This case requires us to decide what evidentiary standards should be applied under Title VII of the Civil Rights Act of 1964, 78 Stat. 253, as amended, 42 U. S. C. § 2000e et seq., in determining whether an employer's practice of committing promotion decisions to the subjective discretion of supervisory employees has led to illegal discrimination.
Petitioner Clara Watson, who is black, was hired by respondent Fort Worth Bank and Trust (the Bank) as a proof operator in August 1973. In January 1976, Watson was promoted to a position as teller in the Bank's drive-in facility. In February 1980, she sought to become supervisor of the tellers in the main lobby; a white male, however, was selected for this job. Watson then sought a position as supervisor of the drive-in bank, but this position was given to a white female. In February 1981, after Watson had served for about a year as a commercial teller in the Bank's main lobby, and informally as assistant to the supervisor of tellers, the man holding that position was promoted. Watson applied for the vacancy, but the white female who was the supervisor of the drive-in bank was selected instead. Watson then applied for the vacancy created at the drive-in; a white male was selected for that job. The Bank, which has about 80 employees, had not developed precise and formal criteria for evaluating candidates for the positions for which Watson unsuccessfully applied. It relied instead on the subjective judgment of supervisors who were acquainted with the candidates and with the nature of the jobs to be filled. All the supervisors involved in denying Watson the four promotions at issue were white.
The District Court addressed Watson's individual claims under the evidentiary standards that apply in a discriminatory treatment case. See McDonnell Douglas Corp. v. Green, 411 U.S. 792 (1973), and Texas Dept. of Community Affairs v. Burdine, 450 U.S. 248 (1981). It concluded, on the evidence presented at trial, that Watson had established a prima facie case of employment discrimination, but that the
A divided panel of the United States Court of Appeals for the Fifth Circuit affirmed in part. 798 F.2d 791 (1986). The majority concluded that there was no abuse of discretion in the District Court's class decertification decisions. In order to avoid unfair prejudice to members of the class of black job applicants, however, the Court of Appeals vacated the portion of the judgment affecting them and remanded with instructions to dismiss those claims without prejudice. The majority affirmed the District Court's conclusion that Watson had failed to prove her claim of racial discrimination under the standards set out in McDonnell Douglas, supra, and Burdine, supra.
Watson argued that the District Court had erred in failing to apply "disparate impact" analysis to her claims of discrimination in promotion. Relying on Fifth Circuit precedent, the majority of the Court of Appeals panel held that "a Title VII challenge to an allegedly discretionary promotion system is properly analyzed under the disparate treatment model rather than the disparate impact model." 798 F. 2d, at 797. Other Courts of Appeals have held that disparate impact analysis may be applied to hiring or promotion systems that involve the use of "discretionary" or "subjective" criteria. See, e. g., Atonio v. Wards Cove Packing Co., 810 F.2d 1477 (CA9) (en banc), on return to panel, 827 F.2d 439
Section 703 of the Civil Rights Act of 1964, 42 U. S. C. § 2000e-2, provides:
Several of our decisions have dealt with the evidentiary standards that apply when an individual alleges that an employer has treated that particular person less favorably than
In Griggs v. Duke Power Co., 401 U.S. 424 (1971), this Court held that a plaintiff need not necessarily prove intentional discrimination in order to establish that an employer has violated § 703. In certain cases, facially neutral employment practices that have significant adverse effects on protected groups have been held to violate the Act without proof
The distinguishing features of the factual issues that typically dominate in disparate impact cases do not imply that the ultimate legal issue is different than in cases where disparate treatment analysis is used. See, e. g., Washington v. Davis, 426 U.S. 229, 253-254 (1976) (STEVENS, J., concurring). Nor do we think it is appropriate to hold a defendant liable for unintentional discrimination on the basis of less evidence than is required to prove intentional discrimination. Rather, the necessary premise of the disparate impact approach is that some employment practices, adopted without a deliberately discriminatory motive, may in operation be functionally equivalent to intentional discrimination.
