II

The Supreme Court has made clear that, for disparate-impact claims, "the plaintiff's burden in establishing a prima facie case goes beyond the need to show that there are statistical disparities in the employer's work force." Wards Cove , 490 U.S. at 656, 109 S.Ct. 2115 (quoting Watson , 487 U.S. at 994, 108 S.Ct. 2777 (plurality opinion)). If the prima facie case required nothing more, employers would face potential liability for "the myriad of innocent causes that may lead to statistical imbalances in the composition of their work forces." Id . at 657, 109 S.Ct. 2115 (quoting Watson , 487 U.S. at 992, 108 S.Ct. 2777 (plurality opinion)). As a result, "disparate-impact liability might cause race to be used and considered in a pervasive way and 'would almost inexorably lead' governmental or private entities to use 'numerical quotas.' " Tex. Dep't of Hous. & Cmty. Affairs v. Inclusive Cmtys. Project, Inc. , --- U.S. ----, 135 S. Ct. 2507, 2523, 192 L.Ed.2d 514 (2015) (quoting Wards Cove , 490 U.S. at 653, 109 S.Ct. 2115 ). But racial balancing is "far from the intent of Title VII," Albemarle Paper Co. v. Moody , 422 U.S. 405, 449, 95 S.Ct. 2362, 45 L.Ed.2d 280 (1975) (Blackmun, J., concurring in judgment), as the statute itself makes clear: "Nothing contained in this subchapter shall be interpreted to require any employer ... to grant preferential treatment ... on account of an imbalance ... with respect to the total number or percentage of persons of any race" employed. 42 U.S.C. § 2000e-2(j). Moreover, if disparate-impact *1259liability effectively compelled racial balancing regardless of the qualification of individual employees, it would be at war with the more fundamental prohibition against disparate treatment "because of ... race," id.§ 2000e-2(a)(1) ; see Ricci v. DeStefano , 557 U.S. 557, 577-85, 129 S.Ct. 2658, 174 L.Ed.2d 490 (2009), and would raise "serious constitutional questions," Inclusive Cmtys. , 135 S. Ct. at 2523 ; see Ricci , 557 U.S. at 594-96, 129 S.Ct. 2658 (Scalia, J., concurring).

For these reasons, Wards Cove required disparate-impact plaintiffs to "begin by identifying the specific employment practice that is challenged," 490 U.S. at 656, 109 S.Ct. 2115 (quoting Watson , 487 U.S. at 994, 108 S.Ct. 2777 (plurality opinion)), and then to prove that it caused the disparate impact, id . at 657, 109 S.Ct. 2115 ("As a general matter, a plaintiff must demonstrate that it is the application of a specific or particular employment practice that has created the disparate impact under attack."); see also Meacham v. Knolls Atomic Power Lab., Inc. , 554 U.S. 84, 100, 128 S.Ct. 2395, 171 L.Ed.2d 283 (2008) (the plaintiff must "do more" than "point to a generalized policy that leads to" a disparate impact (cleaned up)). Congress later codified the requirement that the plaintiff identify "a particular employment practice that causes a disparate impact." 42 U.S.C. § 2000e-2(k)(1)(A)(i). My colleagues recognize this basic governing legal framework. Ante at 1248-50, 1250-51.

Under this framework, the decision to hire or fire workers cannot by itself form the basis for disparate-impact claims. In Smith v. City of Jackson , 544 U.S. 228, 125 S.Ct. 1536, 161 L.Ed.2d 410 (2005), the Supreme Court affirmed summary judgment for an employer charged with using a pay plan that adversely affected older workers. The Court explained that, under Wards Cove , "it is not enough to simply allege that there is a disparate impact on workers, or point to a generalized policy that leads to such an impact." Id . at 241, 125 S.Ct. 1536. Rather, the plaintiffs had to identify a "specific test, requirement, or practice within the pay plan that has an adverse impact on older workers." Id . (emphasis added). Likewise, in Davis v. Cintas Corp. , 717 F.3d 476 (6th Cir. 2013), the Sixth Circuit held that, for hiring claims, the "particular employment practice" supporting the claim "cannot be the hiring system itself." Id . at 496. And, as my colleagues acknowledge, various district courts have applied these principles to conclude that "a RIF is not a particular practice subject to disparate-impact challenge under Title VII." Ante at ----. These decisions advance the basic purpose of requiring a "particular employment practice" in the first place-to prevent employers from facing large exposure for "the myriad of innocent causes that may lead to statistical imbalances." Wards Cove , 490 U.S. at 657, 109 S.Ct. 2115 (quoting Watson , 487 U.S. at 992, 108 S.Ct. 2777 (plurality opinion)).

