Barbara ALIOTTA, et al., Appellants v. Sheila C. BAIR, Chairman, Federal Deposit Insurance Corporation, Appellee, 392 U.S. App. D.C. 239

Opinion for the Court filed by Circuit Judge BROWN. BROWN, Circuit Judge: A group of former employees (class members or Aliotta plaintiffs) of the Federal Deposit Insurance Corporation (FDIC or the Agency) sued the Agency, alleging violation of the Age Discrimination in Employment Act (ADEA), 29 U.S.C. § 633a. Specifically, class members claimed FDIC’s management targeted older employees in a series of downsizings implemented between 1998 and 2005. The district court granted

summary judgment on all claims in FDIC’s favor, determining— after excluding the employees who accepted FDIC’s buyout/early retirement offer from its statistical analysis — that the class members failed to produce evidence from which a jury could reasonably conclude that

(1) FDIC intentionally treated older employees less favorably than younger employees, or (2) that a neutral employment practice fell more harshly on older employees and could not be justified by business necessity. We agree and affirm the judgment of the district court. I The FDIC is an independent federal agency that insures federal bank and savings and loan deposits. It also regulates state-chartered banks, establishes receiverships, and manages assets of failed banking institutions.

FDIC’s workload— especially the workload of the Division of Resolutions and Receiverships (DRR) — is highly correlated with the health of the banking industry: when bank failures increase, FDIC’s workload increases; when bank failures decrease, FDIC’s workload declines. See Aliotta v. Bair, 576 F.Supp.2d 113, 115-16 (D.D.C.2008).

On August 6, 2004, FDIC Chief Operating Officer John Bovenzi sent an e-mail to all FDIC employees entitled “Workforce Planning for the Future.” The memo outlined certain “preliminary conclusions” related to the “2005 planning and budget formulation process,” evaluating industry and technological trends, forecasting the need for greater agility and adaptability by the agency, and stated: “The FDIC of the future will be a smaller, more flexible agency.” Bovenzi explained that “all indicators point[ed] to a smaller FDIC with a somewhat different mix of skills in the future” and warned that some divisions and offices within the Agency might reduce overall staffing levels, while others might have “workload requirements or skill set[ ] imbalances that warrant filling selected vacancies.” Two weeks later, DRR Director Mitchell Glassman sent a follow-up memo to his division’s employees confirming the Agency’s view that changes in the banking industry, advances in technology, and workflow improvements had led to “declining workload and excess staff’ and thus might require “difficult decisions ... regarding the size and structure of [the] division.” This communication was followed by a string of e-mails and memos implementing cross-training plans, voluntary rotational assignments, and other staffing changes, forecasting staff reductions of 500 to 600 positions, and predicting that an involuntary Reduction-in-Force (RIF) would still be required.

In a series of memos in October 2004, FDIC management informed staff it planned to reduce the DRR workforce by 53%, from 515 to 240 positions. Buyouts would be offered to all permanent DRR employees (with the exception of a small group of “Executive Management” employees), as well as to employees throughout the Agency on a more limited basis. The offer would include a cash payment equal to 50% of the employee’s total annual salary, the ability to combine the buyout with regular or early retirement, and no restrictions on the employee’s ability to seek employment in another federal agency.

The buyout period would last from November 2004 to May 2, 2005. Director Glassman’s memo also informed DRR employees they would have the opportunity to apply for crossover opportunities with the Division of Supervision and Consumer Protection (DSC) through the Agency’s Corporate Employee Program (CEP). Lastly, Glassman explained that a RIF would be implemented during 2005 “to involuntarily separate any remaining surplus [DRR] employees.” More than 575 FDIC employees applied for and accepted the buyout. 132 were DRR employees. Another 73 DRR employees transferred to other FDIC divisions.

Moreover, as planned, in April 2005, Glassman announced the RIF would go forward and would be effective September 3, 2005. Glassman informed DRR employees that, “while the outcome of the RIF [was] not known, [his] notice [was] intended to alert [them] to the possibility [they] could be impacted through the RIF process.” As of June 30, 2005, 312 permanent DRR employees were subject to the RIF. 56.1% of them were over age 50.

Those employees who had resigned or retired before June 2005 in connection with the buyout program were not considered in the RIF process. 63 DRR employees were selected for involuntary termination and received RIF Notices terminating their employment, effective September 3, 2005. FDIC terminated 53 of those 63 employees; 7 retired in lieu of separation; and 3 voluntarily resigned after receiving a specific RIF Notice. 233 DRR employees remained after the RIF. In fall and winter 2005-06, the employees filed notices with the Equal Employment Opportunity Commission (EEOC). Am. Compl. ¶ 4, Aliotta v.

Gruenberg, No. 05-02325 (D.D.C. Feb. 8, 2006) (Am. Compl.); see 29 U.S.C. § 633a(d). On December 5, 2005, they filed their complaint in the district court alleging FDIC violated 29 U.S.C. § 633a, the portion of the ADEA applicable to federal employers. See 29 U.S.C. § 633a(e).

