Constitutionality of Disparate-Impact Liability Under Title VII

• Court: Department of Justice Office of Legal Counsel
• Decided: June 9, 2026
• Status: Published

Opinion

(Slip Opinion)

Constitutionality of Disparate-Impact Liability Under Title VII EEOC’s Title VII guidelines are unconstitutional because they contemplate liability based on disparate effects alone, without regard to an employer’s likely intent, and pressure employers to engage in race-based decisionmaking. Properly understood, disparate-impact liability proscribes only those practices that reflect a significant like- lihood of intentional discrimination. The business-necessity defense requires employers to demonstrate only that the chal- lenged practice rationally serves a valid business purpose. Workplace requirements and selection procedures—such as background checks, aptitude tests, and SAT scores—are presumptively job-related. Only irrational or arbitrary practices with no plausible job-relatedness can create disparate-impact liability. Disparate-impact plaintiffs must both establish that the challenged employment practice specifically caused the alleged disparate impact and provide evidence that an equally effective alternative practice causes less disparate impact.

June 9, 2026

MEMORANDUM OPINION FOR THE CHAIR U.S. EQUAL EMPLOYMENT OPPORTUNITY COMMISSION

The U.S. Equal Employment Opportunity Commission (“EEOC”) en- forces federal antidiscrimination laws against covered private and public employers, including Title VII of the Civil Rights Act of 1964, Pub. L. No. 88-352, 78 Stat. 241, 255–57 (codified as amended at 42 U.S.C. § 2000e-2). The Supreme Court has interpreted Title VII to incorporate “disparate-impact liability”—that is, liability based on employment de- cisions that disproportionately affect members of a protected group. See generally Griggs v. Duke Power Co., 401 U.S. 424 (1971). Congress acquiesced in that interpretation when it amended Title VII in 1991. See 42 U.S.C. § 2000e-2(k). The EEOC has likewise incorporated dis- parate-impact liability into its Title VII rules, guidance documents, and technical assistance documents. See Letter for T. Elliot Gaiser, Assistant Attorney General, Office of Legal Counsel, from Andrea R. Lucas, Chair, U.S. Equal Employment Opportunity Commission at app. A (Feb. 2, 2026) (“EEOC Request”). You have asked “whether the disparate impact provisions in Title VII of the Civil Rights Act of 1964, as amended by the Civil Rights Act of 1991, are constitutional as currently interpreted and applied, particularly

1 50 Op. O.L.C. __ (June 9, 2026)

by the [EEOC] in its interpretative rules and guidance documents.” Id. at 1. They are not. EEOC’s existing interpretations, including the Uniform Guidelines on Employee Selection Procedures (“Guidelines”), embrace an unconstitu- tional reading of Title VII. Rather than treating disparate impact as an evidentiary mechanism to smoke out intentional discrimination—impos- ing liability only when disproportionate adverse effects give rise to a strong inference of intentional discrimination—EEOC’s historic inter- pretations contemplate liability based on disproportionately adverse effects alone, without regard to an employer’s likely intent. Because EEOC’s historic approach divorces liability from circumstances giving rise to a strong inference that intentional discrimination occurred, it functions as a qualified racial-proportionality mandate and spurs em- ployers to engage in race-based decisionmaking to avoid liability. That approach is unlawful and unconstitutional. Three corrections to that approach are necessary “to resolve the ten- sion between [disparate-impact] claims under [Title VII] and our color- blind Constitution.” Allen v. Milligan, No. 25A1314, 2026 WL 1552756, at *1 (U.S. June 2, 2026) (per curiam). First, the business-necessity de- fense is not a high bar; it requires employers to demonstrate only that the challenged practice is rational, convenient, or helpful for serving a valid business purpose. Employment practices are presumptively job- related, and only irrational or arbitrary practices with no plausible job-relatedness can create disparate-impact liability. Second, plaintiffs must satisfy a robust causality requirement by demonstrating—both at the pleading stage and beyond—that the challenged employment prac- tice itself (not external factors or other employer practices) caused the alleged disparate impact. Third, plaintiffs must establish with particular evidence that there is an available alternative practice that causes less disparate impact and would be equally effective for serving the em- ployer’s valid business purpose. Based on these principles, EEOC’s cur- rent validation-study and affirmative-action regulations are unlawful. This opinion proceeds in four parts. Part I provides a historical over- view of disparate-impact liability. Part II shows how disparate-impact liability pushes constitutional boundaries. Part III deploys constitutional avoidance to interpret Title VII consistent with necessary safeguards against unconstitutional applications of disparate-impact liability. Part IV addresses two aspects of EEOC’s regulations that violate both Title VII and the Constitution. 2 Constitutionality of Disparate-Impact Liability Under Title VII

I.

A.

In the wake of the Civil War, Congress approved and the states ratified the Fourteenth Amendment to guarantee that “[n]o State” would “deny to any person within its jurisdiction the equal protection of the laws.” U.S. Const. amend. XIV, § 1.1 A “core purpose” of the Fourteenth Amendment’s Equal Protection Clause was to “do[] away with all governmentally imposed discrimination based on race.” Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll., 143 S. Ct. 2141, 2161 (2023) (“SFFA”) (quoting Palmore v. Sidoti, 466 U.S. 429, 432 (1984)). The Amendment’s framers understood that the equal-protection guarantee would extend “over every American citizen, without regard to color,” id. at 2159 (quoting Cong. Globe, 39th Cong., 1st Sess. 2462 (1866) (statement of Rep. James A. Garfield)), because “without this principle of equal justice, . . . there is no republican gov- ernment and none that is really worth maintaining,” id. at 2159 (altera- tion accepted) (quoting Cong. Globe at 2766 (statement of Sen. Jacob Howard)). The framers repeatedly emphasized that the Amendment would “abolish[] all class legislation” that “subject[s] one caste of per- sons to a code not applicable to another.” Cong. Globe at 2766 (state- ment of Sen. Jacob Howard); see also id. at 2961 (statement of Sen. Luke Poland) (expressing Congress’s “intention to uproot and destroy all such partial State legislation”). By contrast, general and facially neutral leg- islation that “was equal” and “impartial to all” posed no concern because “where all of the same class are dealt with in the same way then there is no pretense of inequality.” Id. at 1063–64 (statements of Rep. Thaddeus Stevens). Although the Constitution now guarantees equal treatment, it has never guaranteed equal outcomes. “[A] law, neutral on its face and serv- ing ends otherwise within the power of government to pursue, is [not] invalid under the Equal Protection Clause simply because it may affect

1 The Supreme Court has recognized an equal-protection component in the Constitu-

tion that binds the federal government. See Bolling v. Sharpe, 347 U.S. 497, 499–500 (1954) (rooting the principle in the Due Process Clause of the Fifth Amendment); Wein- berger v. Wiesenfeld, 420 U.S. 636, 638 n.2 (1975) (same); see also United States v. Vaello Madero, 142 S. Ct. 1539, 1544 (2022) (Thomas, J., concurring) (finding “[f]irmer ground . . . in the Fourteenth Amendment’s Citizenship Clause”).

