By: Roberto L. Corrada*
The standard for voluntary affirmative action under Title VII has been in question in recent years. The last United States Supreme Court opinion to directly address the matter is over twenty years old, and the Court’s composition has changed since then. In the years since the last Title VII affirmative action opinion in 1987, Congress has passed the Civil Rights Act of 1991, and the constitutional standard for voluntary affirmative action has been addressed by the Court no fewer than five times. The constitutional standard had been crafted by Justice Sandra Day O’Connor; but with her retirement, both the constitutional (Fourteenth Amendment) and the statutory (Title VII) standards for affirmative action have again been obscured.
A recent case, Ricci v. DeStefano, although primarily a Title VII disparate treatment case, nonetheless contains dicta that sheds some light on the Court’s thinking about Title VII affirmative action. Commentators trying to make sense of the Supreme Court’s confusing decision in the case have debated whether it spells doom for affirmative action or whether, as Professor Charles Sullivan puts it with respect to disparate impact theory, reports of the death of affirmative action as a result of Ricci might be exaggerated. I agree with those scholars who see Ricci as having left the door ajar for affirmative action plans under both constitutional and statutory standards, but for reasons on which other scholars have not focused.
This Article argues that Ricci, while having dealt a blow to disparate impact theory, has not necessarily dealt a fatal blow to affirmative action in the process. Many believe that Ricci has no implications for affirmative action at all since the case’s facts involved no preferences for minorities. However, I believe that dicta in the case suggests how the Court may handle a Title VII affirmative action case in the future, even though I agree that no affirmative action issue was before the Court in Ricci. The key to understanding Ricci and to anticipating the foreseeable future of affirmative action lies in understanding Justice Kennedy’s emerging views, assuming new Justices Sotomayor and Kagan follow relatively liberal paths. Specifically, Justice Kennedy—stepping into the “swing-vote” role formerly held by Justice O’Connor—has adopted key elements of Justice O’Connor’s position on affirmative action: hostile and restrictive, yes, but not entirely opposed to it as are the more conservative members of the Court.
I begin in Part I by looking back at the two Supreme Court Title VII voluntary affirmative action cases: United Steelworkers of America v. Weber and Johnson v. Transportation Agency. I then discuss the legal standard that emerged from those cases, and explore in Part II how that standard might have been affected indirectly by subsequent developments—including case law on affirmative action in the constitutional context, passage of the Civil Rights Act of 1991, Justice O’Connor’s retirement, and the Ricci case. I argue in Part III that the legal standard for Title VII affirmative action has perhaps shifted, and that there are sufficient clues in constitutional case law and in the Ricci case to suggest what the legal standard has become.
I. Private Voluntary Affirmative Action Under Title VII
After the famous 1978 affirmative action decision in Regents of the University of California v. Bakke—involving a public entity and the extent of its ability to craft a quota plan while still avoiding liability under the U.S. Constitution—the Court turned its attention to affirmative action programs implemented by private entities, which are constrained only by Title VII. The Court decided two cases within a decade that established a structural framework for voluntary affirmative action under Title VII. In the first of these, United Steelworkers of America v. Weber, decided the year after Bakke, the employer had established a controversial quota plan reserving half the slots in a specific training program for black workers only. The Gramercy, Louisiana plant for the Kaiser Aluminum & Chemical Corporation was located in an area where black workers made up nearly 40% of the entire workforce. Despite that number, the Gramercy plant had only five skilled craftworkers out of 273 (almost 2%) who were black. As a result, the United Steelworkers, a labor union, negotiated with Kaiser Aluminum to add a quota to ensure that 50% of all new trainees for its in-house training program at the Gramercy plant would be black, since some training was required for skilled craft positions.
The Court upheld the plan. Justice Brennan, writing for the majority, canvassed the language of Title VII, as well as its legislative history, to find that while affirmative action was not mandated by the statute, voluntary plans were permitted within certain bounds. In laying out these bounds, the Court first stated that an employer adopting a voluntary plan must be addressing a traditionally segregated employment opportunity that requires such action. Even then, the plan: (1) must not require the discharge of white workers in order to hire black workers; (2) must not serve as an absolute bar to the advancement of white workers; and (3) must be temporary, in that it can only be used to attain, and not to maintain, racial balance. The Court found that black workers had traditionally been kept out of the apprentice positions that served as a critical prerequisite for skilled craft jobs at the Kaiser Aluminum Gramercy plant. The Court further found that since the quota was 50% for each training program, the plan did not serve as an absolute bar to white workers, nor did it require discharge of those workers. It also found the measure to be temporary, since the plan was expressly going to be terminated upon attainment of its goal—black workers constituting 36% of the skilled craftworker population in the Gramercy plant, reflecting the total percentage of black workers in the relevant labor market.
The prospects for female employees at the Santa Clara Transportation Agency were little better than for black workers at Kaiser Aluminum when the Agency set up its own affirmative action plan for women in skilled craft positions. As is often the case, voluntary affirmative action plans are set up by employers facing potential Title VII liability due to a dearth of women or minorities in particular positions. Though 22% of Agency employees were women, none of the 238 workers in skilled craft positions were women. Women at the Agency occupied positions in which women were traditionally represented, including office, clerical, and paraprofessional jobs. As a result of the skewed demographics of the Santa Clara Agency workforce, the County adopted an affirmative action plan for women, with the goal of eventually getting the workforce to reflect the relevant job-market demographic for women, which was 36%. The plan was explicitly a “goal” plan instead of a quota plan, requiring no particular percentage of female hiring in any given year. In 1980, the Agency hired one woman, whose gender was a factor in her being employed over a male applicant who had ranked a couple of points higher on the oral examination (he had achieved a score of seventy-five to her seventy-three). The Supreme Court’s eventual decision in Johnson v. Transportation Agency upheld the hiring and the affirmative action plan, citing the test established in Weber eight years earlier. In upholding the plan, the Court specifically noted that the target job was in a traditionally segregated job category, that the plan was goal based (even less intrusive on the rights of the majority than was the quota plan in Weber), and that the plan was temporary.