On July 25, 2006, the district court granted the employees’ motion for class certification, defining the class as “[fjormer or present employees of FDIC’s Division of Resolution and Receiverships who were born on a date on or before September 30, 1955 and who, as a result of the 2005 RIF, either accepted a buyout or reduction in grade, or were terminated from their positions in the DRR.” Aliotta v. Gruenberg, 237 F.R.D. 4, 13 (D.D.C.2006). The parties filed cross-motions for

summary judgment in the district court and submitted expert reports providing statistical analyses to support their positions. Analyzing only the 53 involuntary separations, 7 retirements in lieu of involuntary separation, and 3 resignations in lieu of involuntary separation, FDIC’s expert, industrial and organizational psychologist Dr. P.R. Jeanneret, found the average age of the 63 DRR employees affected by the 2005 RIF was 48.28 years. Def.’s Mot. Summ. J., Ex. 27 at 6 (Jeanneret Report). Only 42.9% of the RIF’d employees were above the age of 50. Def.’s Mot. Summ.

J., Ex. 27-1 at 20 (filed Feb. 26, 2008) (Jeanneret Rebuttal). On December 31, 2004 (before the RIF), 59.1% of permanent DRR employees were above the age of 50; on September 17, 2005 (after the RIF), the percentage of above-50 employees had increased slightly to 59.6%. Jeanneret Report at 17. In contradistinction to Dr. Jeanneret’s statistical analysis, class members’ expert, Dr. Lance Seberhagen, included in his calculation of the “RIF-related” impact all employees affected by both the 2004-05 buyouts and the 2005 RIF. Def.’s Mot. Summ. J., Ex. 28 (Seberhagen Report). Dr.

Seberhagen identified a set of “RIF-related separation codes” he believed captured the group of employees harmed. Id. at 3. The group included the codes assigned to voluntary retirements, early retirements, retirements and resignations in lieu of involuntary separation, resignations, terminations of term appointments, and involuntary terminations. Id. at 3, 17 tbl.20. Using those codes, he found that permanent DRR employees above the age of 50 were separated at 139.8% the rate of under-50 DRR employees. Id. at 5. Rejecting Dr. Seberhageris analysis, the district court granted FDIC’s motion for

summary judgment and denied class members’ motion for partial

summary judgment. Aliotta, 576 F.Supp.2d at 115. The court concluded that because employees who accepted FDIC’s buyout offers did so voluntarily, the Agency’s buyout program was not an “adverse employment action” and thus could not be considered as part of the employees’ prima facie discrimination case. Id. at 120-24. Analyzing only the 2005 RIF, the court held class members had failed to adequately rebut FDIC’s proffered nondiscriminatory justifications for the RIF. Id. at 124-28. The court concluded both the disparate treatment and disparate impact claims failed.

Class members filed a motion to alter or amend the judgment, which the district court denied. Aliotta v. Bair, 623 F.Supp.2d 73, 75-76 (D.D.C.2009). This appeal followed. II Before proceeding to the merits, we first address FDIC’s assertion class members waived their right to challenge the district court’s failure to analyze them claims under the appropriate “pattern or practice” framework. FDIC insists class members never claimed before the district court FDIC engaged in a pattern or practice of discrimination. FDIC Br. at 22.

We con-elude class members alleged a pattern or practice claim in their complaint but may nonetheless have failed to preserve it at the

summary judgment stage. However, even assuming they did preserve their pattern or practice claim,

summary judgment was properly granted because the vagaries of the various analytical frameworks were no longer relevant. Plaintiffs alleging age discrimination in violation of the ADEA may seek recovery under both disparate treatment and disparate impact theories of recovery. See Smith v. City of Jackson, 544 U.S. 228, 236-40, 125 S.Ct. 1536, 161 L.Ed.2d 410 (2005). In a disparate treatment claim, plaintiffs seek to prove an employer intentionally treated some people less favorably than others because of their age. See, e.g., Reeves v.

Sanderson Plumbing Prods., Inc., 530 U.S. 133, 141, 120 S.Ct. 2097, 147 L.Ed.2d 105 (2000) (stating plaintiffs age “must have ‘actually played a role in [the employer’s decisionmaking] process and had a determinative influence on the outcome’ ”). By contrast, in a disparate impact claim, plaintiffs challenge employment practices that are “facially neutral in then* treatment of different groups but that in fact fall more harshly on one group than another and cannot be justified by business necessity.” Hazen Paper Co. v.

Biggins, 507 U.S. 604, 609, 113 S.Ct. 1701, 123 L.Ed.2d 338 (1993). “ ‘Proof of discriminatory motive ... is not required under [the] disparate-impact theory.’ ” Id. A. Class Members’ Disparate Treatment Claim Disparate treatment claims brought under the ADEA may involve “an isolated incident of discrimination against a single individual, or ... allegations of a ‘pattern or practice’ of discrimination affecting an entire class of individuals.” Palmer v. Shultz, 815 F.2d 84, 90 (D.C.Cir.1987). In International Brotherhood of Teamsters v.

United States, 431 U.S. 324, 360-62, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977), the Supreme Court created a framework for litigating pattern or practice claims. Pattern or practice cases proceed in two phases. In the initial, or “liability,” phase of a pattern or practice lawsuit, the analysis focuses on whether unlawful discrimination has been the employer’s regular or “systemwide” pattern or practice. Id. at 336, 97 S.Ct. 1843. In order to make out a prima facie case, the plaintiffs must prove “more than the mere occurrence of isolated or ‘accidental’ or sporadic discriminatory acts.” Id.