3 50 Op. O.L.C. __ (June 9, 2026)

a greater proportion of one race than of another.” Washington v. Davis, 426 U.S. 229, 242 (1976). A law that creates an unintentional disparate impact, therefore, does not violate the Constitution’s equal-protection guarantee. Indeed, the Supreme Court has repeatedly emphasized that “[t]he central purpose of the Equal Protection Clause of the Fourteenth Amendment is the prevention of official conduct discriminating on the basis of race.” Id. at 239 (emphasis added); see also SFFA, 143 S. Ct. at 2161 (collecting cases). That is why the touchstone of an equal- protection claim is whether the challenged law can “ultimately be traced to a racially discriminatory purpose.” Davis, 426 U.S. at 240 (emphasis added). “[T]he Fourteenth Amendment guarantees equal laws, not equal results.” Pers. Adm’r of Mass. v. Feeney, 442 U.S. 256, 273 (1979).

B.

Like the Constitution, Title VII’s text guarantees equal treatment, not equal outcomes. Title VII creates a motive-based exception to at-will employment by prohibiting employers, employment agencies, unions, and training programs from discriminating in employment against “any individual” “because of” such individual’s “race, color, religion, sex, or national origin.” 42 U.S.C. § 2000e-2(a)–(d). This prohibition focuses on actions taken because of membership in a protected class, rather than actions that result in disparate effects across protected classes. See Safeco Ins. Co. of Am. v. Burr, 551 U.S. 47, 63–64 & n.14 (2007) (noting that “because of” means “based on” and that “‘based on’ indicates a but- for causal relationship”). Title VII established EEOC to help administer the statute. See 42 U.S.C. § 2000e-4. EEOC quickly took “its interpretation of Title VII a step further than other agencies” and “reasoned that in addition to dis- crimination in employment, it is also an unlawful practice to fail or re- fuse to hire, to discharge or to compensate unevenly on [facially neutral] criteria which prove to have a demonstrable racial effect without a clear and convincing business motive.” Gail L. Heriot, Title VII Disparate Impact Liability Makes Almost Everything Presumptively Illegal, 14 N.Y.U. J.L. & Liberty 1, 31 (2020) (“Heriot, Presumptively Illegal”) (alteration accepted) (quoting Samuel C. Jackson, EEOC vs. Discrimi- nation, Inc., Crisis, Jan. 1968, at 16, 16–17). EEOC’s “own official his- tory . . . records with unusual candor the commission’s fundamental disagreement with its founding charter, especially Title VII’s literal

4 Constitutionality of Disparate-Impact Liability Under Title VII

requirement that the discrimination be intentional” before imposing lia- bility. Hugh Davis Graham, The Civil Rights Era: Origins and Develop- ment of National Policy 1960–1972, at 248 (1990). In Griggs v. Duke Power Co., the Supreme Court controversially in- terpreted Title VII as imposing disparate-impact liability, prohibiting “not only overt discrimination but also practices that are fair in form, but discriminatory in operation.” 401 U.S. at 431. The Court determined that Title VII blocked an employer from “requiring a high school education or passing of a standardized general intelligence test as a condition of employment,” when neither requirement was “significantly related to successful job performance” and both requirements had a disparate im- pact on black applicants and employees. Id. at 425–26. And although the lower courts had found that the employer had not engaged in intentional disparate treatment, the Court reasoned that the “absence of discrimina- tory intent does not redeem employment procedures that operate as ‘built-in headwinds,’” because “Congress directed the thrust of the Act to the consequences of employment practices, not simply the motiva- tion.” Id. at 432 (second emphasis added). In short, the Supreme Court interpreted Title VII as embodying Congress’s supposed purpose to eradicate “artificial, arbitrary, and unnecessary barriers to employment” that “operate invidiously to discriminate on the basis of racial or other impermissible classification.” Id. at 431. Because the Supreme Court believed that the “touchstone” of disparate-impact liability under Title VII is a lack of “business necessity” for challenged employment practices, it emphasized that employers may defend against disparate-impact liability by showing that the practices are “related to job performance.” Id. After Griggs, plaintiffs could allege disparate-impact claims under Title VII based on “practices that are facially neutral in their treatment of different groups but that in fact fall more harshly on one group than another.” Int’l Bhd. of Teamsters v. United States, 431 U.S. 324, 335–36 n.15 (1977). In 1978, EEOC, the Department of Labor, the Department of Justice, and the Civil Service Commission jointly adopted the Uniform Guide- lines on Employee Selection Procedures, which are still in effect today and were designed to “provide a framework for determining the proper use of tests and other selection procedures.” Uniform Guidelines on Em- ployee Selection Procedures, 43 Fed. Reg. 38,290, 38,296 (Aug. 25, 1978) (codified as amended at 29 C.F.R. part 1607). The Guidelines

5 50 Op. O.L.C. __ (June 9, 2026)

define discrimination to include “[t]he use of any selection procedure which has an adverse [i.e., disparate] impact on the hiring, promotion, or other employment or membership opportunities of members of any race, sex, or ethnic group,” “unless the procedure has been validated in accordance with these guidelines.” 29 C.F.R. § 1607.3(A).

C.

In 1989, the Supreme Court clarified its disparate-impact framework for Title VII. Wards Cove Packing Co. v. Atonio concerned an Alaska salmon cannery that exhibited stark racial disparities between the “pre- dominantly white noncannery workers and the predominantly nonwhite cannery employees.” 490 U.S. 642, 647 (1989). The Ninth Circuit had held that the nonwhite cannery employees established a prima facie case of disparate impact against the cannery based “solely on . . . statistics showing a high percentage of nonwhite workers in the cannery jobs and a low percentage of such workers in the noncannery positions.” Id. at 650. The Supreme Court rejected that raw-statistics approach because “[t]he only practicable option for many employers” facing statistical dis- parities “would be to adopt racial quotas, insuring that no portion of their work forces deviated in racial composition from the other portions thereof.” Id. at 652. The Supreme Court instead emphasized the plaintiff’s burden of showing specific causation. To establish a prima facie disparate-impact claim, “‘the plaintiff is . . . responsible for isolating and identifying the specific employment practices that are allegedly responsible for any observed statistical disparities.’” Id. at 656 (quoting Watson v. Ft. Wor. Bank & Tr., 487 U.S. 977, 994 (1988) (plurality opinion)). In other words, “a plaintiff must demonstrate that it is the application of a specific or particular employment practice that has created the disparate impact under attack.” Id. at 657 (emphasis added). This “specific causation requirement” protects defendants from “being potentially liable for ‘the myriad of innocent causes that may lead to statistical imbalances in the composition of their work forces.’” Id. (quoting Watson, 487 U.S. at 992 (plurality opinion)). It thus remanded the case to require “a demonstration that specific elements” of the defendant’s hiring practices actually caused the disparate impact. Id. at 658. A different aspect of Wards Cove provoked a congressional response. The Court also held that “the employer carries the burden of producing

6 Constitutionality of Disparate-Impact Liability Under Title VII

evidence of a business justification for his employment practice,” but “[t]he burden of persuasion . . . remains with the disparate-impact plain- tiff.” Id. at 659. In the Civil Rights Act of 1991, Pub. L. No. 102-166, 105 Stat. 1071, Congress overrode this specific aspect of Wards Cove, see 42 U.S.C. § 2000e-2(k), but left the specific-causation requirement in place, see Tex. Dep’t of Hous. & Cmty. Affs. v. Inclusive Cmtys. Pro- ject, Inc., 576 U.S. 519, 542 (2015). 2 A disparate-impact claim that reaches summary judgment or trial now proceeds in three steps. First, “a complaining party [must] demonstrate[] that [an employer] uses a particular employment practice that causes a disparate impact on the basis of race, color, religion, sex, or national origin.” 42 U.S.C. § 2000e-2(k)(1)(A)(i). Then, the burden shifts to the employer to “demonstrate that the challenged practice is job related for the position in question and consistent with business necessity.” Id. Finally, if the employer makes that showing, the burden shifts back to the complaining party to demonstrate that the employer “refuses to adopt [an equally effective] alternative employment practice” that would re- duce the disparate impact relative to the original employment practice. Id. § 2000e-2(k)(1)(A)(ii). As EEOC now observes, Title VII does not define “when an employment practice is ‘job related for the position in question’ or ‘consistent with business necessity.’” EEOC Request at 1.