Critically, in both the Johnson and Weber cases, the Supreme Court allowed Title VII voluntary plans to be justified using general labor-force statistics. The plans were upheld not only because there was a dearth of minorities or women in the particular jobs that had been traditionally occupied by majority-class workers, but also because the plans had stopping points, or goals, reflected by the minority population in the overall workforce. The Gramercy locale had 40% black workers and the Santa Clara locale had 36% women in their respective labor markets. The Court allowed the plans to be founded on these very general workforce markers.
II. The Fourteenth Amendment Equal Protection Clause’s Public Voluntary Affirmative Action Standard
The key jurist on affirmative action has been Justice Sandra Day O’Connor. On the Court from 1981 until 2006, she was involved in the Johnson case on the Title VII side and wrote for the majority or plurality in City of Richmond v. J.A. Croson Co., Adarand Constructors, Inc. v. Pena, and Grutter v. Bollinger (as well as the concurrence in Gratz v. Bollinger) on the public sector/Fourteenth Amendment side before stepping down from the Court in 2006. Justice O’Connor has been the driving force or had a hand in six of the eight full United States Supreme Court decisions on constitutional and statutory voluntary affirmative action. Although Justice O’Connor is often cited for her majority opinions, which form the body of the Court’s thinking on affirmative action, her concurring opinion in an early case, Wygant v. Jackson Board of Education, may be the best opinion to analyze to understand her thinking on the subject. In Wygant, Justice O’Connor transparently puzzles through what would ultimately become the foundation of her philosophy on affirmative action. Moreover, the Wygant concurrence has taken on even more meaning now, as Justice Anthony Kennedy—O’Connor’s successor as the key vote on affirmative action—has prominently cited to it in his majority opinion in Ricci. It is this notable reliance that makes Ricci suggestive regarding the future of voluntary affirmative action under Title VII.
Wygant v. Jackson Board of Education was decided in 1986 and was the next Supreme Court case on voluntary affirmative action brought under the Fourteenth Amendment’s Equal Protection Clause after Bakke. The Court in Wygant confronted an affirmative action plan that protected newly hired minority teachers from termination. The Board of Education of Jackson, Michigan—in an attempt to redress rampant racial discrimination in teacher hiring—adopted an affirmative action hiring plan, but realized that any layoffs, especially mass layoffs in response to an economic downturn, would quickly erase any affirmative action gains. In response, the Board, working with the teacher’s union, adopted an additional termination-protection plan. That plan provided as follows:
In the event that it becomes necessary to reduce the number of teachers through layoff from employment by the Board, teachers with the most seniority in the district shall be retained, except that at no time will there be a greater percentage of minority personnel laid off than the current percentage of minority personnel employed at the time of the layoff.
The Wygant case was filed after two separate years of layoffs in which some less senior minority teachers were retained, while some more senior majority teachers were let go. The Court struck down the layoff-protection plan as a violation of the Fourteenth Amendment’s Equal Protection Clause. In doing so, the Court inquired as to whether any compelling state interest justified the plan and examined the means used to accomplish that interest. The Jackson Board of Education justified its layoff protection plan in two ways. First, the Board maintained that minority students needed minority role models in teaching positions and pointed to the fact that the percentage of minority teachers was lower than the percentage of minority students in the school. Second, the Board argued that the city’s history of racial discrimination justified the layoff-protection plan as a remedial measure. With respect to the first argument—the “role model” theory—the Court stated that societal discrimination is not enough to justify a role-model approach and that, in any case, the role-model approach had no link to past discrimination by the school district, nor did it have any logical stopping point. With respect to the second argument—remedying past discrimination—the Court found that the layoff-protection plan had been originally instituted without sufficient evidence documenting actual past discrimination by the Jackson School Board. Any showing of discrimination was made only in the context of the lawsuit, after the challenged plan was implemented.
Regardless of the Board’s interest in creating the layoff-protection plan, the Court stated that the plan would fail under the Fourteenth Amendment in any case because it was “not sufficiently narrowly tailored” to meet that interest. According to the Court, while there are times when race must be taken into account in formulating a remedy, the burden imposed on the majority class by race-based remedies must be kept to a minimum to withstand constitutional strict scrutiny. The Court found that hiring goals impose such a minimal burden, presumably because a person who is denied a job has not yet developed the expectation that comes with having the position. According to the Court, though, layoff protection imposes a harsher injury on the majority class because the loss of an existing job is more intrusive than is the denial of a prospective future opportunity.
Justice O’Connor’s concurrence in Wygant lays bare her developing thinking on affirmative action. First, Justice O’Connor emphasizes that she favors voluntary action by employers, and especially public employers, to remedy past discrimination. She agrees with the plurality, however, that rationales based on remedying general societal discrimination or role-model theories are not sufficient bases on which to anchor voluntary efforts. According to O’Connor:
The imposition of a requirement that public employers make findings that they have engaged in illegal discrimination before they engage in affirmative action programs would severely undermine public employers’ incentive to meet voluntarily their civil rights obligations. This result would clearly be at odds with this Court’s and Congress’ consistent emphasis on “the value of voluntary efforts to further the objectives of the law.” The value of voluntary compliance is doubly important when it is a public employer that acts, both because of the example its voluntary assumption of responsibility sets and because the remediation of governmental discrimination is of unique importance.
In Wygant, O’Connor asks by implication, if societal discrimination is not enough to ground voluntary remedial efforts by the government, what would motivate voluntary action short of requiring the government to make out an entire case of discriminatory liability against itself? Justice O’Connor suggests that requiring too much by way of evidence would serve as a strong disincentive to voluntary action. Accordingly, she notes, a required finding of prior antecedent or contemporaneous discrimination is too much. Nonetheless, public employers must have a “sufficient basis” for imposing affirmative action measures.