They must establish, by a preponderance of the evidence, that discrimination “was the company’s standard operating procedure! — ]the regular rather than the unusual practice.” Id.

In this phase, the plaintiffs need not show each individual member of the class “was a victim of the employer’s discriminatory policy,” id. at 360, 97 S.Ct. 1843, since “proof of the pattern or practice supports an inference that any particular employment decision, during the period in which the discriminatory policy was in force, was made in pursuit of that policy,” id. at 362, 97 S.Ct. 1843 (explaining it is presumed that as a member of the class, each plaintiff has been the victim of the discriminatory conduct).

Statistical evidence may suffice to establish a prima facie case if the disparities in treatment are significant. See, e.g., Wagner v. Taylor, 836 F.2d 578, 592 (D.C.Cir.1987); Ledoux v. District of Columbia, 820 F.2d 1293, 1303 (D.C.Cir. 1987). In their amended complaint, the Aliotta plaintiffs alleged a persistent pattern or practice of discrimination spanning almost a decade. See Am. Compl. ¶¶ 56-94.

The recitation included allegations that remarks made by FDIC management were hostile to older employees as well as allegations that buyout offers and RIFs in 2002, 2003, and 2004 were specifically designed to reduce the number of older employees and that the complete sequence of events showed discrimination against employees over the age of 50 was the “regular rather than the unusual” practice at FDIC. Teamsters, 431 U.S. at 336, 97 S.Ct. 1843.

It is nonetheless unclear (at least as to their allegations of disparate treatment) the pattern or practice claim survives on appeal because plaintiffs cannot raise on appeal claims they allege in their complaint but abandon at the

summary judgment stage, see Road Sprinkler Fitters Local Union No. 669 v. Indep. Sprinkler Corp., 10 F.3d 1563 (11th Cir.1994) (declining to address a claim alleged in the complaint but not raised at

summary judgment); Self Directed Placement Corp. v. Control Data Corp., 908 F.2d 462, 466 (9th Cir.1990) (same); see also Edmond v. U.S. Postal Serv., 949 F.2d 415, 422 (D.C.Cir.1991) (stating that while “[t]here is no bright-line rule to determine whether a matter has been properly raised in moving papers, ... when a plaintiffs opposition is less than paradigmatic, ... the question becomes one of sufficiency, i.e., whether in light of the policies behind the rule of waiver plaintiff sufficiently raised the issue below so that waiver should not apply”). In their motion for partial

summary judgment, class members focus only on the 2004-05 buyout and point to no policies or other employment decisions targeting or adversely affecting older employees. See Pls.’ Summ. J. Mem. at 1-4, Aliotta v. Bair, No. 05-02325 (D.D.C. Feb.25, 2008). Nor do they argue there is a material dispute concerning an intentional pattern or practice of discrimination.

More significantly, FDIC challenged the disparate treatment claim and argued it should be analyzed under the McDonnell Douglas framework applicable to individual discrimination claims, not the Teamsters framework, and class members’ opposition did not dispute the Agency’s position. See Def.’s Mot. Summ. J. at 22; Pls.’ Opp’n. at 29 (citing Teamsters only once and for a general proposition applicable to both individual and pattern or practice claims); see, e.g., Muhammad v.

Giant Food Inc., 108 Fed.Appx. 757, 764 (4th Cir.2004) (explaining a passing reference to pattern or practice allegations in plaintiffs’ responses to defendant’s

summary judgment motions and a failure even to cite Teamsters were insufficient to preserve plaintiffs’ arguments that Teamsters applied to their claims). The forfeiture debate seems largely beside the point. The class members’ singular focus on the Teamsters analysis appears to hinge on a distinction without a difference. Once a prima facie case is established, the burden shifts to the employer to rebut the inference of discrimination by showing the employees’ proof is either inaccurate or insignificant. Teamsters, 431 U.S. at 360, 97 S.Ct. 1843.

Failure to rebut the inference moves a pattern and practice case to the remedial stage where each class member must show individual harm. Id. at 361-62, 97 S.Ct. 1843. However, as we explain below, class members’ flawed statistical evidence is fatal to their claims under either framework since it fails to establish any adverse effect on older employees. See Segar v.

Smith, 738 F.2d 1249, 1274 (D.C.Cir.1984) (noting plaintiffs’ statistics must “show a disparity of treatment, eliminate the most common nondiscriminatory explanations of the disparity, and thus permit the inference that, absent other explanation, the disparity more likely than not resulted from illegal discrimination”). Under our decision in Brady v. Office of Sergeant at Arms, 520 F.3d 490, 493 (D.C.Cir.2008), at the

summary judgment stage, “once [an] employer asserts a legitimate, nondiscriminatory reason [for its challenged decision], the question whether the employee actually made out a prima facie case is ‘no longer relevant’ and thus ‘disappear[s]’ and ‘drops out of the picture.’ ” See id. at 494 (explaining that once an employer asserts a legitimate, nondiscriminatory explanation, “the district court need not — and should not— decide whether the plaintiff actually made out a prima facie case”); id. (describing the prima facie case at the

summary judgment stage as “a largely unnecessary sideshow”); see also Jones v. Bernanke, 557 F.3d 670, 678 (D.C.Cir.2009) (explaining “ ‘the question whether the employee made out a prima facie case under the McDonnell Douglas framework ‘is almost always irrelevant’ because ‘by the time the district court considers an employer’s motion for

summary judgment ... the employer ordinarily will have asserted a legitimate, nondiscriminatory reason for the challenged decision’ ’ ”). Thus, once an employer has submitted admissible evidence of a legitimate, non-discriminatory reason for its decisión, any distinction between the burden-shifting frameworks becomes immaterial to the success of a discrimination case.