II.

Since its inception in Griggs, Title VII disparate-impact liability has raised serious equal-protection concerns. See, e.g., Ricci v. DeStefano, 557 U.S. 557, 594 (2009) (Scalia, J., concurring); Heriot, Presumptively Illegal at 31; Richard A. Primus, Equal Protection and Disparate

2 Unlike Title VII, Congress has never amended Title VI to incorporate disparate-

impact liability. As explained in the Department of Justice’s recent rulemaking, Title VI “prohibits specifically intentional discrimination,” “makes no reference to unintentional disparate effects or impact,” and does not “provide any Federal department or agency with authority to prohibit unintentional disparate impact.” Rescinding Portions of De- partment of Justice Title VI Regulations to Conform More Closely with the Statutory Text and to Implement Executive Order 14281, 90 Fed. Reg. 57,141, 57,141 (Dec. 10, 2025) (first citing 42 U.S.C. § 2000d; and then citing Alexander v. Sandoval, 532 U.S. 275, 280 (2001)). Title VI prohibits “only those racial classifications that would violate the Equal Protection Clause” if committed by a government actor. Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 287 (1978) (opinion of Powell, J.); see also SFFA, 143 S. Ct. at 2156–57 n.2.

7 50 Op. O.L.C. __ (June 9, 2026)

Impact: Round Three, 117 Harv. L. Rev. 494, 528 (2003) (“Primus, Round Three”). We find these constitutional concerns well-founded.

The fundamental problem is that disparate-impact liability tends to incent—and even coerce—employers to make race-based decisions to avoid liability or the threat of liability. In Ricci v. DeStefano, for exam- ple, the fire department of New Haven, Connecticut, adopted a facially neutral and “objective examination[]” to “determine which firefighters would be considered for promotions . . . , and the order in which they would be considered.” 557 U.S. at 562. But white firefighters outper- formed other firefighters on the exam. See id. The fire department found itself in an impossible situation: keep the test results and potentially face disparate-impact liability, or toss the results and deny promotions to de- serving candidates based solely on “the statistical racial disparity.” Id. In other words, the fire department could accept the risk of disparate- impact liability or avoid it by taking race-based action—that is, by dis- criminating based on race. Even facially race-neutral actions, when mo- tivated by the purpose of altering racial balance, constitute intentional discrimination against the members of the racial group who are balanced down. After all, “if race played a role in a decision made by a govern- ment actor,” or at the behest of a government actor, then race discrimi- nation has occurred and “strict scrutiny applie[s].” Louisiana v. Callais, 146 S. Ct. 1131, 1146 (2026); see also Coal. for TJ v. Fairfax Cnty. Sch. Bd., 146 S. Ct. 541, 545 (2024) (Alito, J., dissenting from denial of cer- tiorari).3 As Ricci shows, disparate-impact liability under Title VII “place[s] a racial thumb on the scales, often requiring employers to evaluate the ra- cial outcomes of their policies, and to make decisions based on (because of) those racial outcomes.” 557 U.S. at 594 (Scalia, J., concurring). That is a significant problem because nearly every job qualification or hiring

3 To be clear, Ricci held that the fire department was unable to show a “strong basis

in evidence” that disparate-impact liability was likely to result from the department’s use of an “objective examination[],” so its race-based remedial action violated Title VII’s disparate-treatment prohibition. 557 U.S. at 562, 585. This only goes to show that even in the absence of actual disparate-impact liability, the structural incentives created by a disparate-impact regime push employers to take race-conscious—and thus racially discriminatory—remedial actions.

8 Constitutionality of Disparate-Impact Liability Under Title VII

process has a disparate impact on members of one protected group or another. So a “statute designed to establish a color-blind and gender- blind workplace” has been twisted to operate as “a powerful engine of racism and sexism, not merely permitting intentional race- and sex-based discrimination, but often making it, through operation of the legal sys- tem, practically compelled.” Johnson v. Transp. Agency, 480 U.S. 616, 677 (1987) (Scalia, J., dissenting); see also United States v. Brennan, 650 F.3d 65, 95 & n.37 (2d Cir. 2011) (extending Ricci’s logic to sex discrimination and concluding that the defendants’ remedial actions vi- olated Title VII because they “were explicitly race- and sex-based”). By pressuring employers to take race-based actions in the name of proactively addressing potential statistical disparities, disparate-impact liability allows the government to engage in race discrimination indi- rectly. But “what cannot be done directly cannot be done indirectly. The Constitution deals with substance, not shadows,” and the prohibition against racial discrimination is “levelled at the thing, not the name.” Cummings v. Missouri, 71 U.S. (4 Wall.) 277, 325 (1867). “Few princi- ples of law are more firmly stitched into our constitutional fabric than the proposition that a [government] must not discriminate against a per- son because of his race . . . or in any way act to compel or encourage” such discrimination by others. Adickes v. S.H. Kress & Co., 398 U.S. 144, 150–52 (1970); see also Anderson v. Martin, 375 U.S. 399, 404 (1964). Just as “the Federal Government is prohibited from discriminating on the basis of race,” so too is it “prohibited from enacting laws mandating that third parties—e.g., employers, whether private, State, or munici- pal—discriminate on the basis of race,” unless those laws survive strict scrutiny. Ricci, 557 U.S. at 594 (Scalia, J., concurring) (first citing Bol- ling, 347 U.S. at 500; and then citing Buchanan v. Warley, 245 U.S. 60, 78–82 (1917)). And just as it is unconstitutional for the federal govern- ment to “force[] States to engage in the very race-based discrimination that the Constitution forbids,” Callais, 146 S. Ct. at 1142; see also Miller v. Johnson, 515 U.S. 900, 926–27 (1995), so too it is unconstitutional for the federal government to coerce employers to adopt employment policies or make employment decisions motivated by race. Disparate- impact liability does not just raise constitutional doubt in occasional cases; unless narrowly circumscribed, it structurally compels the very racial discrimination that the Constitution forbids.