Justice O’Connor then finds this “sufficient basis” in the statistical analysis approved in Hazelwood School District v. United States, a case involving systemic, pattern, or practice discrimination under Title VII. As she explains, a statistical comparison of the percentage of minority employees in target jobs to the percentage of minorities in the relevant labor market is sufficient to establish a prima facie case of systemic discrimination and should likewise provide a “firm basis” for remedial affirmative action. Justice O’Connor notes that imposing such a strong requirement would neither make the employer automatically liable nor make the affirmative action plan unassailable. Indeed, the statistical finding illustrated in Hazelwood creates a prima facie case of discrimination, but named plaintiffs and anecdotal evidence of discrimination would have to accompany the statistics to actually prove systemic discrimination resulting in liability, thus allowing an employer to make a statistical case without finding actual liability against itself. O’Connor concludes her concurrence by applying her construct to the facts at hand in Wygant. She explains that the statistical comparison of minority teachers to minority students is irrelevant to the issue of employment discrimination: “[I]t is only when it is established that the availability of minorities in the relevant labor pool substantially exceeded those hired that one may draw an inference of deliberate discrimination in employment.”
Justice O’Connor’s thinking on affirmative action, revealed in Wygant, became cemented a few years later in City of Richmond v. J.A. Croson Co. In Croson, the City of Richmond had created an affirmative action plan for hiring minority contractors. As was the case with the City of Jackson in Wygant, Richmond also had its substantial share of racial strife and past discrimination. The City Council decided that since the population of the city was 50% African American, it was only logical that a substantial number of its contractors should be drawn from the ranks of minority-owned enterprises. The City thus established a substantial goal for prime contractors to award 30%, based on total dollar amounts, of their city-project subcontracts to minority-owned enterprises. Justice O’Connor, writing for the plurality, emphasized that to lawfully establish an affirmative action plan, the City had to show actual discrimination with “some specificity before [it could] use race-conscious relief.” According to O’Connor, the comparison to the city population was irrelevant to the issue of proving discrimination in contracting (this was similar to her reaction to the comparison between teacher and student populations advanced in Wygant). Instead, the City should have identified the disparity between two figures—the percentage of dollar amounts awarded to minority contractors by Richmond and the percentage of qualified minority contractors in the relevant market. When these statistics are used, the percentages change markedly. Richmond awarded only 0.67% of its prime contracts to minority firms during the relevant time period. However, evidence indicated that the percentage of qualified minority contractor firms in the national market at the time was only 4.7%, and that a large percentage of those firms were concentrated in just five other states. Even if a standard deviation of greater than two or three were produced statistically in the Croson case (assuming Richmond was representative of the national market), the maximum goal of any affirmative action plan would have to be the market percentage—about 5%.
The bottom line for voluntary affirmative action plans, subject to constitutional scrutiny under the Fourteenth Amendment after Croson, is that they can only be adopted after the relevant governmental unit produces a “firm” or “strong” basis in evidence that actual discrimination has occurred. The firm or strong basis refers to the amount of evidence sufficient to make out a prima facie case of systemic discrimination under Title VII, consistent with the Court’s decision in Hazelwood. This much is made plain by a close analysis of Justice O’Connor’s concurring opinion in Wygant, followed by her plurality opinion in Croson. Importantly, Justice Kennedy, the new swing vote on affirmative action and the author of the majority opinion in Ricci, joined Justice O’Connor in Croson.
III. Ricci v. DeStefano: Predicting a Shift in the Standard for Private Voluntary Affirmative Action Under Title VII
A. Relevant Facts of Ricci
The facts of the Ricci case have now been rehashed dozens of times in scholarly articles. The critical facts for purposes of this Article are the following. The City of New Haven developed and administered officer-promotion exams for lieutenant and captain positions within its fire department. These examinations were developed over a period of time with the involvement of experts. The exams included a written component, worth 60% of the final exam score, and an oral component, worth 40% of the final score. This balance was struck in the collective bargaining agreement between the firefighter union and the City of New Haven. In addition, a City rule required that each promotion went to someone with one of the top three scores on a given exam. For the lieutenant exam, seventy-seven applicants took the exam (forty-three Caucasian, nineteen African American, and fifteen Hispanic), thirty-four of whom passed the exam (twenty-five Caucasian (60%), six African American (30%), and three Hispanic (20%)). For the captain exam, there were forty-one applicants (twenty-five Caucasian, eight African American, and eight Hispanic), twenty-two of whom passed the exam (sixteen Caucasian (65%), three African American (40%), and three Hispanic (40%)). Unfortunately, the tests produced a disparate impact against minority takers under the very rough, traditional disparate impact test known as the “80% rule.” Under the 80% statistical rule, an impact on minorities from a test is disparate for purposes of making out a prima facie case under Title VII if the pass rate for minority-class takers is less than 80% of the pass rate for majority-class takers.
In Ricci, the 80% rule was met. On the lieutenant exam, the pass rate for African-American takers was only about 55% of the Caucasian pass rate (32% vs. 58%), and the pass rate for Latino takers was even lower, at about 34% (20% vs. 58%). On the captain exam, there were similar results. The pass rate for both African Americans and Latinos was 38%, about 60% of the Caucasian pass rate, and also within the 80% requirement. Because this created prima facie disparate impact liability, the City of New Haven chose to discard the test results. In the end, the Supreme Court narrowly held that the City of New Haven should not have discarded the tests, because of the race-based disadvantage or disparate treatment caused to the white firefighters who had taken and passed the tests. The Court found that Title VII would not support a disparate treatment violation (here, discrimination against whites caused by nullifying the test results) in order to address only a prima facie case of disparate impact liability (the City of New Haven had uncovered a prima facie case of disparate impact liability based on the 80% rule, but had not uncovered strong evidence of an actual case of disparate impact liability). The Court stated, however, that test results could be discarded if the City of New Haven had uncovered a “strong basis in evidence” of an actual case of disparate impact liability, but held that no such showing of disparate impact liability existed in the New Haven scenario; while there was an impact, the City had not gone further to analyze whether the tests also failed the “business necessity/job-relatedness” and “alternative means” prongs of the full disparate impact analysis in order to have a “strong basis” for finding liability.