Under either framework, the only relevant question is “whether [the plaintiff] produced evidence sufficient for a reasonable jury to find that the employer’s stated reason was not the actual reason and that the employer intentionally discriminated against [the employee].” Brady, 520 F.3d at 495. Nonetheless, while Brady held that in an individual discrimination case, an employer’s mere assertion of a legitimate, nondiscriminatory explanation renders the question whether the plaintiff made out a prima facie case “almost always irrelevant,” id. at 493, our decision in Segar v.

Smith requires more from an employer in a pattern or practice case. See Segar, 738 F.2d at 1269-70 (explaining that because “the plaintiffs’ initial offer of evidence [in a pattern or practice case] will have been so strong ... the bare articulation of a nondiscriminatory explanation will not suffice to rebut it”). Under Segar, in a pattern or practice case, “the strength of the evidence sufficient to meet [an employer’s] rebuttal burden will typically need to be much higher than the strength of the evidence sufficient to rebut an individual plaintiffs low-threshold McDonnell Douglas showing.” Id.

The Segar court, however, acknowledged that if an employer accused of a pattern or practice of discrimination satisfies its heightened rebuttal burden, the plaintiffs’ prima facie case, as under Brady, becomes irrelevant. See id. at 1273 n. 20 (explaining that “ ‘[w]here the defendant has done everything that would be required of him if the plaintiff had properly made out a prima facie case, whether the plaintiff really did so is no longer relevant’ ” since the district court “ ‘has before it all the evidence it needs [to make the ultimate determination]’ ” (quoting U.S. Postal Serv. Bd. of Governors v.

Aikens, 460 U.S. 711, 715, 103 S.Ct. 1478, 75 L.Ed.2d 403 (1983))); id. at 1270 n. 15 (noting that “class actions often can be viewed as collapsing the prima facie and pretext stages of a suit involving an individual plaintiff’); id. at 1267 (“How far [the] 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.”).

Because FDIC has done more than simply assert a nondiscriminatory explanation for the challenged actions — it also submitted evidence demonstrating that class members’ statistics, after excluding the voluntary buyouts, failed to show even an insignificant disparity between older and younger employees — Segar does not preclude us from applying the rule set forth in Brady. In Segar, the court concluded the rebuttal of the employer, the federal Drug Enforcement Agency (DEA), failed as a matter of law because DEA submitted no admissible evidence to support its purported nondiscriminatory explanation. Id. at 1287-88.

Here, FDIC sought to rebut class members’ prima facie case in two ways. First, the Agency offered a legitimate, nondiscriminatory explanation for the RIF: it implemented the RIF to respond to decreased workload in DRR due to the improved health of the banking industry and to improve the Agency’s responsiveness and efficiency. See Def.’s Mot. Summ. J. at 28. Unlike the employer in Segar, who presented no admissible evidence supporting its nondiscriminatory justification, FDIC submitted numerous communications between Agency officials and employees explaining its nondiscriminatory reasons for the RIF.

See, e.g., Email from DRR Director Mitchell Glass-man to DRR Employees (Aug. 19, 2004) (stating “[r]ecord profitability and capital in the banking industry,” “[i]ndustry consolidation,” “[e]merging technology,” and “improved business processes” had led to “a declining workload and excess staff’ and would require some “difficult decisions” regarding the “size and structure” of DRR); E-mail from FDIC Chief Operating Officer John Bovenzi to FDIC Employees (Oct. 26, 2004) (explaining a RIF in certain divisions would likely be necessary since “staffing levels [were] not justified by current or projected workloads”).

Class members did not, at the

summary judgment stage, and have not, on appeal, pointed to any evidence refuting FDIC’s claim the RIF targeted DRR because of the division’s reduced workload caused by improved conditions in the banking industry. See Aliotta, 576 F.Supp.2d at 125. FDIC’s rebuttal also included an attack on class members’ statistical methodology. FDIC argued the buyout employees should not be included in class members’ disparate impact analysis and submitted reports from its own statistical expert refuting their methodology, see Jeanneret Report at 6, 24; Jeanneret Rebuttal at 9-11.

Unlike DEA’s attack on the plaintiffs’ statistical proof in Segar, 738 F.2d at 1272, FDIC’s alternative statistical analysis demonstrated class members’ statistics could not support an inference of discrimination. See Aliotta, 576 F.Supp.2d at 123 & n. 4, 125-26, 127-28 (holding the voluntary buyouts could not comprise any part of the employees’ case and that, without the buyouts, the employees could show no adverse impact on older employees). FDIC satisfied its rebuttal burden, and class members’ prima facie case is therefore irrelevant.

In order for class members to succeed on their disparate treatment claims, they must have produced evidence sufficient to demonstrate FDIC’s nondiscriminatory reason for the RIF was pretext and that FDIC intentionally discriminated against older workers. See Brady, 520 F.3d at 494. Neither class members’ statistical nor their non-statistical evidence is sufficient. See infra Sections IV, V. B. Class Members’ Disparate Impact Claim Class members’ disparate impact claim is easier to parse.