9 50 Op. O.L.C. __ (June 9, 2026)

Some in the academy4 and the U.S. Commission on Civil Rights 5 have insisted that facially neutral policies burdening minorities trigger dispar- ate-impact liability, but that policies burdening majority groups (such as whites and males) do not. These assertions are wrong. Of course, “[i]f the statute had prescribed this rule expressly” with respect to race, a court “would subject it to strict scrutiny.” Ames v. Ohio Dep’t of Youth Servs., 87 F.4th 822, 828 (6th Cir. 2023) (Kethledge, J., concurring), va- cated and remanded, 145 S. Ct. 1540 (2025). After all, racial preferences and “classifications are simply too pernicious to permit any but the most exact connection between justification and classification.” Gratz v. Bol- linger, 539 U.S. 244, 270 (2003) (citation omitted). And racial classifi- cations are no less “inherently suspect” when made secretly than when made openly. SFFA, 143 S. Ct. at 2163 (citation omitted). We have no doubt that this asymmetric view of Title VII would fail the exacting test of strict scrutiny. Cf. Constitutionality of Race-Based Department of Ed- ucation Programs, 49 Op. O.L.C. __, at *5–8 (Dec. 2, 2025) (“Race- Based Education Programs”). But as the Supreme Court has made clear, disparate-impact liability under Title VII extends equally to members of “minority or majority” groups. Griggs, 401 U.S. at 431. In Ames v. Ohio Department of Youth Services, the Court rejected the holdings of several circuit courts that

4 See Heriot, Presumptively Illegal at 138–40. For example, Professor Charles Sulli-

van “firmly announced” to his students that disparate-impact liability could not benefit “whites and males.” Charles A. Sullivan, The World Turned Upside Down?: Disparate Impact Claims by White Males, 98 Nw. U. L. Rev. 1505, 1506 (2004); see also Primus, Round Three at 528 (“[E]mployment practices with disparately adverse impacts on his- torically dominant classes are, as a matter of law, not actionable under Title VII.”); John J. Donohue III, Understanding the Reasons for and Impact of Legislatively Man- dated Benefits for Selected Workers, 53 Stan. L. Rev. 897, 898 n.2 (2001) (“I conclude that disparate impact analysis will not protect white males as a matter of theory.”); Mar- tha Chamallas, Evolving Conceptions of Equality Under Title VII: Disparate Impact Theory and the Demise of the Bottom Line Principle, 31 UCLA L. Rev. 305, 366–69 (1983) (outlining the “one-sided” nature of disparate-impact analysis and predicting that “courts, in an effort to reduce the intrusion on employer discretion, will continue to limit disparate impact challenges to those brought by minorities”). 5 The U.S. Commission on Civil Rights remarked that disparate-impact liability “cannot sensibly be applied to white males” given the purposes of disparate-impact lia- bility. U.S. Comm’n on Civ. Rts., Affirmative Action in the 1980s: Dismantling the Pro- cess of Discrimination 21 n.20 (1981), https://perma.cc/B6PC-TL8J.

10 Constitutionality of Disparate-Impact Liability Under Title VII

imposed an “additional burden” on majority-group plaintiffs bringing Title VII claims. 145 S. Ct. at 1545 & n.1. Instead, it held that Title VII “establish[ed] the same protections for every ‘individual’—without re- gard to that individual’s membership in a minority or majority group.” Id. at 1546. The Court emphasized that “Congress left no room for courts to impose special requirements on majority-group plaintiffs alone.” Id. True, Ames focused only on “Title VII’s disparate-treatment provision” and did not address disparate-impact claims more generally. Id. (citing 42 U.S.C. § 2000e-2(a)(1)). But it quoted Griggs approvingly for the proposition that “‘discriminatory preference for any group, minority or majority, is precisely and only what Congress has proscribed’ in Title VII.” Id. (alteration accepted) (quoting Griggs, 401 U.S. at 431). And the reasoning and language of Ames apply with equal force to dis- parate-impact claims under Title VII.

Equal application of the disparate-impact framework, however, does not resolve the core constitutional problems that disparate-impact liabil- ity often poses, as the Supreme Court has recognized. In Ricci, the Court took pains to explain that its “statutory holding [did] not address the constitutionality of the measures taken [by the fire department] in pur- ported compliance with Title VII.” 557 U.S. at 584. The Court declined to decide whether “the strong-basis-in-evidence standard would satisfy the Equal Protection Clause” or “whether a legitimate fear of disparate impact is ever sufficient to justify discriminatory treatment under the Constitution.” Id. As Justice Scalia explained, Ricci had “merely post- pone[d] the evil day on which the Court will have to confront the ques- tion: Whether, or to what extent, are the disparate-impact provisions of Title VII of the Civil Rights Act of 1964 consistent with the Constitu- tion’s guarantee of equal protection?” Id. at 594 (Scalia, J., concurring). And he predicted that “the war between disparate impact and equal pro- tection [would] be waged sooner or later.” Id. at 595–96. The Supreme Court’s decision six years later in Inclusive Communi- ties further confirms our analysis. There, the Court controversially ex- tended disparate-impact liability to the Fair Housing Act, but acknowl- edged the “serious constitutional questions that might arise” if “liability were imposed based solely on a showing of a statistical disparity.” 576 U.S. at 540 (emphasis added). The Court therefore stressed that

11 50 Op. O.L.C. __ (June 9, 2026)

disparate-impact liability must be “limited in key respects” and outlined various safeguards. Id. “Without adequate safeguards,” the Court ex- plained, “disparate-impact liability might cause race to be used and con- sidered in a pervasive way” that would raise “serious constitutional ques- tions.” Id. at 542. According to the Court, these safeguards delineate the outer constitutional limits of disparate-impact liability, because without them, “potential defendants may” have a strong incentive to “adopt ra- cial quotas” to avoid liability. Id. at 543.6 At the same time, the Supreme Court has indicated that unequal results may be evidence of discriminatory motive, with the strength of the in- ference depending on (among other things) the proffered rationale for the challenged policy. See, e.g., DHS v. Regents of the Univ. of Cal., 140 S. Ct. 1891, 1915 (2020) (plurality opinion) (disparate impact is “[p]os- sible evidence” of intentional discrimination); Village of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 269–70 (1977) (re- jecting an inference of discriminatory purpose, despite arguably dispar- ate impacts). Using statistical disparities as one piece of evidence to prove a “pattern or practice” of intentional employment discrimination fits this mold. See 42 U.S.C. § 2000e-6(a). True, Griggs was decided in the face of factual findings that there was an “absence of discriminatory intent.” 401 U.S. at 432. But there was at least a serious risk of intentional discrimination, 7 and the Supreme Court

6 The Supreme Court further admonished lower courts to “avoid interpreting dispar-

ate-impact liability to be so expansive as to inject racial considerations into every . . . decision,” Inclusive Communities., 576 U.S. at 543, and cautioned that “[r]emedial or- ders that impose racial targets or quotas might raise . . . difficult constitutional ques- tions” too, id. at 545. Remedial orders that consider race must satisfy strict scrutiny and serve the compelling governmental interest in “remediating specific, identified instances of past [intentional] discrimination that violated the Constitution or a statute.” Callais, 146 S. Ct. at 1152 (quoting SFFA, 143 S. Ct. at 2141). Mere “compliance with federal laws” such as Title VII “cannot automatically be a compelling interest”; after all, “Adarand Constructors held a federal statute unconstitutional precisely because it re- quired public officials to make use of race, and the statute was not itself supported by a compelling governmental interest.” Biondo v. City of Chicago, 382 F.3d 680, 684 (7th Cir. 2004) (emphasis omitted). 7 The trial court’s findings were controversial, considering that Duke Power “jobs

[were] rigidly segregated by race,” the company maintained “segregated facilities” until plaintiffs filed their EEOC complaint, and the company imposed a new written test re- quirement only on “the same date that Title VII became effective.” Robert Belton, The Crusade for Equality in the Workplace: The Griggs v. Duke Power Story 109–10 (2014). Yet the trial court “reject[ed] entirely the evidence of discriminatory practices [that