B. Ricci’s Holding Limited to Tests Already Taken
The Court’s decision in Ricci must be interpreted narrowly for the full decision (holding and dicta) to make sense. The square holding of the case is that when a test has a disparate impact, the employer must be able to show a “strong basis in evidence” for Title VII disparate impact liability from the lower-passing group before that employer may discard the test results; only by following this proof structure may the employer avoid potential disparate treatment liability to the higher-passing group. As related to tests already taken, a “strong basis in evidence” means an employer finding of potential disparate impact liability, as opposed to a mere prima facie case. The public employer must show not only that the tests had a disparate impact on the basis of race, but also that there was no adequate business necessity for the tests, or if there was an adequate business necessity, that no alternative measure or test that accomplished the employer’s business goals, but with less impact, was available.
There are solid indications in the Ricci decision that the holding is limited to tests that have already been administered. First, the Court explicitly states that the violation occurred in “discarding the test results,” and not in the efforts of the City to create a fair test. Second, despite its overall holding in Ricci, the Court takes great pains to explain, in dicta, that voluntary actions to remedy discrimination (short of discarding tests, apparently) are important and would be chilled if a public employer had to find actual disparate impact liability against itself before attempting to remedy racial discrimination. For example, in answering petitioner arguments urging that compliance cannot ever be a defense unless actual disparate impact liability is shown first, the Court states as follows:
Again, this is overly simplistic and too restrictive of Title VII’s purpose. The rule petitioners offer would run counter to what we have recognized as Congress’s intent that “voluntary compliance” be “the preferred means of achieving the objectives of Title VII.” Forbidding employers to act unless they know, with certainty, that a practice violates the disparate-impact provision would bring compliance efforts to a near standstill. Even in the limited situations when this restricted standard could be met, employers likely would hesitate before taking voluntary action for fear of later being proven wrong in the course of litigation and then held to account for disparate treatment.
Obviously, the foregoing statement by the Court would make no sense if the latter part of the decision—which required the City of New Haven to have strong evidence of potential disparate impact liability before discarding tests—applied to all attempts by an employer to eradicate past racial discrimination, including through affirmative action programs. Indeed, the Court here cites to Justice O’Connor’s concurring opinion in Wygant, in which she suggests that the City of Jackson, Michigan might have escaped liability for its affirmative action plan (limiting minority-teacher terminations) if it had based its remedial efforts on a statistical showing sufficient to make out a prima facie case of systemic discrimination liability, rather than just on an impact case. As shown prior, Justice Kennedy subscribes to Justice O’Connor’s thinking on this issue. Third, the Court explains its decision about the discarded tests by invoking the reliance interest of the test takers (mainly related to the effort involved in studying for a particular test)—an interest that would not exist if the remedial actions were taken at the design stage, pre-administration.
C. The Future of Title VII Affirmative Action
An interesting question after Ricci is what may an employer voluntarily do under Title VII if it faces (as did the City of New Haven) a vastly segregated workforce and desires to address the discrimination? What does Ricci’s “strong basis in evidence” rule mean, if anything, in the affirmative action context? We know, after Ricci, that the employer cannot discard test results that have an adverse impact against a protected class unless the employer can show it faces potential disparate impact liability if it were to keep the results. We also know that an employer can take any action to ensure that its testing or selection criteria are fair. But what if an employer chooses to address the segregation by implementing affirmative action remedies—say a goal plan for hiring or promotion, like the one used in Johnson? The relevant case law for Title VII affirmative action is Johnson and Weber—discussed earlier in this Article—which allow such plans based on general labor force statistics, rather than on a statistical test that analyzes the makeup of the workplace in relation to the makeup of the qualified labor pool. However, those cases are now dated and likely do not reflect the current thinking of the Court in these matters.
If the Ricci holding is limited to discarding tests already taken, are there any clues in the case’s dicta about how the Court perceives Title VII affirmative action programs—voluntary remedial preferences to address workplace racial disparities? It turns out there are. For example, it seems clear that a majority of the Court still favors voluntary remedial action, including affirmative action by employers, and thinks that such action is consistent with Title VII and the Equal Protection Clause of the Fourteenth Amendment. The Ricci Court cites twice to Croson and once to Justice O’Connor’s concurring opinion in Wygant to underscore the importance of voluntary remedial actions, and affirmative action, as a part of Title VII compliance. Of course, the citation to Justice O’Connor’s concurring opinion in Wygant, I argue, is also an intentional nod to her view that it is particularly important for the government to take the lead in voluntary compliance efforts, given the especially pernicious history of the government’s role in racial discrimination. Wygant is an affirmative action case, even though Ricci is not. The Court cites to these cases also to emphasize the “strong basis in evidence” idea as having come from Croson and O’Connor’s concurring opinion in Wygant. The Ricci decision thus strongly suggests the introduction of a new legal standard for Title VII affirmative action, forged in the context of the already-existing standard for affirmative action under the Equal Protection Clause of the U.S. Constitution. Instead of citing to Johnson or Weber regarding the proper Title VII analysis for voluntary remedial action, the Court cites constitutional affirmative action precedent. As the Court states:
In searching for a standard that strikes a more appropriate balance, we note that this Court has considered cases similar to this one, albeit in the context of the Equal Protection Clause of the Fourteenth Amendment. The Court has held that certain government actions to remedy past racial discrimination—actions that are themselves based on race—are constitutional only where there is a “‘strong basis in evidence’” that the remedial actions were necessary. This suit does not call on us to consider whether the statutory constraints under Title VII must be parallel in all respects to those under the Constitution. That does not mean the constitutional authorities are irrelevant, however. Our cases discussing constitutional principles can provide helpful guidance in this statutory context.
The Court appears to be signaling a shift in its standard for affirmative action under Title VII. The Ricci case announces a “strong basis in evidence” standard and explains that, in the context of tests already taken, a city must have evidence of imminent disparate impact liability before it can discard such tests. However, the Court goes on to state that applying such a standard to all voluntary remedial actions would chill these efforts, and indicates that for these other efforts, a “strong basis in evidence” is consistent with standards already developed under the Equal Protection Clause. The Court expressly states that these cases can serve as useful guidance under Title VII.