In Segar, we held a class of plaintiffs alleging a pattern or practice of discrimination may also challenge the disparate impact of specific employment practices. Segar, 738 F.2d at 1266-67. To establish a prima facie disparate impact claim under the ADEA, a plaintiff is not required to offer evidence the employer’s action was the result of discriminatory intent, see Krodel v. Young, 748 F.2d 701, 709 (D.C.Cir.1984), but need only offer statistical evidence of a kind and degree sufficient to show the employment decision disproportionately impacts older employees, id.; see also Roger, 98 F.3d at 639.

As we explained in Segar, by challenging the effect of specific employment practices, plaintiffs alleging disparate impact, like those in a disparate treatment pattern or practice case, are alleging the employer’s practices have had a “systemic adverse effect” on members of the class. See Moore v.

Summers, 113 F.Supp.2d 5, 19 (D.D.C.2000) (noting “ ‘an important point of convergence’ between disparate treatment and disparate impact claims exists in class actions ... [b]ecause both ... claims ‘are attacks on the systemic results of employment practices [and] proof of each claim will involve a showing of disparity between the minority and majority groups in an employer’s workforce’ ” (quoting Segar, 738 F.2d at 1267)). In their amended complaint, class members allege the 2005 RIF “had a discriminatory impact against plaintiffs and other employees over the age of 50.” Am. Compl. ¶ 86. At the

summary judgment stage, they again argued FDIC’s actions disparately affected older employees and offered statistical evidence to support their claim. See Pis.’ Opp’n at 9-13. The district court, however, concluded class members’ statistics were invalid and established no disparate impact. Aliotta, 576 F.Supp.2d at 126-28. In the alternative, the district court held FDIC had articulated a reasonable factor other than age, to wit, the “reduced workload” in DRR, that the class members failed to rebut. Id. at 127.

Class members unsuccessfully challenged the court’s holding in their motion to alter or amend the judgment and now continue to defend their claim of disparate impact on appeal. Ill The foregoing analysis reveals class members may not have preserved a distinct pattern and practice claim, but they assert both disparate treatment and disparate impact. Both class member claims are premised almost' entirely upon the statistical findings of their expert, Dr. Seberhagen.

In order for class members to show a disparate effect on older workers, they must combine the effects of the involuntary terminations resulting from the 2005 RIF with the effects of the voluntary retirements from the 2004-05 buyout offers. But, as the district court properly concluded, id. at 120-24, class members cannot include as evidence of discrimination the statistics of a group of employees who, because they voluntarily accepted a buyout, suffered no adverse employment action. Without the inclusion of the voluntary terminations, class members’ claims of discrimination collapse.

The statistical impact of the involuntary RIF terminations reveals a disparate effect on younger, not older, employees, see Jeanneret Rebuttal at 14-16, 19-20 tbls.2, 3 & 4. Under either a disparate treatment or disparate impact theory of discrimination, plaintiffs must show they suffered an adverse employment action. See, e.g., Barnette v. Chertoff, 453 F.3d 513, 515 (D.C.Cir.2006); see also Baloch v. Kempthome, 550 F.3d 1191, 1196 (D.C.Cir.2008) (same).

This court has defined an “adverse employment action” as “a significant change in employment status, such as hiring, firing, failing to promote, reassignment with significantly different responsibilities, or a decision causing significant change in benefits.” Douglas v. Donovan, 559 F.3d 549, 552 (D.C.Cir.2009). “Thus, not everything that makes an employee unhappy is an actionable adverse action.” Id. “ ‘[Resignations or retirements are presumed to be voluntary....”’ Veitch v. England, 471 F.3d 124, 134 (D.C.Cir.2006) (Rogers, J., concurring); see also Keyes v.

District of Columbia, 372 F.3d 434, 439 (D.C.Cir.2004); Henn v. Nat’l Geographic Soc’y, 819 F.2d 824, 828 (7th Cir.1987). This includes “buyout” plans. See, e.g., Terban v. Dep’t of Energy, 216 F.3d 1021, 1023-24 (Fed.Cir.2000). In certain cases, the doctrine of constructive discharge enables an employee to overcome the presumption of voluntariness and demonstrate she suffered an adverse employment action by showing the resignation or retirement was, in fact, not voluntary. See, e.g., Rowell v. BellSouth Corp., 433 F.3d 794, 805 (11th Cir.2005); Vega v. Kodak Caribbean, Ltd., 3 F.3d 476, 480 (1st Cir.1993).

The test for constructive discharge is an objective one: whether a reasonable person in the employee’s position would have felt compelled to resign under the circumstances. See Bodnar v.

Synpol, Inc., 843 F.2d 190, 194 (5th Cir.1988) (stating constructive discharge claim “relies on an objective test to evaluate what otherwise appears to be voluntary conduct by an employee”); Rowell, 433 F.3d at 803 (describing test). “The ‘voluntariness’ question ... turns on such things as: did the person receive information about what would happen in response to the choice? [W]as the choice free from fraud or other misconduct? [D]id the person have an opportunity to say no?” Henn, 819 F.2d at 828 (holding plaintiffs who accepted early retirement buyout could prevail on ADEA claim only by showing the employer “manipulated the options so that they were driven to early retirement not by its attractions but by the terror of the alternative”); see Bodnar, 843 F.2d at 192-94 (analyzing allegedly coercive factors in employer’s early retirement offer and concluding employees had failed to proffer “objective evidence that working conditions had become so intolerable as to force [employees’] resignation”).