12 Constitutionality of Disparate-Impact Liability Under Title VII

accordingly focused on “the removal of artificial, arbitrary, and unnec- essary barriers” that “operate invidiously to discriminate on the basis of racial or other impermissible classification.” Id. at 431. Griggs’s invo- cation of Aesop’s fable about the “offer of milk to the stork and the fox” underscores the point about likely intent. Id. The fabled offer was inten- tionally discriminatory because the offeror knew and wanted the other animal to be unable to drink the milk. Reading Griggs in context and together with the Supreme Court’s more recent precedent requires “updat[ing] the framework” for dispar- ate-impact liability “to ensure a constitutional reading and application of” Title VII. Callais, 146 S. Ct. at 1157, 1161 (updating the framework for claims under section 2 of the Voting Rights Act). As Justice Scalia suggested in Ricci, a properly tailored disparate-impact scheme might constitutionally operate as “an evidentiary tool” to “smoke out” practices that present a significant likelihood of intentional discrimination. 557 U.S. at 595 (Scalia, J., concurring); see also City of Rome v. United States, 446 U.S. 156, 177–78 (1980). This is a similar approach to the one that the Supreme Court took in Callais, which held that section 2 of the Voting Rights Act of 1965, Pub. L. No. 89-110, 79 Stat. 437, “imposes liability only when the circum- stances give rise to a strong inference that intentional discrimination occurred.” 146 S. Ct. at 1156. There, Congress had amended section 2 to prohibit certain discriminatory results, and courts for decades had con- strued broadly the statute’s vague terms. See id. at 1146–47. But “properly interpreted,” the applicable standard reached only results giv- ing rise to an objectively “strong inference” of intentional discrimina- tion, thereby avoiding both the difficulties of proving subjective intent and the risk of reading the statute to induce the type of race-based deci- sionmaking that it was meant to prevent. Id. at 1155–56. As several members of the Supreme Court have acknowledged, a plaintiff’s “spe- cial burden” to demonstrate “a racial motivation” in that context is com- parable to requiring “evidence of racially discriminatory intent.” Id. at 1177–78 & n.5 (Kagan, J., dissenting, joined by Sotomayor & Jackson, JJ.) (citation omitted).

plaintiffs’ attorneys] had put before him.” David J. Garrow, Toward a Definitive History of Griggs v. Duke Power Co., 67 Vand. L. Rev. 197, 213 (2014) (book review).

13 50 Op. O.L.C. __ (June 9, 2026)

III.

With these considerations in mind, we must interpret Title VII to avoid finding the statute unconstitutional if its text is reasonably suscep- tible to a saving construction. “If an otherwise acceptable construction of a statute would raise serious constitutional problems, and where an alternative interpretation of the statute is fairly possible, we are obligated to construe the statute to avoid such problems.” Legal Authorities Sup- porting the Activities of the National Security Agency Described by the President, 30 Op. O.L.C. 1, 34 (2006) (cleaned up) (quoting INS v. St. Cyr, 533 U.S. 289, 299–300 (2001)); see, e.g., Whether Eluding Inspec- tion Under 8 U.S.C. § 1325(a)(2) Is a Continuing Offense, 49 Op. O.L.C. __, at *10–12 (June 21, 2025). Constitutional avoidance is a tool “to avoid not only the conclusion that [a statute] is unconstitutional but also grave doubts upon that score.” United States v. Jin Fuey Moy, 241 U.S. 394, 401 (1916); accord United States v. Palomar-Santiago, 141 S. Ct. 1615, 1622 (2021). Applying Griggs, Wards Cove, Inclusive Communities, and Callais, we identify three limiting principles on disparate-impact liability that prevent a constitutional collision under current Supreme Court prece- dent. We emphasize, as the Supreme Court has, that no single limiting principle is wholly sufficient; rather, each caveat is essential to avoid the conclusion that Title VII violates the Constitution’s equal-protection guarantee. First, the business-necessity defense to liability must provide defend- ants significant “leeway to state and explain the valid interest served by their policies.” Inclusive Communities, 576 U.S. at 541. Title VII re- quires employers to “demonstrate that the challenged practice is job re- lated for the position in question and consistent with business necessity.” 42 U.S.C. § 2000e-2(k)(1)(A)(i). Workplace requirements and selection procedures—such as background checks, aptitude tests, knowledge- based tests, SAT scores, high-school graduation requirements, or blind auditions—are presumptively job-related.8 Only practices that establish

8 To take just one example, EEOC’s employer guidance warns against basing em-

ployment decisions on criminal background checks that fail to accurately “predict who will be a responsible, reliable, or safe employee” due to “problems that may be more common among people of a certain race.” U.S. Equal Emp. Opportunity Comm’n & Fed. Trade Comm’n, EEOC-NVTA-0000-38, Background Checks: What Employers

14 Constitutionality of Disparate-Impact Liability Under Title VII

truly “artificial, arbitrary, and unnecessary barriers”—i.e., those that have no plausible job-relatedness—can create liability. Inclusive Com- munities, 576 U.S. at 543 (quoting Griggs, 401 U.S. at 431). Thus, a challenged practice need only be a “reasonable” way of accomplishing a “valid” interest. Id. at 541. Business necessity must be interpreted “flexibly to mean ‘appropriate’ and ‘helpful’ to achieving a certain end.” Application of the Randolph-Sheppard Act to the United States Mint, 48 Op. O.L.C. __, at *7 (Sept. 9, 2024) (quoting Ayestas v. Davis, 584 U.S. 28, 44 (2018)); see also M‘Culloch v. Maryland, 17 U.S. (4 Wheat.) 316, 413, 418 (1819) (“‘[N]ecessary’ . . . frequently imports no more than that one thing is convenient, or useful, or essential to another.”). If construed more stringently, the business-necessity defense would turn disparate-impact liability into a quota or proportionality requirement ra- ther than “an evidentiary tool” used to “smoke out” intentional discrim- ination. Ricci, 557 U.S. at 595 (Scalia, J., concurring). An appropriately low threshold to establish a valid business-necessity defense allows for “competing explanations” of racial “disparities” to determine whether the challenged policy is an innocent business deci- sion or the “functional[] equivalent to intentional discrimination.” Wat- son, 487 U.S. at 987. A sufficiently strong “risk of purposeful discrimi- nation,” City of Rome, 446 U.S. at 177, exists only “when all legitimate reasons” for a challenged policy “have been eliminated as possible rea- sons,” such that “an inference of discrimination” is likely, Furnco Con- str. Corp. v. Waters, 438 U.S. 567, 577 (1978) (discussing analogous defense to disparate-treatment actions). A “justification which is reason- ably related to the achievement of some legitimate goal” dispels an in- ference based on the adverse effects of a challenged policy. Id. at 578.9

Need to Know (Mar. 11, 2014), https://www.eeoc.gov/laws/guidance/background- checks-what-employers-need-know [https://perma.cc/RG33-DFBY]. Such disparate- impact guidance is inconsistent with the strong presumption of job -relatedness and the constitutionally mandated leeway for demonstrating business necessity. 9 Notwithstanding the low threshold for showing business necessity, the creation of

a racially diverse workplace does not qualify as such, because it is not race -neutral. See 42 U.S.C. § 2000e-2(k)(2) (“[D]emonstration that an employment practice is required by business necessity may not be used as a defense against a claim of intentional dis- crimination.”); see also id. § 2000e-2(m). There is no “diversity” exception to this rule. See Massey v. Borough of Bergenfield, 169 F.4th 188, 199 (3d Cir. 2026) (“[A] diversity preference is not a legitimate, non-discriminatory reason for an employment decision.”); Taxman v. Bd. of Educ. of the Twp. of Piscataway, 91 F.3d 1547, 1558–59 (3d Cir. 1996) (en banc). Basing employment decisions on the racial preferences of clients, customers,