Those citations to Croson, and perhaps especially to Justice O’Connor’s concurring opinion in Wygant, invoke O’Connor’s idea of a prima facie statistical case of systemic discrimination (as opposed to impact) as being strong enough to ground voluntary remedial efforts. Therefore, “strong basis in evidence” means—consistent with Justice O’Connor’s concurring opinion in Wygant and majority opinion in Croson—a statistical showing disciplined by a technical analysis (minimally a standard-deviation test and maximally a multiple-linear-regression analysis) in affirmative action or voluntary remediation cases in which a test has not yet been given. Croson and Wygant are not testing cases.
After Ricci, I believe that the City of New Haven could take more aggressive affirmative action measures (instead of, or simultaneously with, changing its testing criteria for future promotions). For example, the City of New Haven could use a statistical test to compare the percentage of minority firefighter officers to the percentage of minority firefighters (or even the hiring pool from which New Haven firefighters are drawn). If the result is greater than two or three standard deviations, the City can take affirmative action to fix the problem. The City could institute a goal plan to reach a percentage of minority firefighters consistent with the percentage in the appropriate labor pool. In addition, the City could provide free study materials to minorities, additional training for minority officer candidates, and could act to step up minority recruitment. In short, the City could implement any number of preferences for minorities in order to address the systemic discrimination that apparently exists in the fire department, and that would be revealed by the application of more searching statistical methods.
* Professor, University of Denver Sturm College of Law. The author thanks Professor Wendy Parker and the Wake Forest Law Review for an informative, well-run, and impressively well-attended Symposium. The author thanks Charles Sullivan, Steve Willborn, Alan Chen, Michael Selmi, Justin Driver, Kimberly West-Faulcon, David Schwartz, Randy Wagner, and the Colorado Employment and Labor Law Faculty (Melissa Hart, Martin Katz, Scott Moss, Helen Norton, Nantiya Ruan, and Catherine Smith) for their comments on this Article. All errors are the author’s.
. Although it is perhaps a question of some debate, for purposes of this Article, “affirmative action” involves only voluntary efforts by an employer to remedy past discrimination in a race-conscious way by adopting goals, or possibly even quotas, or by creating preferences on the basis of race or gender. “Affirmative action” does not encompass employer attempts to ensure that selection criteria apply to all persons equally and that such criteria do not discriminate against minorities. See Helen Norton, The Supreme Court’s Post-Racial Turn Towards a Zero-Sum Understanding of Equality, 52 Wm. & Mary L. Rev. 197, 244–46 (2010) (explaining the legal distinctions between these two types of programs). But see George Rutherglen, Ricci v DeStefano: Affirmative Action and the Lessons of Adversity, 2009 Sup. Ct. Rev. 83, 110–11 (examining different approaches to disparate impact theory and concluding that certain “forms of race-conscious action,” such as mandatory affirmative action plans and readjustment of test scores, “are too coercive, and perhaps too clear, to fit the long-standing consensus on affirmative action”).
. 129 S. Ct. 2658 (2009).
. See generally Charles A. Sullivan, Ricci v. DeStefano: End of the Line or Just Another Turn on the Disparate Impact Road?, 104 Nw. U. L. Rev. 411 (2010).
. See, e.g., Rutherglen, supra note 1, at 83, 94–95. These scholars cite to Justice Ginsburg’s dissent in the case, in which she wrote that “New Haven’s action, which gave no individual a preference, ‘was simply not analogous to a quota system or a minority set-aside where candidates, on the basis of their race, are not treated uniformly.’” Ricci, 129 S. Ct. at 2696 (Ginsburg, J., dissenting) (quoting Ricci v. DeStefano, 554 F. Supp. 2d 142, 157 (D. Conn. 2006) (internal quotation marks omitted)).
. 443 U.S. 193 (1979).
. 480 U.S. 616 (1987).
. Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 269–71 (1978).
. 443 U.S. 193 (1979). For important background information and context relating to the Weber case, especially regarding the United Steelworkers’ involvement in affirmative action efforts in government contracting, see generally Deborah Malamud, The Story of United Steelworkers of America v. Weber, in Employment Discrimination Stories 173 (Joel Wm. Friedman ed., 2006).
. Weber, 443 U.S. at 197–98.
. See id. at 208 (noting that both Title VII and the challenged affirmative action plan “were structured to ‘open employment opportunities for Negroes in occupations which have been traditionally closed to them’” (quoting 110 Cong. Rec. 6548 (1964) (statement of Sen. Hubert Humphrey))).
. Id. at 198–99, 222–23.
. Johnson v. Transp. Agency, Santa Clara Cnty., Cal., 480 U.S. 616, 621 (1987).
. See id. at 653 (O’Connor, J., concurring).
. Id. at 621 (majority opinion).
. Id. at 635; United Steelworkers of Am. v. Weber, 443 U.S. 193, 208–09 (1979).
. Johnson, 480 U.S. at 635–36; Weber, 443 U.S. at 208–09.
. There has been some suggestion that the Civil Rights Act of 1991 changed and hardened the standard for affirmative action, but that does not seem to be the case. Section 116 of the 1991 Act expressly states that the amendments have no impact on affirmative action. Civil Rights Act of 1991, Pub. L. No. 102-166, § 116, 105 Stat. 1071, 1079. Despite a memorandum in the Act’s legislative history by then-Senator Robert Dole suggesting that codification of the mixed-motive standard created a hurdle for race-conscious action of any kind, 137 Cong. Rec. S15,477 (daily ed. Oct. 30, 1991) (memorandum of Sen. Robert Dole), federal courts have rejected this view (most likely because there was not enough guidance from Congress on the issue). See, e.g., Gilligan v. Dep’t of Labor, 81 F.3d 835, 840 (9th Cir. 1996); Officers for Justice v. Civil Serv. Comm’n, 979 F.2d 721, 725 (9th Cir. 1992); Hannon v. Chater, 887 F. Supp. 1303, 1316–18 (N.D. Cal. 1995). Section 106 of the Act prohibits race-norming tests (altering or modifying test results on the basis of race), but that provision has been narrowly applied and likely only prohibits what the Supreme Court already condemned in Connecticut v. Teal, 457 U.S. 440 (1982). Civil Rights Act of 1991 § 106; 42 U.S.C. § 2000e-2(l) (2006). See Chi. Firefighters Local 2 v. City of Chicago, 249 F.3d 649, 656 (7th Cir. 2001) (holding that section 106 does not apply to test “banding,” or treating all scores within a certain range the same way); Hayden v. Cnty. of Nassau, 180 F.3d 42, 53 (2d Cir. 1999) (holding that section 106 does not apply to an employer’s attempts to create a test with the slightest possible adverse impact on racial minorities); Alfred W. Blumrosen, Society in Transition IV: Affirmation of Affirmative Action Under the Civil Rights Act of 1991, 45 Rutgers L. Rev. 903, 908–09, 913 (1993); Nelson Lund, The Law of Affirmative Action in and After the Civil Rights Act of 1991: Congress Invites Judicial Reform, 6 Geo. Mason L. Rev. 87, 89–91 (1997).