Perhaps the most obvious examples of such functional equivalence have been found where facially neutral job requirements necessarily operated to perpetuate the effects of intentional discrimination that occurred before Title VII was enacted. In Griggs itself, for example, the employer had a history of overt racial discrimination that predated the enactment of the Civil Rights Act of 1964. 401 U. S., at 426-428. Such conduct had apparently ceased thereafter, but the employer continued to follow employment policies that had "a markedly disproportionate" adverse effect on blacks. Id., at 428-429. Cf. Teamsters, supra, at 349, and n. 32. The Griggs Court found that these policies, which involved the use of general aptitude tests and a high school diploma
This Court has repeatedly reaffirmed the principle that some facially neutral employment practices may violate Title VII even in the absence of a demonstrated discriminatory intent. We have not limited this principle to cases in which the challenged practice served to perpetuate the effects of pre-Act intentional discrimination. Each of our subsequent decisions, however, like Griggs itself, involved standardized employment tests or criteria. See, e. g., Albemarle Paper Co. v. Moody, 422 U.S. 405 (1975) (written aptitude tests); Washington v. Davis, supra (written test of verbal skills); Dothard v. Rawlinson, 433 U.S. 321 (1977) (height and weight requirements); New York City Transit Authority v. Beazer, 440 U.S. 568 (1979) (rule against employing drug addicts); Connecticut v. Teal, 457 U.S. 440 (1982) (written examination). In contrast, we have consistently used conventional disparate treatment theory, in which proof of intent to discriminate is required, to review hiring and promotion decisions that were based on the exercise of personal judgment or the application of inherently subjective criteria. See, e. g., McDonnell Douglas Corp. v. Green, supra (discretionary decision not to rehire individual who engaged in criminal acts against employer while laid off); Furnco Construction Corp. v. Waters, 438 U.S. 567 (1978) (hiring decisions based on personal knowledge of candidates and recommendations); Texas Dept. of Community Affairs v. Burdine, supra (discretionary decision to fire individual who was said not to get along with co-workers); United States Postal Service
Our decisions have not addressed the question whether disparate impact analysis may be applied to cases in which subjective criteria are used to make employment decisions. As noted above, the Courts of Appeals are in conflict on the issue. In order to resolve this conflict, we must determine whether the reasons that support the use of disparate impact analysis apply to subjective employment practices, and whether such analysis can be applied in this new context under workable evidentiary standards.
The parties present us with stark and uninviting alternatives. Petitioner contends that subjective selection methods are at least as likely to have discriminatory effects as are the kind of objective tests at issue in Griggs and our other disparate impact cases. Furthermore, she argues, if disparate impact analysis is confined to objective tests, employers will be able to substitute subjective criteria having substantially identical effects, and Griggs will become a dead letter. Respondent and the United States (appearing as amicus curiae) argue that conventional disparate treatment analysis is adequate to accomplish Congress' purpose in enacting Title VII. They also argue that subjective selection practices would be so impossibly difficult to defend under disparate impact analysis that employers would be forced to adopt numerical quotas in order to avoid liability.
We are persuaded that our decisions in Griggs and succeeding cases could largely be nullified if disparate impact analysis were applied only to standardized selection practices. However one might distinguish "subjective" from "objective" criteria, it is apparent that selection systems that combine both types would generally have to be considered subjective in nature. Thus, for example, if the employer in Griggs had consistently preferred applicants who had a high school diploma
We are also persuaded that disparate impact analysis is in principle no less applicable to subjective employment criteria than to objective or standardized tests. In either case, a facially neutral practice, adopted without discriminatory intent, may have effects that are indistinguishable from intentionally discriminatory practices. It is true, to be sure, that an employer's policy of leaving promotion decisions to the unchecked discretion of lower level supervisors should itself raise no inference of discriminatory conduct. Especially in relatively small businesses like respondent's, it may be customary and quite reasonable simply to delegate employment decisions to those employees who are most familiar with the jobs to be filled and with the candidates for those jobs. It does not follow, however, that the particular supervisors to whom this discretion is delegated always act without discriminatory intent. Furthermore, even if one assumed that any such discrimination can be adequately policed through disparate treatment analysis, the problem of subconscious stereotypes and prejudices would remain. In this case, for example, petitioner was apparently told at one point that the teller position was a big responsibility with "a lot of money . . . for blacks to have to count." App. 7. Such remarks may not prove discriminatory intent, but they do suggest a lingering form of the problem that Title VII was enacted to combat. If an employer's undisciplined system of subjective decisionmaking has precisely the same effects as