15 50 Op. O.L.C. __ (June 9, 2026)

Second, any disparate-impact claim must satisfy a robust causality requirement to avoid reading Title VII as imposing an unconstitutional quota system. See Inclusive Communities, 576 U.S. at 542–43. To establish a prima facie case of disparate impact, a plaintiff must “demonstrate[]” that the “particular” policy actually “causes a disparate impact.” 42 U.S.C. § 2000e-2(k)(1)(A)(i) (emphasis added); see also Wards Cove, 490 U.S. at 657–58. Shielding businesses from litigation over “racial disparities they did not create” helps ensure that they will not resort to impermissible “racial quotas” to stave off lawsuits. Inclu- sive Communities, 576 U.S. at 542–43. If businesses “could be haled into court and forced to engage in the expensive and time-consuming task of defending” their policies based on the mere existence of “racial[] imbal- ance[s]” without a clear causal link between the policy and the outcome, then “racial quotas” would become the “only practicable option.” Wards Cove, 490 U.S. at 652. Such quotas are per se unconstitutional. See Race- Based Education Programs at *4 (citing Grutter v. Bollinger, 539 U.S. 306, 334 (2003)).10 To avoid such an unconstitutional result, plaintiffs must plead the specific employment practice challenged and plausibly allege factual content showing that the practice causes the disparate im- pact. Cf. Ashcroft v. Iqbal, 556 U.S. 662, 682–83 (2009). And at later stages of litigation, the plaintiff must continue to shoulder the burden of proving causation. Third, the plaintiff must offer a viable alternative to the challenged practice that accomplishes the employer’s legitimate goals “just as well.” Callais, 146 S. Ct. at 1159. “[B]efore rejecting a business

or coworkers constitutes intentional race discrimination. And workplace requirements aimed at fostering racial diversity involve “artificial, arbitrary, and unnecessary barri- ers” that have no plausible job-relatedness. Inclusive Communities, 576 U.S. at 543 (ci- tation omitted). In narrow circumstances, “a bona fide occupational qualification” re- lated to religion, sex, or national origin may be “reasonably necessary to the normal operation of [a] particular business or enterprise,” but even there, race and ethnicity cannot be bona fide occupational qualifications. 42 U.S.C. § 2000e-2(e). For similar reasons, racial or ethnic diversity cannot be a legitimate business goal that defeats a disparate-impact challenge to a facially neutral policy. “[T]here is no room under cur- rent Title VII doctrine for diversity preferences to be justified as business necessities.” Kingsley R. Browne, Title VII and Diversity, 14 Nev. L.J. 806, 824 (2014). 10 A robust causality requirement is also essential for practical reasons. It would be impossible to apply the business-necessity defense unless the challenger identified a specific practice causing the disparate impact, because otherwise the employer would not know what practice it needed to justify.

16 Constitutionality of Disparate-Impact Liability Under Title VII

justification . . . , a court must determine that a plaintiff has shown that there is ‘an available alternative practice that has less disparate impact and serves the entity’s legitimate needs.’” Inclusive Communities, 576 U.S. at 533 (alterations accepted) (emphasis added) (quoting Ricci, 557 U.S. at 586); see also 42 U.S.C. § 2000e–2(k)(1)(C) (acquiescing “to the concept of ‘alternative employment practice’”). And, of course, any pro- posed alternative must be “equally effective” in achieving the em- ployer’s legitimate goals, including with respect to “the cost or other burdens.” Wards Cove, 490 U.S. at 661 (quoting Watson, 487 U.S. at 998 (plurality opinion)); cf. Callais, 146 S. Ct. at 1159 (requiring plain- tiffs who allege racial vote dilution to, at minimum, submit an alternative map showing how the state could achieve “all” of its “legitimate district- ing objectives . . . just as well”); Alexander v. S.C. State Conf. of the NAACP, 144 S. Ct. 1221, 1249 (2024) (concluding that an alternative map must show how the state “‘could have achieved its legitimate polit- ical objectives’ . . . while producing ‘significantly greater racial bal- ance’” (citation omitted)). A plaintiff must identify an alternative em- ployment practice and prove both that it would have less disparate impact and be equally effective. It is only the refusal to adopt an equally effective and administrable alternative “without a similarly undesirable racial effect” that permits a strong inference that the challenged employment policy is “a ‘pretext’ for discrimination.” Albemarle Paper Co. v. Moody, 422 U.S. 405, 425 (1975) (citation omitted); see also Callais, 146 S. Ct. at 1156 (Section 2 of the Voting Rights Act “imposes liability only when the circumstances give rise to a strong inference that intentional discrimination occurred.”). When an alternative comes with no downside, it is more likely that a defendant adopted the challenged policy “at least in part ‘because of’” its disparate “effects.” Feeney, 442 U.S. at 279; see also Wards Cove, 490 U.S. at 660–61. But that conclusion of likely discriminatory intent does not follow if a business declines to adopt an alternative that is less effective, more burdensome, or more costly. In that instance, it is far more likely that the business rejected the alternative on those legitimate, race-neutral grounds than because the alternative has less of a disparate impact. See Watson, 487 U.S. at 998 (plurality opinion). The plaintiff thus bears the burden to “rul[e] out the competing expla- nation” for the challenged business practice by showing that it “cannot

17 50 Op. O.L.C. __ (June 9, 2026)

be explained” apart from intentional discrimination. See Callais, 146 S. Ct. at 1157–58 (citation omitted). And for the same reasons, if the cost or effectiveness of an alternative is reasonably debatable, that alone would not create sufficient risk of intentional discrimination to justify subordinating the employer’s assessment simply to achieve a different racial outcome. Just as the employer must receive significant latitude to define its business goals, so too must the employee bear a correspond- ingly high burden to identify and prove an alternative. He cannot simply rely on debatable grounds of cost or efficacy, because that would smug- gle a stricter necessity standard into the last analytic step. Properly understood, disparate-impact liability under Title VII serves only as an “evidentiary device aimed at ferreting out” practices that re- flect a significant likelihood of intentional discrimination. Primus, Round Three at 498 & n.24. Nothing in our opinion today precludes the use of statistical evidence as one evidentiary factor from which to infer intentional discrimination based on a protected characteristic. See Ar- lington Heights, 429 U.S. at 265–68 (identifying several factors to con- sider); Coalition for TJ, 146 S. Ct. at 545 (Alito, J., dissenting from de- nial of certiorari). For example, statistics may be used as an evidentiary device supporting a “pattern or practice” claim of intentional employ- ment discrimination. 42 U.S.C. § 2000e-6(a); see Teamsters, 431 U.S. at 338–40 (concluding that statistical evidence, considered alongside rele- vant facts and circumstances, including personal experiences, may be sufficient to establish a prima facie case of intentional discrimination).

IV.

With that interpretation of Title VII in mind, we conclude that EEOC’s opinions, interpretative rules, and guidance documents imple- menting disparate-impact liability violate both Title VII and the Consti- tution.11 Because EEOC is the entity established by Congress to enforce Title VII against covered private and public employers, regulated parties

11 In Smith v. City of Jackson, 544 U.S. 228 (2005), the Supreme Court extended

disparate-impact liability to age-discrimination claims under the Age Discrimination in Employment Act of 1967 (“ADEA”), see id. at 240. But unlike the protected character- istics listed in Title VII, age is not a “suspect class.” Mass. Bd. of Ret. v. Murgia, 427 U.S. 307, 313 (1976) (per curiam). Disparate-impact liability in that context does not raise the same problems as in Title VII, and we do not purport to address disparate- impact liability under the ADEA here.