. See Girardeau A. Spann, The Dark Side of Grutter, 21 Const. Comment. 221, 226–27 (2004).
. 488 U.S. 469, 476 (1989).
. 515 U.S. 200, 204 (1995).
. 539 U.S. 306, 311 (2003).
. 539 U.S. 244, 276 (2003) (O’Connor, J., concurring).
. See Goodwin Liu, The Bush Administration and Civil Rights: Lessons Learned, 4 Duke J. Const. L. & Pub. Pol’y 77, 97 (2009) (discussing Justice O’Connor’s role in affirmative action cases).
. 476 U.S. 267, 284 (1986) (O’Connor, J., concurring in part and concurring in the judgment).
. See Norton, supra note 1, at 248; Ilya Shapiro, A Faint-Hearted Libertarian at Best: The Sweet Mystery of Justice Anthony Kennedy, 33 Harv. J.L. & Pub. Pol’y 333, 348 (2010) (book review) (“[A]t the very least it is safe to say that, for the foreseeable future, the outcome of race cases will all depend upon Justice Kennedy.”).
. See Wygant, 476 U.S. at 273.
. Id. at 298 (Marshall, J., dissenting).
. Id. at 270 (plurality opinion).
. Id. at 290 (O’Connor, J., concurring in part and concurring in the judgment).
. Id. at 290 (citations omitted) (quoting Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 364 (1978) (Brennan, J., concurring in the judgment in part and dissenting in part)) (citing S. Rep. No. 92-415, at 10 (1971) (accompanying the amendments extending coverage of Title VII to the States) (“Discrimination by government . . . serves a doubly destructive purpose. The exclusion of minorities from effective participation in the bureaucracy not only promotes ignorance of minority problems in that particular community, but also creates mistrust, alienation, and all too often hostility toward the entire process of government.”)).
. Id. at 291 (“As is illustrated by this case, public employers are trapped between the competing hazards of liability to minorities if affirmative action is not taken to remedy apparent employment discrimination and liability to nonminorities if affirmative action is taken. Where these employers, who are presumably fully aware both of their duty under federal law to respect the rights of all their employees and of their potential liability for failing to do so, act on the basis of information which gives them a sufficient basis for concluding that remedial action is necessary, a contemporaneous findings requirement should not be necessary.”).
. Id. at 294 (citing Hazelwood Sch. Dist. v. United States, 433 U.S. 299, 308 (1977)).
. Hazelwood, 433 U.S. at 301. The U.S. Supreme Court’s approved standard for proving discrimination through statistics is a standard-deviation (binomial-distribution) analysis, although a more sophisticated linear-regression analysis is also acceptable. See Bazemore v. Friday, 478 U.S. 385, 399–400 (1986) (Brennan, J., concurring in part) (explaining the role of regression analysis); Castaneda v. Partida, 430 U.S. 482, 496 & n.17 (1977) (explaining and applying binomial distribution and standard deviation); Hazelwood, 433 U.S. at 311 & n.17 (same); Dianne Avery et al., Employment Discrimination Law 205–10 (8th ed. 2010) (explaining the various statistical measures used by litigants in proving employment discrimination claims). Lesser statistical measures than these will not suffice statistically to prove discrimination since they do not meet the statistical significance rule, which requires a showing that any disparity is not due to mere chance. Id. at 205. For more on Hazelwood, see generally Stewart J. Schwab & Steven L. Willborn, The Story of Hazelwood: Employment Discrimination by the Numbers, in Employment Discrimination Stories, supra note 8, at 37, 37–63.
. Wygant, 476 U.S. at 292 (“[I]n order to provide some measure of protection to the interests of its nonminority employees and the employer itself in the event that its affirmative action plan is challenged, the public employer must have a firm basis for determining that affirmative action is warranted. Public employers are not without reliable benchmarks in making this determination. For example, demonstrable evidence of a disparity between the percentage of qualified blacks on a school’s teaching staff and the percentage of qualified minorities in the relevant labor pool sufficient to support a prima facie Title VII pattern or practice claim by minority teachers would lend a compelling basis for a competent authority such as the School Board to conclude that implementation of a voluntary affirmative action plan is appropriate to remedy apparent prior employment discrimination.” (emphasis added)).
. Id. (“If a voluntary affirmative action plan is subsequently challenged in court by nonminority employees, those employees must be given the opportunity to prove that the plan does not meet the constitutional standard this Court has articulated. However, as the plurality suggests, the institution of such a challenge does not automatically impose upon the public employer the burden of convincing the court of its liability for prior unlawful discrimination; nor does it mean that the court must make an actual finding of prior discrimination based on the employer’s proof before the employer’s affirmative action plan will be upheld. In ‘reverse discrimination’ suits, as in any other suit, it is the plaintiffs who must bear the burden of demonstrating that their rights have been violated.” (citation omitted)).
. See Hazelwood, 433 U.S. at 311 & n.17, 312. Though the Court has indicated in dicta that statistics alone may be enough to prove actionable discrimination, this statement may be limited to egregious cases in which minority hiring is nonexistent. See, e.g., Int’l Bhd. of Teamsters v. United States, 431 U.S. 324, 342 n.23 (1977) (“[T]he company’s inability to rebut the inference of discrimination came not from a misuse of statistics but from ‘the inexorable zero.’”); see also 2 Barbara T. Lindemann & Paul Grossman, Equal Emp’t Opportunity Comm., Am. Bar Ass’n, Employment Discrimination Law 2306 (C. Geoffrey Weirich ed., 4th ed. 2007) (“Courts recognize that evidence of the ‘inexorable zero’—a failure to hire any members of a protected class—by itself may support an inference of intentional discrimination under the disparate treatment theory.” (footnotes omitted)).
