Sandra Day O'Connor
Sandra Day O'Connor; drawing by David Levine


Is affirmative action unconstitutional? Does it violate the Fourteenth Amendment’s guarantee of “equal protection of the laws” for universities to give preference to blacks and other minorities in the fierce competition for student places, as the best of our universities have done for thirty years? In 1978, the late Justice Lewis Powell, in his opinion in the Supreme Court’s famous Bakke decision, ruled that racial preferences are permissible if their purpose is to improve racial diversity among students, and if they do not stipulate fixed minority quotas, but take race into account as one factor among many. Since four other justices in that case would have upheld even a quota system, five of the nine agreed that plans meeting Powell’s tests were constitutional.

Many lawyers fear that the Supreme Court will soon reconsider its Bakke ruling, however, and declare that any racial preference in an admissions process is, after all, unconstitutional. In 1996, the Fifth Circuit Court of Appeals, in the Hopwood case, struck down the Texas Law School’s affirmative action plan, and two of the three judges in the panel declared that recent Supreme Court decisions about affirmative action policies in areas other than education have already in effect overruled Bakke, so that all university affirmative action is now unconstitutional.1

The Fifth Circuit’s decision had immediate and, in the view of the Texas Law School’s faculty, disastrous results: that school had admitted thirty-one black students in 1996, but it enrolled only four in 1997. The Supreme Court refused to review the decision, but the Center for Individual Rights, a Washington, D.C.-based organization that had facilitated the Hopwood litigation, has already filed a new lawsuit in Michigan challenging the University of Michigan’s affirmative action plan, and other suits can be expected in other jurisdictions. The Supreme Court will have to rule on the matter soon.

It will be not only ironic but sad if the Court reverses its own longstanding ruling now, because dramatic evidence of the value of affirmative action in elite higher education has just become available. Critics of the policy have long argued, among other things, that it does more harm than good, because it exacerbates rather than reduces racial hostility, and because it damages the minority students who are selected for elite schools where they must compete with other students whose test scores and other academic qualifications are much higher than their own. But a new study-The Shape of the River by William G. Bowen and Derek Bok-draws on a huge database of information about student records and histories, and on sophisticated statistical techniques, not only to refute those claims but to demonstrate the contrary.2 According to the River study, affirmative action has achieved remarkable success: it has produced higher rates of graduation among black college students, more black leaders in industry, the professions, and community and neighborhood service, and more sustained interaction and friendship among different races than would otherwise have been possible. (I have discussed the findings and implications of this study, in detail, in the last issue of The New York Review.3 ) If the Supreme Court declares affirmative action unconstitutional, the study declares, black enrollment in elite universities and colleges will be sharply reduced, and scarcely any black students will be admitted to the best law and medical schools. 4 That would be a huge defeat for racial harmony and justice. Will the Supreme Court rule that the Constitution requires us to accept that defeat?

The Fifth Circuit judges are convinced that it will, and if we are to understand why they think so, and why so many commentators fear that they are right, we must explore the apparatus of legal doctrines and distinctions that the Court has developed, over the last several decades, to assist it in applying the equal protection clause I mentioned; for this is one of those instances, created by our constitutional system, in which America’s social and political future hinges on careful legal analysis.

The equal protection clause does not, of course, protect citizens from all legal distinctions or classifications that work to their disadvantage. Government must decide which medical research to support, which art to subsidize, which industries or products to protect by tariffs or other trade policy, which businesses to regulate for environmental reasons, where to locate a new army base or airport or a new nuclear waste dump, and thousands of other matters that will affect the fates and fortunes of different citizens very differently. Officials make such decisions for a variety of reasons. In principle, they should aim at decisions that, though they benefit some citizens and disadvantage others, are in the general interest of the community as a whole. In practice, interest-group politics often play a crucial part: an industry that is denied protection or selected for regulation may have lost its legislative battle, not because a different decision would have been less in the public interest, but because it lacked the political power, on this occasion, to force that different decision.


The equal protection clause is violated, not whenever some group has lost an important decision on the merits of the case or through politics, but when its loss results from its special vulnerability to prejudice or hostility or stereotype and its consequent diminished standing-its second-class citizenship-in the political community. The clause does not guarantee each citizen that he will benefit equally from every political decision; it guarantees him only that he will be treated as an equal-with equal concern and respect-in the political processes and deliberations that produce those decisions.