Throughout this litigation, the plaintiffs have predicated their disparate-impact claim on "the RIF, as a whole." Appellants' Br. at 20. But as my colleagues explain, "RIF" is simply a "shorthand for downsizing a workforce." Ante at ----. Thus, the gravamen of the plaintiffs' claim is that a set of layoffs caused the racial composition of the Agency's workforce to change. The plaintiffs do not link that change to anything besides the layoffs. So, they have not identified any "particular employment practice" that caused the adverse impact, and their claim runs afoul of a key Title VII teaching: "a Title VII plaintiff does not make out a case of disparate impact simply by showing that, at the bottom line, there is racial imbalance in the work force." Wards Cove , 490 U.S. at 657, 109 S.Ct. 2115 (quotation marks omitted).

*1260In response, the plaintiffs argue that neither pre-1991 Title VII cases (such as Wards Cove ) nor age-discrimination cases (such as Smith ) apply here. Before 1991, Wards Cove governed disparate-impact claims under both Title VII and the Age Discrimination in Employment Act (ADEA). See Smith , 544 U.S. at 240, 125 S.Ct. 1536. The operative ADEA provision, 29 U.S.C. § 623(a)(2), tracks 42 U.S.C. § 2000e-2(a)(2), the original Title VII provision that Griggs construed to give rise to disparate-impact liability. But in the Civil Rights Act of 1991, Congress disapproved certain aspects of Wards Cove , see Pub. L. No. 102-166, § 2(2), 105 Stat. at 1071; expanded disparate-impact liability under Title VII, see id . § 105, 105 Stat. at 1074-75; and left the ADEA unchanged. As a result, the plaintiffs argue, neither Wards Cove nor ADEA cases now govern disparate-impact analysis under Title VII.

The plaintiffs' argument confuses different aspects of the disparate-impact framework. What Congress disapproved was Wards Cove 's formulation of the business-necessity defense, see Pub. L. No. 102-166, § 3(2), 105 Stat. at 1071, which it narrowed. Compare Wards Cove , 490 U.S. at 659, 109 S.Ct. 2115 ("there is no requirement that the challenged practice be 'essential' or 'indispensable' to the employer's business"), and id . (burden of persuasion "remains with the disparate-impact plaintiff"), with 42 U.S.C. § 2000e-2(k)(1)(A)(i) (employer must "demonstrate that the challenged practice is job related for the position in question and consistent with business necessity"). But far from disapproving Wards Cove 's requirement that the plaintiff identify the "specific or particular employment practice that has created the disparate impact under attack," 490 U.S. at 657, 109 S.Ct. 2115, Congress codified that holding, in requiring the plaintiff to "demonstrate[ ]" that the employer "uses a particular employment practice that causes a disparate impact," 42 U.S.C. § 2000e-2(k)(1)(A)(i). The plaintiffs further try to distinguish a "specific" practice under Wards Cove from a "particular" practice under the statute. But that distinction is insubstantial; "specific" and "particular" are synonyms, and Wards Cove used them as such. See 490 U.S. at 657, 109 S.Ct. 2115. Moreover, since 1991, the Supreme Court repeatedly has cited Wards Cove 's holding on the need for a "specific" or "particular" employment practice as good law, see, e.g. , Inclusive Cmtys. , 135 S. Ct. at 2522-23 ; Wal-Mart Stores, Inc. v. Dukes , 564 U.S. 338, 357, 131 S.Ct. 2541, 180 L.Ed.2d 374 (2011), and other courts have recognized it as such, see, e.g., Reyes v. Waples Mobile Home Park Ltd. P'ship , 903 F.3d 415, 425 (4th Cir. 2018) ; Tabor v. Hilti, Inc. , 703 F.3d 1206, 1223 (10th Cir. 2013). Finally, courts addressing the statutory requirement of a "particular employment practice" have cited Title VII and ADEA precedents interchangeably. See, e.g. , Davis , 717 F. 3d at 496-97.

The plaintiffs claim support from three cases, but none helps their position. In Aliotta v. Bair , 614 F.3d 556 (D.C. Cir. 2010), we did not decide whether a RIF qualified as a specific employment practice that could support disparate-impact liability. Instead, we affirmed summary judgment for the employer because the plaintiffs had shown no adverse impact in any event. See id . at 569-70 ("the RIF disproportionately affected younger employees"). Moreover, Council 31 v. Ward , 978 F.2d 373 (7th Cir. 1992), and Shollenbarger v. Planes Moving & Storage , 297 F. App'x 483 (6th Cir. 2008), considered disparate-impact claims predicated not on RIFs as such, but on specific decisions through which the RIFs had been implemented. See id . at 486 ("We conclude that the challenged employment practice of subjecting only certain [predominantly female]

*1261departments to the RIF had a legitimate business justification."); Council 31 , 978 F.2d at 379 (permitting challenge to "the initial decision to concentrate the layoffs in Chicago," where "black employees are concentrated"). Neither case supports the plaintiffs' disparate-impact challenge to "the RIF, as a whole."