18 Constitutionality of Disparate-Impact Liability Under Title VII

must have confidence that its guidance accurately reflects the best inter- pretation of the law. See Loper Bright Enters. v. Raimondo, 144 S. Ct. 2244, 2266 (2024) (holding that “if [an agency’s interpretation] is not the best, it is not permissible”). We now address two unconstitutional and unlawful features of that implementation structure.

The Guidelines unconstitutionally saddle employers with an atextual and onerous validation requirement for employment practices that are presumptively job-related and matters of business necessity. Under this framework, validation is the main way to show business necessity and job relatedness. But the Guidelines’ validation structure imposes numer- ous additional burdens on nearly all employers and radically deviates from Title VII’s burden-shifting structure. If “a total selection process [as defined in the Guidelines] has an ad- verse impact,” 29 C.F.R. § 1607.15(A)(3)(a), then an employer must adduce “evidence of both the validity and utility of [that] selection procedure,” id. § 1607.5(G). Not only that, but the employer must also fine-tune its analysis to identify the specific cause of the disparity—im- properly putting the burden on the employer to show specific causation rather than the challenger. See id. §§ 1607.4(C), 1607.15(A)(2)(a). In other words, under the Guidelines, the employer must undertake a vali- dation study to “prove that the test or other criterion is positively corre- lated with successful job performance and also that the correlation is strong enough to justify the test or criterion’s disparate impact.” Heriot, Presumptively Illegal at 57. The Guidelines impose detailed require- ments for validation studies by providing “[g]eneral standards,” 29 C.F.R. § 1607.5, evidentiary documentation standards, id. § 1607.15, and “[t]echnical standards,” id. § 1607.14. For evidentiary documenta- tion alone, the Guidelines enumerate both “essential” and “desirable” standards. Id. § 1607.15(A)(3)(c). A validation study lacking infor- mation deemed “essential” is considered “incomplete,” unless the em- ployer provides a sufficient justification for omitting it. See id. A review of only the “essential” elements illustrates the Guidelines’ overbearing granularity. The Guidelines impose 34 essential require- ments for criterion-related validity, 19 for content validity, and 25 for

19 50 Op. O.L.C. __ (June 9, 2026)

construct validity. See id. § 1607.15(B)–(D).12 These requirements in- clude providing a “full description of all criteria on which data were col- lected and means by which they were observed, recorded, evaluated, and quantified,” id. § 1607.15(B)(5); a description of the “[m]easures of cen- tral tendency (e.g., means) and measures of dispersion (e.g., standard deviations and ranges) for all selection procedures,” id. § 1607.15(B)(8); and any “statistical adjustments, such as for less then [sic] perfect relia- bility or for restriction of score range in the selection procedure or crite- rion,” id.13 To the typical employer, these requirements are “incompre- hensible.” Heriot, Presumptively Illegal at 56. Their complexity explains why employers generally rely on secondary sources that are themselves lengthy and “highly-technical” to comprehend the Guidelines’ byzantine requirements. See id. at 57 & n.131 (describing a popular 192-page “tome”). These validation-study guidelines are inconsistent with Title VII’s business-necessity defense, properly understood. See Inclusive Commu- nities, 576 U.S. at 541. Although the Guidelines’ validation processes might find support under an exceedingly stringent interpretation of the business-necessity defense, 42 U.S.C. § 2000e-2(k)(1)(A)(i) (requiring employers to “demonstrate that the challenged practice is job related for the position in question and consistent with business necessity”), that interpretation is neither the best nor the only permissible way to read the statute. That is why, to avoid interpreting Title VII as imposing an un- constitutional duty on employers, the Supreme Court has said that an employer need only show that the challenged policy is a “reasonable”

12 Criterion-related validity is “[d]emonstrated by empirical data showing that the

selection procedure is predictive of or significantly correlated with important elements of work behavior,” 29 C.F.R. § 1607.16(F); content validity is “[d]emonstrated by data showing that the content of a selection procedure is representative of important aspects of performance on the job,” id. § 1607.16(D); and construct validity is “[d]emonstrated by data showing that the selection procedure measures the degree to which candidates have identifiable characteristics which have been determined to be important for suc- cessful job performance,” id. § 1607.16(E). 13 EEOC states that “[t]hese guidelines do not require a user to conduct validity stud-

ies of selection procedures where no adverse impact results.” 29 C.F.R. § 1607.1(B). But an employer frequently will not know whether an adverse impact on any particular group has resulted from the employer’s procedures until it performs a disparate-impact study. And virtually every employment practice has at least some adverse impact on some protected group. EEOC’s regulations therefore have the effect of injecting racial considerations into the evaluation of nearly all employment practices.

20 Constitutionality of Disparate-Impact Liability Under Title VII

way of accomplishing a “valid interest.” Inclusive Communities, 576 U.S. at 541. We follow the Court’s lead on the most harmonious reading of Title VII and the Constitution. The Guidelines sprint in the opposite direction, nitpicking and second- guessing eminently reasonable employment practices and contemplating an exceedingly high burden to validate an employment policy. See 29 C.F.R. § 1607.3(A) (labeling unlawful any policy that has a disparate impact “unless the procedure has been validated in accordance with these guidelines”). Rather than screening for “artificial, arbitrary, and unnecessary barriers” that have no plausible job-relatedness, Inclusive Communities, 576 U.S. at 543 (quoting Griggs, 401 U.S. at 431), the Guidelines’ validation studies require an employer to follow somewhere between 19 and 34 “essential” requirements just for documenting evi- dence—let alone abiding by other requirements, such as the “technical standards” in performing the validation, see 29 C.F.R §§ 1607.14, 1607.15(B)–(D). The pages upon pages of labyrinthine regulations gov- erning validation studies hardly give employers the necessary “leeway” to justify their employment practices based on “reasonable” and “valid . . . private priorities.” Inclusive Communities, 576 U.S. at 541, 544. As a result, employers not only are dragooned into developing costly validation studies but also are coerced to engage in even costlier litiga- tion over whether the study passes muster under the Guidelines. See, e.g., Lopez v. City of Lawrence, 823 F.3d 102, 112–14 (1st Cir. 2016). These stifling demands make it less likely that employers will experiment with “innovative employment practices.” Heriot, Presumptively Illegal at 57. Even worse, such onerous burdens pressure employers to avoid liability through race-based decisionmaking. An employer who satisfies EEOC’s validation-study requirements has done far more than Ti- tle VII’s business-necessity defense demands—and has been given far less leeway than the Constitution requires.

EEOC’s Affirmative Action Guidelines run further into unconstitu- tional territory, purporting to authorize—and expressly encouraging— racial preferences (what the regulations refer to as “affirmative action”) in response to actual or anticipated disparate impacts. See 29 C.F.R. § 1608.1 et seq. The regulations claim that it is the “clear . . . intent” of Title VII “to encourage voluntary affirmative action.” Id. § 1608.1(a).