But the Fourteenth Amendment therefore poses a special difficulty for the courts that must enforce it: it requires them to judge not merely the consequences of legislation for different groups, but the motive behind that legislation. Was the law that injures this or that group the product of a forbidden, prejudiced attitude toward that group, or of more benign motives? It is extremely difficult to attribute motives and attitudes to general legislation, not just because it is difficult to identify the psychological states of different individual legislators and other officials, but for the deeper reason that it is often unclear how we should translate those individual motives—and the motives and attitudes of the constituents in whose interests the legislation has supposedly been adopted—into an overall motive that we can attribute to the legislation itself.5

In some cases, that judgment seems easy, at least in retrospect. The Court rightly decided, in 1954, that racial school segregation violated the equal protection rights of black children, because segregation signaled their inferiority and exclusion. It rightly decided, in 1996, that a Colorado state constitutional amendment forbidding any local antidiscrimination protection for homosexuals violated the equal protection rights of members of that group, because, as Justice Anthony Kennedy said, “the amendment seems inexplicable by anything but animus toward the class it affects….” 6

Other cases, however, are much more difficult to assess. Does a local rent-control ordinance, for example, express a theory about wise and fair housing management or a special hostility toward landlords as a class? It seems silly to invite judges to review the political sociology of every piece of legislation that anyone challenges, because they have neither the time nor the equipment for such studies. It also seems dangerous to democracy, because judges might overrule democratic decisions on the barest speculation of improper motives.7

The courts have instead tried to approach the question of motive indirectly, through doctrines intended to “smoke out” improper motives by concentrating on the apparently more objective question of a law’s actual effects. They subject all political decisions that are challenged on equal protection grounds to an initial threshold classification. If a decision imposes serious disadvantages on what the Supreme Court has called a “suspect” class-a class, according to one prominent definition, that is “saddled with such disabilities, or subjected to such a history of purposeful unequal treatment, or relegated to such a position of political powerlessness as to command extraordinary protection from the majoritarian political process”8 -then the decision is to be subject to “strict scrutiny.” This means that it must be rejected as violating the equal protection clause unless the disadvantage can be shown to be essential in order to protect some “compelling” governmental interest. But if those whom a law disadvantages do not form such a “suspect” class-if they are only the members of a particular business or profession or the residents of a particular area, and are not different from their fellow citizens in any way historically associated with prejudice or antipathy-then that law is subject to only a “relaxed” scrutiny: it is constitutional unless it can be demonstrated to serve no purpose or point at all.

The initial assignment of any particular law or decision to one or the other of these “levels of scrutiny” has almost always proved final. As one leading commentator put it long ago, strict scrutiny is “‘strict’ in theory and fatal in fact,”9 because almost no interest has seemed sufficiently “compelling” to justify imposing further disadvantage on a suspect class,10 and “relaxed” scrutiny is in effect no scrutiny at all, because some purpose or other can always be attributed to even the most inane legislation.

So lawyers considering the constitutionality of affirmative action programs naturally begin by asking whether such programs should be initially classified as requiring strict or only relaxed scrutiny. But they have great difficulty answering that question, because neither choice seems fully appropriate. On the one hand, affirmative action plans seem entitled to relaxed scrutiny, because though they use racial classifications, the group they mainly disadvantage-white applicants to colleges and universities-do not constitute a “suspect class,” that is, a class that has been the victim of prejudice. But race is so closely associated with bias and favoritism that some racial classifications which seem benign on the surface might turn out, after a closer look, to be constitutionally offensive. Black municipal councils might conceivably have acted to favor black businesses out of racial solidarity, for example, or to punish innocent whites for the past racial crimes of their ancestors; a university admissions scheme that gives preference to blacks might conceivably have been constructed to reduce the number of Asian-Americans or Jews admitted.


Careful inspection would almost always disclose such improper motives—statistics could show whether any such group was disproportionately represented among the applicants displaced by affirmative action-but relaxed scrutiny would not permit that inspection. On the other hand, subjecting racial classifications that benefit “suspect” groups to the same standards of strict scrutiny as those classifications that impose further damage on those groups seems insensitive to the important moral differences between those two aims. It also seems perverse because, as the River study apparently demonstrates, affirmative action is one of the most effective weapons we have against the racism that strict scrutiny is designed to thwart.

So affirmative action presents a great challenge to the conventional doctrine, and lawyers and judges have suggested different responses to that challenge. The most direct-and I believe the most appealing-response would be to declare the level-of-scrutiny strategy inapposite to the problem. That strategy, as it has historically been understood and used, is designed to identify types of legislation that by their nature involve either so high a risk of invidious discrimination that invidiousness should be nearly irrevocably presumed or so low a risk that its possibility should be nearly irrevocably dismissed. Race- sensitive programs that are, on their face, designed to help a disadvantaged racial group fall into neither of these categories, and it is procrustean to try to force them into one or the other.