Finally, the plaintiffs seek to challenge the RIF under 42 U.S.C. § 2000e-2(k)(1)(B)(i), which provides that if the plaintiff "can demonstrate to the court that the elements of a respondent's decisionmaking process are not capable of separation for analysis," then "the decisionmaking process may be analyzed as one employment practice." In opposing summary judgment, the plaintiffs never invoked that provision, much less sought to create a triable issue on whether the Agency's decisionmaking was "capable of separation for analysis." The plaintiffs thus have forfeited this possible basis for attacking the RIF as a whole. See, e.g. , Chichakli v. Tillerson , 882 F.3d 229, 234 (D.C. Cir. 2018).

III

My colleagues do not seek to defend the plaintiffs' arguments on their own terms, but rather to recast them. They repackage the plaintiffs' broad challenge to the RIF as a much narrower attack on "the Agency's choices to (a) target the SWA and SSA job categories for elimination; and (b) allow managers to make putatively individualized, discretionary and subjective choices of which positions to winnow from other units." Ante at 1250. They thus describe this case as one involving "the practices through which an employer implements a RIF," ante at 1250, along the lines contemplated by Council 31 and Shollenbarger .

What my colleagues describe is not the case that the plaintiffs presented, either below or on appeal. Instead, the plaintiffs have consistently framed their lawsuit as a challenge to the RIF writ large. In the district court, they opposed summary judgment by arguing that "a RIF conducted due to budget cuts is a facially neutral employment practice that can be considered under disparate impact theory." Pls.' Mem., ECF Doc. 148 at 38. Then, they stated unequivocally that "the specific employment practice in this case is the RIF." Id . at 40. On appeal, the plaintiffs continued to direct their challenge to "the RIF, as a whole." Appellants' Br. at 20; accord Appellants' Reply Br. at 6 ("the RIF is the 'particular employment practice' that resulted in a disparate impact based on race"); id . at 8 ("the discriminatory 'particular employment practice' that disparately impacted Plaintiffs-Appellants was the RIF itself"). And at oral argument, the plaintiffs repeated this point no fewer than six times. See Oral Arg. 2:38 ("a RIF is a particular employment practice"); 4:24 ("the RIF was a particular employment practice"); 5:56 (challenging "the decision to lay off employees"); 9:57 ("The target is the layoff decision"); 10:23 ("plaintiffs have met their prima facie duty to identify the RIF as a particular employment practice"); 13:12 ("the RIF does meet Title VII"). As the plaintiffs pursued it, this case was about the RIF itself.

Moreover, in the district court, the plaintiffs expressly disclaimed both of the challenges now suggested by my colleagues. Far from alleging that the SWA and SSA positions were targeted for elimination, the plaintiffs posited that these positions were replaced by a new, functionally identical position of Family Support Worker (FSW). Pls.' Mem., ECF Doc. 148 at 8-11. Furthermore, the plaintiffs described the RIF as an "agency-wide" layoff driven by budget cuts, id . at 37-38, and they argued that "[n]othing" in the record *1262"indicated that the RIF targeted particular positions," id . at 6. As for subjective judgments, the plaintiffs said in no uncertain terms, in a bolded argument heading: "The RIF was not the result of subjective decision-making." Id . at 39. The plaintiffs made these disclaimers not once, but several times. See, e.g. , id . at 3 ("there is not a single piece of evidence on the record that supports" the premise that the RIF was concentrated in "specific offices and divisions within the agency" (quotation marks omitted)); id . ("the agency undertook a neutral cost-cutting move rather than a targeted realignment"); id . at 40 ("Defendant simply cannot show that the RIF was a targeted subjective process"); Pls.' Stmt. Disputed Material Facts, ECF Doc. 148-1 at 3 ("Nothing ... indicated that the RIF targeted particular positions or was conducted as a result of individual decisions."); id . at 4 ("The personnel cuts were not driven by a targeted realignment of the Agency."). Even on appeal, the plaintiffs never sought to predicate their claim on the putative targeting of SWA and SSA positions. They do now briefly argue that the RIF was implemented through an ad hoc system of "subjective" decisionmaking, Appellants' Br. at 25-27, but parties cannot change positions on appeal, see Keepseagle v. Perdue , 856 F.3d 1039, 1053-54 (D.C. Cir. 2017).