Mere uncertainty due to the threat of a RIF layoff does not translate into a constructive discharge. See Adams v. Lucent Techs., Inc., 284 Fed.Appx. 296, 301-02 (6th Cir.2008) (unpublished table decision) (holding plaintiffs’ uncertainty regarding the effect of a potential merger on their jobs did not establish early retirement offer constituted constructive discharge); Vega, 3 F.3d at 480-81 (noting nothing in the record indicated refusing early retirement meant employees would be discharged or subjected to intolerable working conditions).

Class members argue the district court could not consider the voluntariness of the buyouts — an individual question — until the remedial phase of their pattern and practice claim and that even if the question of voluntariness could be addressed during the liability phase, the court resolved it incorrectly.

The former argument is specious; the district court considered voluntariness not in determining the remedial issue whether any individual employee was entitled to compensation, but rather in determining whether the statistical analysis proffered by the class members showing a disparate number of older employees accepted the buyout could “comprise any part of [their] prima facie case of discrimination.” Aliotta, 576 F.Supp.2d at 123-24. A class-wide voluntariness inquiry is appropriate for that purpose. That leaves the question of whether FDIC’s buyout offers were voluntary.

Class members argue they were not and thus constitute an “adverse employment action” on which they premise liability under the ADEA. Employees’ Br. at 15-17, 53-59. Accordingly, they contend any analysis of the sufficiency of their proof should include those employees who accepted the buyout. Id. at 35-38. After reviewing the Agency’s reorganization charts and seniority lists, class members say many older DRR employees were convinced they faced a “near certainty of being terminated in a RIF” if they did not accept the buyout. Id. at 17.

Because the employees “reasonably believed they were going to lose their jobs” if they did not accept early retirement, they were essentially “compelled” to accept the buyout. Id. at 15, 57-58. The employees’ decisions to accept the buyout, class members argue, were motivated not by the attractiveness of the offer, but rather by the “terror of the alternative.” Id. at 17. We are not persuaded.

Undoubtedly, the employees who accepted buyout offers faced a difficult decision: they could leave the Agency early and receive an incentive payment and benefits, or they could choose to stay and face the risk of termination in the RIF. But, senior employees were not faced with “an impermissible take-it-or-leave-it choice between retirement or discharge,” see Rowell, 433 F.3d at 805, nor were they otherwise compelled to accept the buyout.

First, with the possible exception of a few individual employees who claim the size of the reduction and the veterans and seniority preferences of their co-workers guaranteed they would not survive a RIF, see Employees’ Br. at 15-16, employees considering whether to accept the buyout could do no more than speculate that they might be terminated.

Although a RIF seemed inevitable, see E-mail from FDIC COO John Bovenzi to FDIC Employees (Oct. 26, 2004) (indicating the “necessary staffing reductions [in DRR] ... c[ould not] be accomplished entirely through voluntary departures”) (Bovenzi E-mail), it was impossible for DRR employees to know how many employees would be subject to the RIF. That number was dependent on retirements, transfers to other divisions within the Agency, and general attrition. Moreover, it was impossible for employees deciding whether to leave voluntarily to know exactly who would be RIF’d.

As noted swpra, FDIC’s RIF process gives preference to certain types of employees — for example, veterans and those with seniority. Without knowing whether certain employees with those preferences would be subject to the RIF, it was impossible for employees to calculate their chances of surviving the RIF — a chance that improved if a greater number of preference employees accepted the buyout and dropped out of the competition for positions. Moreover, the “bump” and “retreat” rights of FDIC employees subject to a RIF are complex, see 5 C.F.R. § 351.701; Aliotta v. Bair, Decl. of Pamela K.

Mergen, Lead Human Resources Specialist at FDIC, No. 05-cv-02325 (D.D.C. Feb. 21, 2008); see also Benjamin Franklin Am. Legion Post No. 66 v. U.S. Postal Serv., 732 F.2d 945, 946 n. 1 (D.C.Cir.1983) (“The process by which RIF procedures work is quite complex.”), making it almost impossible for any individual employee to know beforehand whether she will be terminated. Class members’ purported “Hobson’s choice” between retirement and termination, Employees’ Br. at 54, might never materialize. Furthermore, employees were not pressured into accepting the offer. Cf. James v.

Sears, Roebuck & Co., 21 F.3d 989, 992-94 (10th Cir.1994) (evidence employees were pressured into accepting buyout and early retirement plan by employers’ threats to fire them was sufficient to establish constructive discharge). They were given detailed information about the terms of the buyout and were allowed several months to decide whether to accept it, see Bovenzi Email. See, e.g., Bodnar, 843 F.2d at 193-94 (holding fifteen days to consider early retirement offer was “ample time” for an employee to consult attorney and examine options).