Having decided that disparate impact analysis may in principle be applied to subjective as well as to objective practices, we turn to the evidentiary standards that should apply in such cases. It is here that the concerns raised by respondent have their greatest force. Respondent contends that a plaintiff may establish a prima facie case of disparate impact through the use of bare statistics, and that the defendant can rebut this statistical showing only by justifying the challenged practice in terms of "business necessity," Griggs, 401 U. S., at 431, or "job relatedness," Albemarle Paper Co., 422 U. S., at 426. Standardized tests and criteria, like those at issue in our previous disparate impact cases, can often be justified through formal "validation studies," which seek to determine whether discrete selection criteria predict actual on-the-job performance. See generally id., at 429-436. Respondent warns, however, that "validating" subjective selection criteria in this way is impracticable. Some qualities — for example, common sense, good judgment, originality, ambition, loyalty, and tact — cannot be measured accurately through standardized testing techniques. Moreover, success at many jobs in which such qualities are crucial cannot itself be measured directly. Opinions often differ when managers and supervisors are evaluated, and the same can be said for many jobs that involve close cooperation with one's co-workers or complex and subtle tasks like the provision of
We agree that the inevitable focus on statistics in disparate impact cases could put undue pressure on employers to adopt inappropriate prophylactic measures. It is completely unrealistic to assume that unlawful discrimination is the sole cause of people failing to gravitate to jobs and employers in accord with the laws of chance. See Sheet Metal Workers v. EEOC, 478 U.S. 421, 489 (1986) (O'CONNOR, J., concurring in part and dissenting in part). It would be equally unrealistic to suppose that employers can eliminate, or discover and explain, the myriad of innocent causes that may lead to statistical imbalances in the composition of their work forces. Congress has specifically provided that employers are not required to avoid "disparate impact" as such:
We do not believe that disparate impact theory need have any chilling effect on legitimate business practices. We recognize, however, that today's extension of that theory into the context of subjective selection practices could increase the risk that employers will be given incentives to adopt quotas or to engage in preferential treatment. Because Congress has so clearly and emphatically expressed its intent that Title VII not lead to this result, 42 U. S. C. § 2000e-2(j), we think it imperative to explain in some detail why the evidentiary standards that apply in these cases should serve as adequate safeguards against the danger that Congress recognized.
First, we note that the plaintiff's burden in establishing a prima facie case goes beyond the need to show that there are statistical disparities in the employer's work force. The plaintiff must begin by identifying the specific employment practice that is challenged. Although this has been relatively easy to do in challenges to standardized tests, it may sometimes be more difficult when subjective selection criteria are at issue. Especially in cases where an employer combines subjective criteria with the use of more rigid standardized rules or tests, the plaintiff is in our view responsible for isolating and identifying the specific employment practices that are allegedly responsible for any observed statistical disparities. Cf. Connecticut v. Teal, 457 U.S. 440 (1982).
Once the employment practice at issue has been identified, causation must be proved; that is, the plaintiff must offer statistical evidence of a kind and degree sufficient to show that the practice in question has caused the exclusion of applicants for jobs or promotions because of their membership in a protected group. Our formulations, which have never
A second constraint on the application of disparate impact theory lies in the nature of the "business necessity" or "job relatedness" defense. Although we have said that an employer has "the burden of showing that any given requirement must have a manifest relationship to the employment in question," Griggs, 401 U. S., at 432, such a formulation should not be interpreted as implying that the ultimate burden of proof can be shifted to the defendant. On the contrary, the ultimate burden of proving that discrimination against a protected group has been caused by a specific employment practice remains with the plaintiff at all times.