The district court clearly identified what the plaintiffs were and were not challenging. As it explained, the plaintiffs "framed the RIF itself " as the challenged employment practice, Blue Mountain Mushroom Co., Inc. v. Monterey Mushroom, Inc. , 246 F. Supp. 2d 394, 396 (E.D. Pa. 2005), and they affirmatively argued "that the RIF was not the result of targeted, subjective decisionmaking," id . at 395. The court further observed: "Plaintiffs might have framed the purported neutral employment practice as the elimination of the SSA and SSW positions," but they "did not elect to do so." Id . at 397. The court thus did not embrace the sweeping conclusion that worries my colleagues-that any "method of implementing" a RIF is immune from disparate-impact scrutiny. Ante at 1251. Rather, the court simply rejected the challenge presented to it, while not addressing other challenges that the plaintiffs could have made but did not.

My colleagues respond that because the Agency's specific "processes" for implementing the RIF are "properly before the court," we are not "limited to the particular legal theories advanced by the parties." Ante at 1253 (quoting Kamen v. Kemper Fin. Servs., Inc. , 500 U.S. 90, 99, 111 S.Ct. 1711, 114 L.Ed.2d 152 (1991) ). But Kamen simply holds that, if the parties properly present an "issue or claim," forfeiture cannot force a court to decide it under an incorrect "construction of governing law." 500 U.S. at 99, 111 S.Ct. 1711. Here, the claims challenging the Agency's specific processes are not properly before the court, for the plaintiffs repeatedly disavowed them. The plaintiffs claim only that "the RIF, as a whole" violated Title VII, and we may readily evaluate that claim based on our own assessment of the governing law. Moreover, the plaintiffs' challenge to "the RIF, as a whole" cannot be deemed an "umbrella claim," Lebron v. Nat'l R.R. Passenger Corp. , 513 U.S. 374, 381, 115 S.Ct. 961, 130 L.Ed.2d 902 (1995) (quotation marks omitted), encompassing challenges to the RIF's constituent parts. That theory runs headlong into Title VII, which prohibits disparate-impact challenges to an employer's overall "decisionmaking process" unless the plaintiff shows that its elements are "not capable of separation for analysis," 42 U.S.C. § 2000e-2(k)(1)(B)(i) -a showing that the plaintiffs did not even attempt below. Despite my colleagues' creative efforts, the governing forfeiture rule here is a pedestrian one:

*1263Where distinct employment practices are at issue, plaintiffs must separately preserve challenges to each one. See, e.g. , Brooks v. Grundmann , 748 F.3d 1273, 1278-79 (D.C. Cir. 2014) ; Sellers v. Deere & Co. , 791 F.3d 938, 943 n.4 (8th Cir. 2015).

The requirement of precisely identifying the challenged employment practice is no mere pleading quibble. To the contrary, it has substantial consequences in any Title VII case. Here, for example, a challenge to the alleged targeting of SWA and SSA positions would have benefitted substantially fewer prospective class members, thus also substantially reducing the expected aggregate recovery for the class. It also would have rendered irrelevant the agency-wide statistics developed by the plaintiffs' own expert, given the "essential requirement" that "the data concern those persons subject to the challenged employment practice." Carpenter v. Boeing Co. , 456 F.3d 1183, 1196 (10th Cir. 2006) ; see also Sengupta v. Morrison-Knudsen Co. , 804 F.2d 1072, 1076 (9th Cir. 1986) ("The impact of a practice on the protected class should generally be measured against the actual pool of employees affected by that practice.").1 Likewise, a challenge to the practice of permitting individual supervisors to evaluate subordinates subjectively would have raised a host of difficulties-including objections that "merely proving that the discretionary system has produced a racial or sexual disparity is not enough" to establish a prima facie case, Wal-Mart , 564 U.S. at 357, 131 S.Ct. 2541 ; that there are obvious business justifications for permitting subjective assessments of employees, id . at 355, 131 S.Ct. 2541 ; and that localized decisionmaking forecloses the possibility of class certification, id . at 348-60, 131 S.Ct. 2541. In this context as elsewhere, our adversarial system holds litigants to their tactical choices, because it presumes that "the parties know what is best for them, and are responsible for advancing the facts and arguments entitling them to relief." Greenlaw v. United States , 554 U.S. 237, 244, 128 S.Ct. 2559, 171 L.Ed.2d 399 (2008) (quotation marks omitted).

In short, the plaintiffs challenged nothing more specific than the RIF. Because the RIF is not a "particular employment practice" within the meaning of Title VII, the plaintiffs failed to establish a prima facie case of disparate-impact discrimination. Accordingly, I respectfully dissent from Part II.A of the Court's opinion, but join the balance of Part II.