21 50 Op. O.L.C. __ (June 9, 2026)

The regulations then purport to authorize employers to “take affirmative action based on an analysis which reveals facts constituting actual or potential adverse impact, if such adverse impact is likely to result from existing or contemplated practices.” Id. § 1608.3(a). The regulations establish “three elements” for an “affirmative action plan or program.” Id. § 1608.4. First, employers should conduct a “[r]ea- sonable self analysis” to “determine whether employment practices do, or tend to, . . . result in adverse impact.” Id. § 1608.4(a). “In conducting a self analysis,” the regulations note, “the employer . . . should be con- cerned with the effect on its employment practices of circumstances which may be the result of discrimination by other persons or institu- tions.” Id. Second, employers should have a “reasonable basis” for using racial preferences, including any evidence that their employment prac- tices “[h]ave or tend to have an adverse effect.” Id. § 1608.4(b)(1). Im- portantly, the regulations make clear that “[i]t is not necessary [to] es- tablish a violation of title VII” before adopting racial preferences because a “reasonable basis exists . . . whether there exists arguable de- fenses to a title VII action.” Id. § 1608.4(b). Third, employers may take “reasonable action,” which “may include goals and timetables or other appropriate employment tools which recognize the race, sex, or national origin of applicants or employees” and “the adoption of practices which will eliminate . . . actual or potential adverse impact.” Id. § 1608.4(c). “[A]ffirmative action programs may in design and execution be race, color, sex, or ethnic conscious.” Id. § 1607.4(E). Federal endorsement of racial preferences is inconsistent with Ti- tle VII and prohibited by the Constitution. Nothing in Title VII requires or authorizes employers to adopt racial preferences. To the contrary, Ti- tle VII disclaims any interpretation requiring an employer “to grant pref- erential treatment to any individual or to any group because of the race, color, religion, sex, or national origin of such individual or group on account of [a statistical] imbalance” among existing employees or appli- cants from the available work force. 42 U.S.C. § 2000e-2(j). What is more, the regulations’ endorsement of racial preferences in certain circumstances—though such circumstances are limited—encourages employers to consider “the effect . . . of discrimination by other persons or institutions,” 29 C.F.R. § 1608.4(a), ignoring disparate-impact liabil- ity’s “robust causality requirement,” Inclusive Communities, 576 U.S. at 542; see Wards Cove, 490 U.S. at 656. And by encouraging racial

22 Constitutionality of Disparate-Impact Liability Under Title VII

preferences without an “establish[ed] . . . violation of title VII,” 29 C.F.R. § 1608.4(b), the regulations violate Supreme Court precedent that requires a much stronger showing before allowing race-conscious reme- dial action, see Ricci, 557 U.S. at 585 (concluding that Title VII requires at least a “strong basis in evidence” that disparate impact would result before an employer takes race-based remedial measures); see also id. at 584 (reserving the question of whether even “the strong-basis-in- evidence standard would satisfy the Equal Protection Clause”). The Supreme Court had initially indicated that Title VII’s prohibition against racial discrimination did not categorically “condemn all private, voluntary, race-conscious affirmative action plans,” and permitted their use under certain, limited circumstances. United Steelworkers of Am. v. Weber, 443 U.S. 193, 208 (1979); see also Johnson, 480 U.S. at 626–27. Some might argue that the Court made this allowance “at a time when this Court often paid insufficient attention to the language of statutory provisions,” Callais, 146 S. Ct. at 1146; that it is inconsistent with the logic of Ames; and that it has been fatally undercut by the Court’s more recent decisions concerning racial preferences, see SFFA, 143 S. Ct. at 2209 (Gorsuch, J., concurring) (linking Title VI’s prohibition against “intentionally treating any individual worse even in part because of his race, color, or national origin, . . . without regard to any other reason or motive the recipient [of federal funds] might assert,” to Title VII’s counterpart provisions). At a minimum, though, the Court has held that race-based decisionmaking is permissible only to remediate “specific in- stances of past discrimination,” Callais, 146 S. Ct. at 1153, and we con- clude that the regulations’ encouragement for employers to consider race sweeps too broadly to meet that standard. Even if Title VII had authorized employers to favor some races over others (contrary to the statutory text) and an employer had a strong basis in evidence to believe that a disparate impact would result but for so- called “affirmative action,” Congress could not constitutionally require or pressure employers to use such racial preferences. See SFFA, 143 S. Ct. at 2168–70; see also Anderson, 375 U.S. at 404 (holding that the government cannot encourage unlawful discrimination). After all, the regulations’ adoption of “goals and timetables . . . which recognize the race, sex, or national origin of applicants or employees,” 29 C.F.R. § 1608.4(c), places the federal government’s imprimatur on the imper- missible “[c]lassifying and assigning” of benefits and burdens based on

23 50 Op. O.L.C. __ (June 9, 2026)

protected characteristics, SFFA, 143 S. Ct. at 2166 (quoting Parents In- volved in Cmty. Schs. v. Seattle Sch. Dist. No. 1, 551 U.S. 701, 735 (2007)). The regulations’ use of racial preferences must fail because Congress has never demonstrated that “racial classifications” under Title VII are necessary to “remediat[e] specific, identified instances of past discrimi- nation that violated the Constitution or a statute.” Race-Based Education Programs at *6 (quoting SFFA, 143 S. Ct. at 2162); see also Callais, 146 S. Ct. at 1152–53 (same). It has never made “particularized eviden- tiary findings” to demonstrate a compelling interest in requiring employ- ers to adopt such racial or ethnic classifications. Race-Based Education Programs at *7. The racial-preference regulations also fail narrow tailoring because they function as an open-ended racial-balancing mechanism. They en- courage racial quotas, lack any definite endpoint, rely on arbitrary and undefined racial categories, and impose burdens on other racial groups. See id. at *7–8. The guidelines purport to authorize racial “goals and timetables” without any definite stopping point. 29 C.F.R. § 1608.4(c); see also Callais, 146 S. Ct. at 1160–61 (confirming that race-based preferences, “even if permissible” when adopted, cannot “extend indef- initely into the future”); SFFA, 143 S. Ct. at 2165–66 (emphasizing that racial classifications “must be limited in time” and “have a logical end point” (citation omitted)). Indeed, the regulations invite employers to adjust workforce composition until a racial target is reached, which is the hallmark of a quota. The endorsement of racial preferences likewise relies on “overbroad” and “arbitrary” racial and ethnic categories that are both under- and overinclusive (such as “Asian” and “Hispanic”). Compare 29 C.F.R. § 1607.4(B), with SFFA, 143 S. Ct. at 2167–68.14 Finally, an employer’s use of racial preferences in accordance with the regulations shifts the costs of those preferences onto members of other protected groups. Putting a thumb on the racial scale in this manner would alone make the regulations’ endorsement of racial preferences un- lawful and unconstitutional.

14 Consistent with the fact that the Supreme Court has recognized that racial and eth-

nic categories such as “Hispanic” are overbroad and arbitrary, Congress can prohibit employers from engaging in practices that would consider (or likely consider) those improper traits. Requiring employers to affirmatively consider them, by contrast, runs afoul of the narrow tailoring that the Supreme Court requires.

24 Constitutionality of Disparate-Impact Liability Under Title VII

V.

Disparate-impact liability as currently interpreted by EEOC is uncon- stitutional. We interpret Title VII to conform to the safeguards on dis- parate-impact liability that the Supreme Court has indicated are consti- tutionally required. Setting an appropriate threshold for business necessity, requiring robust causality, and demanding evidence of an equally effective alternative practice that causes less disparate impact are essential prerequisites to any imposition of liability. EEOC’s regula- tions and guidance fail to incorporate these safeguards in key respects, and the validation-study and affirmative-action regulations conflict with Title VII and the Constitution.

T. ELLIOT GAISER Assistant Attorney General Office of Legal Counsel

JOSHUA J. CRADDOCK Deputy Assistant Attorney General Office of Legal Counsel