Our cases make it clear that employers are not required, even when defending standardized or objective tests, to introduce formal "validation studies" showing that particular criteria predict actual on-the-job performance. In Beazer, for example, the Court considered it obvious that "legitimate employment goals of safety and efficiency" permitted the exclusion of methadone users from employment with the New York City Transit Authority; the Court indicated that the "manifest relationship" test was satisfied even with respect to non-safety-sensitive jobs because those legitimate goals were "significantly served by" the exclusionary rule at issue in that case even though the rule was not required by those goals. 440 U. S., at 587, n. 31. Similarly, in Washington v. Davis, the Court held that the "job relatedness" requirement was satisfied when the employer demonstrated that a written test was related to success at a police training academy "wholly aside from [the test's] possible relationship to actual performance as a police officer." 426 U. S., at 250. See also id., at 256 (STEVENS, J., concurring) ("[A]s a matter of law, it is permissible for the police department to use a test
In the context of subjective or discretionary employment decisions, the employer will often find it easier than in the case of standardized tests to produce evidence of a "manifest relationship to the employment in question." It is self-evident that many jobs, for example those involving managerial responsibilities, require personal qualities that have never been considered amenable to standardized testing. In evaluating claims that discretionary employment practices are insufficiently related to legitimate business purposes, it must be borne in mind that "[c]ourts are generally less competent than employers to restructure business practices, and unless mandated to do so by Congress they should not attempt it." Furnco Construction Corp. v. Waters, 438 U. S., at 578. See also Zahorik v. Cornell University, 729 F.2d 85, 96 (CA2 1984) ("[The] criteria [used by a university to award tenure], however difficult to apply and however much disagreement they generate in particular cases, are job related.. . . It would be a most radical interpretation of Title VII for a court to enjoin use of an historically settled process and plainly relevant criteria largely because they lead to decisions which are difficult for a court to review"). In sum, the high standards of proof in disparate impact cases are sufficient in our view to avoid giving employers incentives to modify any normal and legitimate practices by introducing quotas or preferential treatment.
We granted certiorari to determine whether the court below properly held disparate impact analysis inapplicable to a subjective or discretionary promotion system, and we now hold that such analysis may be applied. We express no opinion as to the other rulings of the Court of Appeals.
Neither the District Court nor the Court of Appeals has evaluated the statistical evidence to determine whether petitioner
It is so ordered.
JUSTICE KENNEDY took no part in the consideration or decision of this case.
JUSTICE BLACKMUN, with whom JUSTICE BRENNAN and JUSTICE MARSHALL join, concurring in part and concurring in the judgment.
I agree that disparate-impact analysis may be applied to claims of discrimination caused by subjective or discretionary selection processes, and I therefore join Parts I, II-A, II-B, and III of the Court's opinion. I am concerned, however, that the plurality mischaracterizes the nature of the burdens this Court has allocated for proving and rebutting disparate-impact claims. In so doing, the plurality projects an application of disparate-impact analysis to subjective employment practices that I find to be inconsistent with the proper evidentiary standards and with the central purpose of Title VII. I therefore cannot join Parts II-C and II-D. I write separately to reiterate what I thought our prior cases had made plain about the nature of claims brought within the disparate-impact framework.
The plurality's discussion of the allocation of burdens of proof and production that apply in litigating a disparate-impact claim under Title VII of the Civil Rights Act of 1964, 78 Stat. 253, as amended, 42 U. S. C. § 2000e et seq., is flatly
The plurality's suggested allocation of burdens bears a closer resemblance to the allocation of burdens we established for disparate-treatment claims in McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802-804 (1973), and Texas Dept. of Community Affairs v. Burdine, 450 U.S. 248, 252-256 (1981), than it does to those the Court has established for disparate-impact claims. Nothing in our cases supports the plurality's declaration that, in the context of a disparate-impact challenge, "the ultimate burden of proving
The violation alleged in a disparate-treatment challenge focuses exclusively on the intent of the employer. See Teamsters v. United States, 431 U.S. 324, 335, n. 15 (1977) (in disparate-treatment challenge "[p]roof of discriminatory motive is critical"). Unless it is proved that an employer intended to disfavor the plaintiff because of his membership in a protected class, a disparate-treatment claim fails. A disparate-impact claim, in contrast, focuses on the effect of the employment practice. See id., at 336, n. 15 (disparate-impact claims "involve employment practices that are facially neutral in their treatment of different groups but that in fact fall more harshly on one group than another"). Unless an employment practice producing the disparate effect is justified by "business necessity," ibid., it violates Title VII, for "good intent or absence of discriminatory intent does not redeem
In McDonnell Douglas and Burdine, this Court formulated a scheme of burden allocation designed "progressively to sharpen the inquiry into the elusive factual question of intentional discrimination." Texas Dept. of Community Affairs v. Burdine, 450 U. S., at 255, n. 8. The plaintiff's initial burden of establishing a prima facie case of disparate treatment is "not onerous," id., at 253, and "raises an inference of discrimination only because we presume these acts, if otherwise unexplained, are more likely than not based on the consideration of impermissible factors." Furnco Construction Corp. v. Waters, 438 U.S. 567, 577 (1978).