. Wygant, 476 U.S. at 294.
. Id. (citing Hazelwood, 433 U.S. at 308).
. 488 U.S. 469 (1989).
. Id. at 504. Justice Scalia has characterized the standard as requiring a “strong basis in evidence.” See Concrete Works of Colo., Inc. v. City & Cnty. of Denver, Colo., 540 U.S. 1027, 1029 (2003) (Scalia, J., dissenting from denial of petition for writ of certiorari) (quoting Shaw v. Hunt, 517 U.S. 899, 910 (1996)). Justice O’Connor characterizes the standard as a requirement for a “firm” basis in evidence in Wygant, but then refers to a test of statistical significance as requiring a “strong” basis in evidence in Croson. Compare Wygant, 476 U.S. at 286 (O’Connor, J., concurring), with Croson, 488 U.S. at 510 (plurality opinion) (quoting Wygant, 476 U.S. at 277 (plurality opinion)). Justice Kennedy refers to the same basic proof requirement in Ricci as a “strong basis in evidence.” Ricci v. DeStefano, 129 S. Ct. 2658, 2664 (2009). In the end, it doesn’t matter that much whether the requirement is characterized as “firm” or “strong,” so long as it is understood to be the same standard.
. Croson, 488 U.S. at 503. To Justice O’Connor, only this comparison could give rise to the proper inference of discrimination. See id. This is essentially the standard that Justice O’Connor outlines in Wygant, as evidenced by her concurring opinion in that case.
. See id. at 500 (quoting Wygant, 476 U.S. at 277). For a discussion of the “firm” versus “strong” language, see supra note 69.
. See id. at 476. Particularly noteworthy is the fact that Justice Kennedy joins Part V of Justice O’Connor’s plurality opinion in Croson, in which Justice O’Connor details how the City of Richmond might have proceeded on its own to rectify discrimination using the proper labor-pool analysis. See id. at 509–11.
. See, e.g., Norton, supra note 1, at 216–18; Rutherglen, supra note 1, at 83–91; Sullivan, supra note 3, at 414–15, 418–19.
. Ricci v. DeStefano, 129 S. Ct. 2658, 2665 (2009).
. There are two types of disparate impact cases: those cases based on a single selection criterion, see, e.g., Griggs v. Duke Power Co., 401 U.S. 424, 433–34, 436 (1971) (possession of a high school diploma or passage of intelligence test), and those cases based on multiple selection criteria, typically involving subjective elements, like interviews, see, e.g., Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 982, 989–91 (1988) (supervisory interviews and recommendations). In a single selection criterion case, a specified selection criterion is used by the employer to make hiring or promotion decisions. The single criterion could be a written or oral examination (as in Ricci), a typing test, or even a height and weight requirement. When test results are examined, differing pass rates of race groups can be analyzed. If there is a great disparity, which for policy purposes exists with a rate for one group of takers that is less than 80% of the pass rate for the highest-passing group, the test is viewed as defective and will have to be dropped unless the employer can show that the criterion is job related and consistent with business necessity. Those questioning the test can still prevail even if the employer shows business necessity if they can produce a better test, presumably one roughly achieving the employer’s business goals, but with less of a disparate impact. 1 Lindemann & Grossman, supra note 62, at 122–24, 128–32.
In the multiple selection criteria case, there are a variety of elements that go into selection of employees. Typically included among these are subjective elements, like interview scores. In these cases, it may be impossible to point to a single component that produces a disparate impact in hiring. When this happens, employers turn to a basic statistical test (sometimes called a standard-deviation test, a binomial-distribution test, or simply a “z” test), or an even more sophisticated statistical test called a multiple-linear-regression analysis, to examine the importance of different factors involved in hiring, including, most significantly, the chances that the disparity in hiring can be produced by mere chance or happenstance. In the case of a standard deviation test, a result greater than three standard deviations from the norm means the probability that the result was produced by mere chance is 1% or less. See id. at 122–32.
The standard-deviation and linear-regression tests are tests of statistical significance. These tests are more accurate at showing that a test is problematic than is the 80% rule, which is more of a rough guide that has the virtue of being easy to apply. See id. at 128–32. In Ricci, the City of New Haven did not analyze the test results using a test of statistical significance. The City chose to make its decision about the tests solely on the basis of the 80% rule. See Ricci, 129 S. Ct. at 2677–78.
. Ricci, 129 S. Ct. at 2678.
. Id. at 2666, 2678. The higher the pass-rate differential, the better the case for a disparate impact claim.
. As explained supra note 83, the 80% rule is the crudest of measures of disparate impact. A better statistical test would be a regression analysis, and it is possible that the crudeness of the measure may have subconsciously played a part in the Court’s decision making. Scrutinized closely based on the facts of the case, the holding in Ricci is actually, literally, that a prima facie showing of disparate impact based solely on the 80% rule is not a strong enough basis in evidence to set aside tests already taken.
. Ricci, 129 S. Ct. at 2677–78, 2681.
. Id. at 2678 (“Based on the degree of adverse impact reflected in the results, respondents were compelled to take a hard look at the examinations to determine whether certifying the results would have had an impermissible disparate impact. The problem for respondents is that a prima facie case of disparate-impact liability—essentially, a threshold showing of a significant statistical disparity, and nothing more—is far from a strong basis in evidence that the City would have been liable under Title VII had it certified the results. That is because the City could be liable for disparate-impact discrimination only if the examinations were not job related and consistent with business necessity, or if there existed an equally valid, less-discriminatory alternative that served the City’s needs but that the City refused to adopt. We conclude there is no strong basis in evidence to establish that the test was deficient in either of these respects.” (emphasis added) (citation omitted)). Rather than remand the case at this point for further findings, though, the Court itself proceeds to engage in the more detailed analysis. In so doing, the Court strangely applied mere rationality as the burden for the City to satisfy in these other parts of the disparate impact cause of action. See id. at 2678–81; Melissa Hart, Procedural Extremism: The Supreme Court’s 2008–2009 Labor and Employment Cases, 13 Emp. Rts. & Emp. Pol’y J. 253, 262–63 (2009); Norton, supra note 1, at 218–28; Sullivan, supra note 3, at 422–25. In the end, and despite the Supreme Court’s analysis, there really is no reason to believe that these tests identified the best supervisors for the firefighting position, a point that may well have been fleshed out more carefully on remand.