Employees were not threatened with lower pay or benefits if they did not accept the buyout, and they had the option of applying for transfer to other FDIC divisions. In fact, 73 DRR employees applied for and accepted transfers to other positions within the Agency before the RIF began. The district court’s rejection of class members’ statistics also implies, as suggested by FDIC in the record, that the statistics were based on a flawed methodology and therefore not probative of whether the Agency intentionally discriminated against older employees.

As demonstrated by our discussion above, there are multiple reasons an older employee presented with a buyout offer of early retirement in advance of an impending RIF might choose to accept the offer. The employee might feel she has no choice because being involuntarily terminated in the RIF is inevitable. Alternatively, the employee may simply believe the early retirement offer is such a good deal she voluntarily chooses to take advantage of the buyout incentives. Dr. Seberhagen’s statistics, however, do not appear to consider employee choice.

If his statistics do not control for this important explanatory variable, they tell us nothing about why older employees took the buyouts, and are therefore not relevant to determining whether FDIC discriminated against them. See, e.g., Garcia v.

Johanns, 444 F.3d 625, 635 (D.C.Cir.2006) (noting appellants’ statistical analysis was “analytically flawed” because it “did not incorporate key relevant variables connecting disparate impact to [the employer’s] decisionmaking criteria”); Segar, 738 F.2d at 1261 (“The choice of proper explanatory variables determines the validity of the regression analysis.”); see also Jeanneret Rebuttal at 10-11 (arguing there are “no valid conclusions to be drawn from Dr.

Seberhagen’s work” because his statistics “[do not] attempt[] to segregate key variables for analysis” and that because he “lump[ed] all the[ ] outcomes together and assessfed] only the bottom line result,” his analysis “yields no reliable inferences about the process he was purporting to study”). Class members also argue our decision in Schmid v. Frosch, 680 F.2d 248 (D.C.Cir.1982) (per curiam), justifies inclusion of those employees who accepted the buyout in their statistical analysis. Employees’ Br. at 34-35.

They argue that under Schmid, the group of employees adversely affected by a RIF includes all employees “hurt” by the RIF. Id. at 35. The “threat[]” of termination facing employees considering whether to accept the buyout, they argue, was sufficient “harm” to constitute an adverse employment action. Id. Class members thus argue the employees who accepted the buyout were just as much “victims” of FDIC’s discriminatory policy as those employees who refused the buyout and suffered involuntary termination in the RIF. Class members construe our holding in Schmid far too broadly.

In Schmid, we concluded the group of employees actually “hurt” by the RIF and thus “probative” of the plaintiffs age discrimination claim were those who had received RIF notices and were either separated or downgraded as a result. 680 F.2d at 250-51; see also id. at 251 n. 8 (noting the “usefulness [of plaintiffs statistics] depends on all the surrounding facts and circumstances”). The employees the class members seek to include here never received RIF notices; they left FDIC voluntarily before any RIF notices were issued.

This distinction is not, as class members suggest, “immaterial.” The statistical analysis in Schmid is therefore entirely distinguishable. Accepting an employer’s offer of voluntary early retirement may often be beneficial to older or more senior employees. See Henn, 819 F.2d at 826, 828; Smith v. World Ins. Co., 38 F.3d 1456, 1461 (8th Cir.1994). An employer should therefore not be deterred from taking voluntary measures to reduce its workforce, especially where the need for involuntai'y reduction measures depends on the success of the employer’s efforts to encourage voluntary responses.

Routinely including the statistical impact of voluntary terminations in the assessment of disparate impact would discourage employers from offering incentives for voluntary exits from the workforce. To be sure, where there is evidence an employer’s voluntary measures are motivated by nothing more than a desire to rid the company of older employees, such incentives may become both undesirable and unlawful. Nonetheless, as we discuss below, class members have put forward no evidence demonstrating FDIC’s buyout plan was motivated by discriminatory intent.

IV Having concluded FDIC’s voluntary buyouts were not adverse employment actions and thus should not be considered as part of class members’ case, we analyze only the independent effect of the 2005 RIF itself and find that, once the buyouts are excluded, their case effectively collapses. The remaining statistical evidence supports neither of their claims. The average age of those employees separated by the RIF was 48.28 years. Jeanneret Report at 6. 62% of the RIF’d employees were under age 50. Jeanneret Rebuttal at 13.

Moreover, the RIF’d population was statistically significantly younger than the population from which it was drawn. While 56.1% of permanent DRR employees subject to the RIF were over 50, only 42.9% of those actually RIF’d were 50 or older. Id. at 14. Between December 2004 (before the RIF) and December 2005 (after the RIF), the average age of DRR employees remained constant at 52.10 years of age. Id. at 12; Jeanneret Report at 16 ex.4, and the average age of the overall FDIC workforce increased slightly from 46.63 years to 46.93. Defs.’ Mot. Summ. J., Ex. 23 at 27 tbl.F.

Thus, the relevant statistics do not support the employees’ theory that the RIF disproportionately affected older employees. If anything, the evidence establishes exactly the opposite — that the RIF disproportionately affected younger employees. V In addition to their flawed statistical analysis, class members argue certain statements made by FDIC officials raise a reasonable inference of discriminatory bias against older employees.