The prima facie case of disparate impact established by a showing of a significant statistical disparity is notably different. Unlike a claim of intentional discrimination, which the McDonnell Douglas factors establish only by inference, the disparate impact caused by an employment practice is directly established by the numerical disparity. Once an employment practice is shown to have discriminatory consequences, an employer can escape liability only if it persuades the court that the selection process producing the disparity has " `a manifest relationship to the employment in question.' " Connecticut v. Teal, 457 U.S. 440, 446 (1982), quoting Griggs v. Duke Power Co., 401 U. S., at 432. The plaintiff in such a case already has proved that the employment practice has an improper effect; it is up to the employer to prove that the discriminatory effect is justified.
Intertwined with the plurality's suggestion that the defendant's burden of establishing business necessity is merely one of production is the implication that the defendant may satisfy this burden simply by "producing evidence that its employment practices are based on legitimate business reasons." Ante, at 998. Again, the echo from the disparate-treatment cases is unmistakable. In that context, it is enough for an employer "to articulate some legitimate, nondiscriminatory reason" for the allegedly discriminatory act in order to rebut the presumption of intentional discrimination. McDonnell Douglas, 411 U. S., at 802. But again the plurality misses a key distinction: An employer accused of discriminating intentionally need only dispute that it had any such intent — which it can do by offering any legitimate, nondiscriminatory justification. Such a justification is simply not enough to legitimize a practice that has the effect of excluding a protected class from job opportunities at a significantly disproportionate rate. Our cases since Griggs make
Precisely what constitutes a business necessity cannot be reduced, of course, to a scientific formula, for it necessarily involves a case-specific judgment which must take into account the nature of the particular business and job in question. The term itself, however, goes a long way toward establishing the limits of the defense: To be justified as a business necessity an employment criterion must bear more than an indirect or minimal relationship to job performance. See Dothard v. Rawlinson, 433 U. S., at 331-332 (absent proof that height and weight requirements directly correlated with amount of strength deemed "essential to good job performance," requirements not justified as business necessity); Albemarle Paper Co. v. Moody, 422 U. S., at 431, quoting the Equal Employment Opportunity Commission's (EEOC's) Uniform Guidelines on Employee Selection Procedures, 29 CFR § 1607.4(c) (1974) ("The message of these Guidelines is the same as that of the Griggs case — that discriminatory tests are impermissible unless shown, by professionally acceptable methods, to be `predictive of or significantly correlated with important elements of work behavior which comprise or are relevant to the job' "). Cf. Washington v. Davis, 426 U.S. 229, 247 (1976) (Title VII litigation "involves a more probing judicial review, and less deference to the seemingly reasonable acts of [employers] than is appropriate under the Constitution where special racial impact, without discriminatory purpose, is claimed"). The criterion must directly relate to a prospective employee's ability to perform the job effectively. And even where an employer
I am also concerned that, unless elaborated upon, the plurality's projection of how disparate-impact analysis should be applied to subjective-selection processes may prove misleading. The plurality suggests: "In the context of subjective or discretionary employment decisions, the employer will often find it easier than in the case of standardized tests to produce evidence of a `manifest relationship to the employment in question.' " Ante, at 999. This statement warrants further comment in two respects.
As explained above, once it has been established that a selection method has a significantly disparate impact on a protected class, it is clearly not enough for an employer merely to produce evidence that the method of selection is job related. It is an employer's obligation to persuade the reviewing court of this fact.