. Ricci, 129 S. Ct. at 2664–65, 2676.
. Id. at 2673 (citing 42 U.S.C. § 2000e-2(k)(1)(A) (2006)).
. See Hart, supra note 89, at 263 (“The majority [in Ricci] drew a line between: 1) voluntary compliance efforts that seek to avoid disparate impact in the creation and administration of employment tests; and 2) practices and the evaluation of test scores after the tests have been taken. The former are not subject to the Court’s new approach. Only after a test has been taken—when the actual racial make-up of the results is known—will an employer be at risk of disparate treatment liability.”). Professor Hart cites to Professor Sullivan for this proposition. Id. at 263 n.58; see Sullivan, supra note 3, at 417 (interpreting the holding of Ricci “to mean that the employer could have adopted its testing . . . to minimize its disparate impact, even though it could not invalidate a test, once it was given, for that reason”); see also Norton, supra note 1, at 237–39 (suggesting that a narrow holding in Ricci that is limited to tests already taken is a distinct possibility).
. Ricci, 129 S. Ct. at 2664 (“As a result, the City’s action in discarding the tests was a violation of Title VII.”).
. Id. at 2674; see also Hart, supra note 89, at 263; Norton, supra note 1, at 235–39; Sullivan, supra note 3, at 417–18.
. See Ricci, 129 S. Ct. at 2674–76.
. Id. at 2674 (citations omitted).
. Nor do I believe that a citation here to a concurring opinion is simply a case of Justice Kennedy or his clerks inserting just any supporting citation. If that were the case, why cite to a concurring opinion? The Croson cite alone would certainly suffice as support. I believe Justice Kennedy signals here that he agrees with Justice O’Connor’s vision and philosophy of affirmative action, as applied in Part V of the Croson decision certainly, but also as conceived and explained in Justice O’Connor’s concurring opinion in Wygant.
. Ricci, 129 S. Ct. at 2681 (“The injury arises in part from the high, and justified, expectations of the candidates who had participated in the testing process on the terms the City had established for the promotional process.”); see also Sullivan, supra note 3, at 418 (“The majority in Ricci repeatedly referred to the white firefighters’ expectations of, and reliance on, the use of the test as a promotion method, neither of which would exist if the employer’s disparate impact calculations occurred early in the process.”); cf. Richard Primus, The Future of Disparate Impact, 108 Mich. L. Rev. 1341, 1345 (2010) (characterizing the Ricci Court’s narrow and rigid holding on the testing issue as explainable because of the “visible victims” involved). This is consistent, too, with the Court’s hesitancy in affirmative action cases to allow remedies that deprive others of actual jobs, as in Wygant, in which the Court found that depriving somebody of an existing job based on affirmative action violates least-restrictive-means analysis. Wygant v. Jackson Bd. of Educ., 476 U.S. 267, 279–80 (1986). Of course, the deprivation in Ricci was a promotion, not loss of a job, but the Court’s statement about candidates’ reliance interest in the test process shows the Court more willing to put this on the “loss” rather than the “fail to get” side of the ledger. The question, of course, has yet to be decided in an affirmative action context.
. See Norton, supra note 1, at 245–46; Sullivan, supra note 3, at 417–18.
. See supra notes 19–27 and accompanying text.
. See supra notes 8–30 and accompanying text.
. Weber and Johnson were decided in 1979 and 1987, thirty-two and twenty-four years ago, respectively. In that time, both the Court and its general thinking about affirmative action, revealed in its constitutional decisions, have changed considerably. While federal courts have consistently applied these precedents in the Title VII context, many have distinguished Weber and Johnson on the basis that in those cases virtually no blacks or women had been hired into target jobs, lessening the requirement of a statistical showing of apparent discrimination to justify affirmative action. See Cynthia L. Estlund, Putting Grutter to Work: Diversity, Integration, and Affirmative Action in the Workplace, 26 Berkeley J. Emp. & Lab. L. 1, 12 (2005).
. See supra note 29 and accompanying text.
. See Norton, supra note 1, at 246–48.
. In this argument, I part company with Linda L. Arakawa and Michele Park Sonen, and the position they take in Note, Caught in the Backdraft: The Implications of Ricci v. DeStefano on Voluntary Compliance and Title VII, 32 U. Haw. L. Rev. 463 (2010). They argue that there is apparently no room for use of statistics to engage in voluntary remedial efforts after the introduction of Ricci’s “strong basis in evidence” standard. Id. at 464–65. They also argue that the use of statistical tools to ground affirmative action is consistent with past Supreme Court precedent, as authored by Justice O’Connor. Id. at 481–82. I agree with them about the latter argument, but not the former. I believe, as I argue in this Part, that the Ricci Court in dicta upholds and supports those prior constitutional opinions on the scope of employer voluntary remedial efforts. Arakawa and Park Sonen argue additionally that the “strong basis in evidence” standard should be rejected in favor of a standard that would allow a prima facie case of disparate impact liability using sophisticated statistical measures to ground voluntary efforts or affirmative action. Id. at 482–83. While I believe they are on the right track, I argue, instead, that the Court leaves the door open for voluntary efforts based on statistical measurements that establish a prima facie case of systemic discrimination, as explained by Justice O’Connor in her concurring opinion in Wygant. Arakawa and Park Sonen attempt to reconcile Ricci and their own proposed standard with Weber and Johnson. I argue that the Court is signaling a shift that would allow voluntary affirmative action, but to a greater degree under the more rigid constitutional standard, squarely inconsistent with the prior analysis established under Weber and Johnson.
. Ricci v. DeStefano, 129 S. Ct. 2658, 2675 (2009) (citations omitted).