Instead, I believe, judges should inspect such plans, when they are challenged in litigation, on a more case-by-case basis: they should use, as Justice Thurgood Marshall once recommended, a “sliding-scale” approach in order to decide whether there is any convincing evidence that the racial classification actually does reflect prejudice or hostility of the kind forbidden by the equal protection clause.11 Such an approach would take into account, among other pertinent factors, the character of the groups benefited and disadvantaged by the program, the racial or other character of the officials who have designed and will administer the plan, and whether the plan aims at a goal-like educational diversity, for example-that has historically been recognized as appropriate for the institution in question. It is true that this case-by-case approach to the affirmative action problem would require more judicial work and provide less predictability and guidance for lower courts, at least initially, until new rules of thumb and doctrinal strategies began to emerge. But any initial loss in predictability would be more than outweighed by the more accurate discrimination between valuable and invidious policies that greater flexibility would allow.

Supreme Court justices have disagreed for many years about whether to abandon the levels-of-scrutiny approach for affirmative action and, if not, about which level to choose. In two cases, the Court tried to solve the problem by defining an “intermediate” level of scrutiny, which requires that an affirmative action plan be shown to serve an “important” but not necessarily a “compelling” interest.12 But in recent cases, chiefly through a series of opinions written by Justice O’Connor, the Court has decided that all racial classifications, including those that are apparently designed to favor rather than injure suspect groups, are subject to strict scrutiny. In 1986, in the Croson case, the Court struck down a Richmond, Virginia, city council plan that required city contractors to subcontract at least 30 percent of the dollar amount of any contract to minority-owned firms.13 Richmond called its plan “remedial,” and said it had adopted the plan “for the purpose of promoting wider participation by minority business enterprises in the construction of public projects.”

O’Connor ruled that Richmond could properly claim a “compelling” interest in rectifying the continuing effects of past discrimination only if it had itself been the author of the injustice, either directly, by its own discriminatory practices, or “as a ‘passive participant’ in a system of racial exclusion practiced by elements of the local construction industry”; and she held that the city had not shown that its plan was carefully tailored to rectify only the effects of its own direct or passive discrimination. It could not satisfy strict scrutiny, she said, by claiming an interest in achieving a racially more diverse local construction industry, because there might be many reasons, other than the continuing effects of past discrimination, why a particular race was underrepresented in a particular industry, and it was not a permissible aim of government to pursue racial diversity or proportionality for its own sake.

The Fifth Circuit judges, in their Hopwood opinion striking down the Texas Law School plan, relied mainly on the Supreme Court’s Croson decision to justify their claim that university affirmative action plans are now unconstitutional. The Texas Law School argued that its affirmative action plan was justified, even under a strict scrutiny test, because, among other things, affirmative action was necessary in order to produce a racially diverse student body-the goal that Powell had approved in Bakke. But the judges said that Croson and other cases had in effect overruled Powell’s principle. These decisions had established the new rule that no state institution may use a racial classification for any purpose except to remedy the continuing effects of its own direct or indirect discrimination. The law school could not satisfy that test, the judges said, because it had ceased discriminating against minorities many years ago. Are those judges right that Croson and later cases have had that dramatic and devastating consequence? That is a crucial question for the future of American education and society, and it is important that the public understand the actual force of these Supreme Court precedents.


In fact, the strict scrutiny test, as it has often been formulated in textbooks and judicial opinions, can be interpreted in two very different ways. One of these (which I shall call the “overriding necessity” version) supports the Fifth Circuit judges’ opinion that the Supreme Court has already in effect declared all university affirmative action unconstitutional. The other (which I shall call the “rebuttal” version) refutes that opinion. When we inspect the recent Supreme Court decisions that the Fifth Circuit judges cited with the distinction between these two interpretations of strict scrutiny in mind, we find that though three of the present Supreme Court justices-Chief Justice Rehnquist and Justices Scalia and Thomas-prefer the overriding necessity version, O’Connor’s key opinions presuppose the rebuttal version instead. We also find that the rebuttal version much better fits the opinions of the five remaining justices as well.

The two versions are based on strikingly different assumptions about the constitutional status of racial classifications. According to the first version, that of overriding necessity, any racial classification imposed by any branch of government for any purpose whatever automatically violates the equal protection clause in principle. A racial classification can therefore be tolerated only if it is absolutely necessary either as the only available means for that branch to end its own past and continuing racial discrimination, or to forestall some danger of such dramatic urgency-in Justice Scalia’s words, “a social emergency rising to the level of imminent danger to life and limb”14—that we must overlook a grave constitutional wrong in order to avoid that danger. If the Croson decision is properly understood as endorsing that version of the strict scrutiny test, then the Fifth Circuit judges were right. Racial diversity in a student body may be an important academic and social goal, but pursuing that goal would not justify overlooking a serious violation of the Fourteenth Amendment. Underrepresentation of minorities in a law school class does not present an “imminent danger to life and limb.”