While the formal validation techniques endorsed by the EEOC in its Uniform Guidelines may sometimes not be effective in measuring the job-relatedness of subjective-selection
The fact that job-relatedness cannot always be established with mathematical certainty does not free an employer from its burden of proof, but rather requires a trial court to look to different forms of evidence to assess an employer's claim of business necessity. And while common sense surely plays a part in this assessment, a reviewing court may not rely on its own, or an employer's, sense of what is "normal," ante, at 999, as a substitute for a neutral assessment of the evidence presented. Indeed, to the extent an employer's "normal" practices serve to perpetuate a racially disparate status quo, they clearly violate Title VII unless they can be shown to be necessary, in addition to being "normal." See Griggs v. Duke Power Co., 401 U. S., at 430 ("[P]ractices, procedures, or tests neutral on their face, and even neutral in terms of intent, cannot be maintained if they operate to `freeze' the [discriminatory] status quo").
The plurality's prediction that an employer "will often find it easier" ante, at 999, to justify the use of subjective practices as a business necessity is difficult to analyze in the abstract. Nevertheless, it bears noting that this statement
Allowing an employer to escape liability simply by articulating vague, inoffensive-sounding subjective criteria would disserve Title VII's goal of eradicating discrimination in employment. It would make no sense to establish a general rule whereby an employer could more easily establish business
While subjective criteria, like objective criteria, will sometimes pose difficult problems for the court charged with assessing the relationship between selection process and job performance, the fact that some cases will require courts to develop a greater factual record and, perhaps, exercise a greater degree of judgment, does not dictate that subjective-selection processes generally are to be accepted at face value, as long as they strike the reviewing court as "normal and legitimate." Ante, at 999.
JUSTICE STEVENS, concurring in the judgment.
The question we granted certiorari to decide, though extremely important, is also extremely narrow. It reads as follows:
Essentially for the reasons set forth in Parts II-A and II-B of JUSTICE O'CONNOR's opinion, I agree that this question must be answered in the affirmative. At this stage of the proceeding, however, I believe it unwise to announce a "fresh" interpretation of our prior cases applying disparate-impact analysis to objective employment criteria. See ante, at 994. Cases in which a Title VII plaintiff challenges an employer's practice of delegating certain kinds of decisions to the subjective discretion of its executives will include too many variables to be adequately discussed in an opinion that does not focus on a particular factual context. I would therefore postpone any further discussion of the evidentiary standards set forth in our prior cases until after the District Court has made appropriate findings concerning this plaintiff's prima facie evidence of disparate impact and this defendant's explanation for its practice of giving supervisors discretion in making certain promotions.
Courts have also referred to the "standard deviation" analysis sometimes used in jury-selection cases. See, e. g., Rivera v. Wichita Falls, 665 F.2d 531, 536, n. 7 (CA5 1982) (citing Casteneda v. Partida, 430 U.S. 482 (1977)); Guardians Association of New York City Police Dept. v. Civil Service Comm'n of New York, 630 F.2d 79, 86, and n. 4 (CA2 1980) (same), cert. denied, 452 U.S. 940 (1981). We have emphasized the useful role that statistical methods can have in Title VII cases, but we have not suggested that any particular number of "standard deviations" can determine whether a plaintiff has made out a prima facie case in the complex area of employment discrimination. See Hazelwood School Dist. v. United States, 433 U.S. 299, 311, n. 17 (1977).
Nor has a consensus developed around any alternative mathematical standard. Instead, courts appear generally to have judged the "significance" or "substantiality" of numerical disparities on a case-by-case basis. See Clady, supra, at 1428-1429; B. Schlei & P. Grossman, Employment Discrimination Law 98-99, and n. 77 (2d ed. 1983); id., at 18-19, and n. 33 (Supp. 1983-1985). At least at this stage of the law's development, we believe that such a case-by-case approach properly reflects our recognition that statistics "come in infinite variety and . . . their usefulness depends on all of the surrounding facts and circumstances." Teamsters v. United States, 431 U.S. 324, 340 (1977).
"(i) that he belongs to a racial minority; (ii) that he applied and was qualified for a job for which the employer was seeking applicants; (iii) that, despite his qualifications, he was rejected; and (iv) that, after his rejection, the position remained open and the employer continued to seek applicants from persons of complainant's qualifications."