They assert then-Chairman Donald Powell’s comment made in 2001 or 2002 to a group of employees that he “want[ed] young people around [him] ... [because] they have all the innovative ideas” and a statement made by the then-Deputy Chairman of the FDIC, Donald Greer, in a 1995 magazine

article that he would like to “keep some of the youngest and brightest people who are moving up in the ranks” support the inference that FDIC targeted DRR for reduction in 2004-05 because it had the highest proportion of older employees among the Agency’s divisions. Employees’ Br. at 47-50. The district court dismissed Chairman Powell’s statements as unsupportive of class members’ claims because they did not present any evidence Powell actually made the alleged statement. Aliotta, 576 F.Supp.2d at 124-25. We agree with the district court.

Class members’ response on appeal that “Powell did not deny saying it,” Employees’ Br. at 48, does not persuade us otherwise. Lastly, class members contend FDIC’s Corporate Employee Program demonstrates FDIC’s 2004-05 downsizing efforts were not intended to respond to a reduced workload but rather to purge the Agency of older DRR employees and replace them with younger ones. Employees’ Br. at 43-46.

The district court rejected class members’ theory, concluding that because the positions created under the CEP were for workers assuming different responsibilities in different departments than the employees, the two groups were “not so similarly situated as to support the proposition that the FDIC conducted the voluntary buyout, transfers and RIF as an elaborate ruse to flush the agency of senior staff.” Aliotta, 576 F.Supp.2d at 128. Again, the district court got it exactly right.

CEP recruits — the vast majority of whom were under age 50, see Seberhagen Report at 12 tbl.14 (noting 201 out of 214 new hires in 2005 were under 50) — did receive some training in DRR functions. But they were hired specifically to pursue DSC examiner commissions. Moreover, DRR, in particular, hired only a handful of new employees during 2005, Seberhagen Report at 13 tbl.16; Jeanneret Rebuttal at 15, even though it reduced its workforce of over 500 by more than half, Jeanneret Report at 16.

Thus, although at first glance FDIC’s recruitment of new, young workers while simultaneously separating others because of its reduced workload might raise suspicions of discrimination, a closer analysis reveals no evidence the Agency’s actions were inspired by improper motives. VI For the foregoing reasons, the judgment of the district court is Affirmed. .

Section 633a requires that "[a]ll personnel actions affecting employees or applicants for employment who are at least 40 years of age ... in executive agencies ... shall be made free from any discrimination based on age.” 29 U.S.C. § 633a(a). . A "reduction-in-force” is an administrative procedure that allows agencies to eliminate jobs and reassign or terminate employees who occupied the abolished positions. . Reductions-in-Force are governed by 5 C.F.R. pt. 351 and FDIC's RIF Circular 2100.4. See FDIC Br. at 11.

The process requires two rounds of competition and provides employees who might otherwise be terminated with certain "bump” and "retreat” rights. See 5 C.F.R. § 351.701. The process favors veterans, as well as employees with seniority and job experience within the agency. See id. §§ 351.501-504. FDIC is also required to notify employees likely to be affected once a decision is made to conduct a RIF and must send specific RIF notices to employees selected for a RIF action. See id. § 351.801(a)(1); Def.’s Mot. Summ. J., Ex. 21, Aliotta v. Bair, No. 05-02325 (D.D.C. Feb.25, 2008) (Def.'s Mot. Summ. J.).

The employees do not allege FDIC did not conduct its 2005 RIF in accordance with federal regulations or its own Agency guidelines. . Although neither this court nor the Supreme Court has addressed the question whether the ADEA authorizes disparate impact claims against federal employers, we need not resolve the issue in this case since we conclude class members have failed to demonstrate any adverse effect on older employees. See City of Jackson, 544 U.S. at 239-40, 125 S.Ct. 1536 (holding only that the ADEA authorizes disparate impact claims against employers under 29 U.S.C. § 623, a

section that does not apply to federal employers); Koger v. Reno, 98 F.3d 631, 639 & n. 2 (D.C.Cir.1996) (declining to decide whether disparate impact analysis applies to age discrimination claims because plaintiffs failed to establish a prima facie case); see also Aliotta, 576 F.Supp.2d at 126 n. 7 (noting "[m]embers of the D.C. District Court remain divided on the issue” of whether a plaintiff may allege disparate impact under the ADEA against federal employers).

For the same reason, we need not resolve whether the "business necessity” test for rebutting a disparate impact claim under Title VII or the less strict "reasonable factor other than age” test for rebutting a disparate impact claim against a private-sector employer under the ADEA, see City of Jackson, 544 U.S. at 243, 125 S.Ct. 1536 (explaining distinction between the tests), would apply if indeed such a claim may lie against a federal employer under § 633a. .

In Teamsters, the plaintiffs brought their "pattern or practice” discrimination claims under Title VII of the Civil Rights Act of 1964. 431 U.S. at 328, 97 S.Ct. 1843. Nevertheless, this court has applied the Teamsters framework to ADEA cases. See, e.g., Schuler v. PricewaterhouseCoopers, LLP, 514 F.3d 1365, 1370 (D.C.Cir.2008). . The district court did not discuss the then-Deputy Chairman’s alleged statement, but even if class members provided sufficient evidence he actually made the statement, it is insufficient, on its own, to establish proof of discriminatory intent. See, e.g., Bevan v.

Honeywell, 118 F.3d 603, 610 (8th Cir.1997) (noting "stray remarks of nondecisionmakers ... are not sufficient ..., standing alone, [to] raise an inference of discrimination”).