The second, rebuttal, version of the strict scrutiny test rests on very different premises. It does not presuppose that every racial classification violates the Fourteenth Amendment, even in principle, and it therefore does not assume that no racial classification is tolerable unless it is required by some emergency sufficiently grave to justify overlooking a constitutional wrong. It assumes that racial classifications violate the equal protection clause only when they have been generated by the unacceptable attitudes of prejudice or stereotype that the clause outlaws. But it also supposes that since race has so often proved a ground of prejudice and favoritism, it is a wise constitutional strategy to impose a strict burden of proof on any institution employing such a classification, by demanding that the institution produce evidence of a proper motive that is sufficiently compelling to rebut any realistic suspicion that unacceptable motives were actually responsible.

This rebuttal version is much more demanding than the sliding-scale test I said that I prefer, because the rebuttal version sets the standard of proof very high.15 But it is nevertheless much more flexible than the overriding necessity version, because whether an institution is able to rebut all suspicion by pointing to some legitimate goal or interest that its racial classification serves depends not just on the intrinsic urgency of that goal, considered abstractly, but on all the concrete circumstances. It depends, among other things, on whether the goal has been part of the institution’s traditional responsibilities-for example, diversity among university students-whether the classification seems carefully designed to serve that goal, whether the institution might have other, less respectable, motives for what it has done, and on any other factors that might arouse or quiet suspicion given all the facts of the case. So a Supreme Court decision that Richmond could not satisfy strict scrutiny by declaring that it was aiming to improve racial diversity in the construction industry would not entail, or even strongly suggest, that the Texas Law School could not satisfy that test by pointing out that its admissions policies improve racial diversity in its classrooms. The grounds and character of suspicion might be so different in the two cases that a goal that fails to rebut the suspicion in one case would not fail in the other.

The rebuttal reading is much easier to justify on constitutional principle than the overriding necessity reading. The latter assumes, as I said, that the equal protection clause automatically forbids all racial classifications no matter what purposes they serve, and there is no warrant for that understanding in any plausible theory of constitutional interpretation. The Fourteenth Amendment does not mention race, and we have no reason to think that those who drafted and endorsed that amendment meant to forbid all racial classifications outright. On the contrary, many of them themselves voted for and endorsed a variety of racial classifications, including even racial segregation in public schools.

It is true that the equal protection clause lays down a general principle of political morality, and that its contemporary interpreters must make moral judgments if they are to remain faithful to that general principle.16 If racial classifications were inherently morally wrong, then they might well be deemed unconstitutional for that reason. But (as I argued in the first of these articles) racial classifications are not inherently wrong, any more than are any other classifications based on physical or genetically grounded properties. The rebuttal version of the strict scrutiny test is therefore the strongest version that the text and point of the Constitution can plausibly be thought to authorize: if the circumstances of some otherwise lawful government action that employs racial criteria are such as to rebut all genuine trace of suspicion that improper motives have been at work, the Court has no license for intervening to halt that action.

The three justices I cited-Scalia, Rehnquist, and Thomas-have nevertheless indicated that they will insist on something like the overriding necessity reading. In his concurring opinion in the Croson case, Scalia, for example, said that the only interest he would recognize as compelling, apart from the “life and limb” emergency I mentioned, is a community’s interest in eliminating “their own maintenance of a system of unlawful racial classification.”17 But there is ample evidence that the six other present justices-so far as they would subject affirmative action to strict scrutiny at all-would prefer a reading much closer to the rebuttal one.

O’Connor’s Croson opinion was starkly different from Scalia’s. True, she said that “classifications based on race carry a danger of stigmatic harm. Unless they are strictly reserved for remedial settings, they may in fact promote notions of racial inferiority and lead to a politics of racial hostility.”18 But that is carefully guarded language—“may” is not “will,” and the River study suggests that O’Connor’s concern was not justified in the case of higher education-and it is best understood as explaining why racial classifications that are not remedial in the narrowest sense must be subjected to particularly careful examination.

In any case, it would certainly be wrong to conclude that O’Connor meant that no institution could ever use racial classifications except in that narrowly remedial way.19 For she made plain that the strict scrutiny she proposed was not intended to replace a careful, case-by-case examination, designed to “smoke out” illegitimate uses of race, with a flat, mechanical rule striking down all plans that did not meet a simple a priori test. “Absent searching judicial inquiry into the justification for such race-based measures,” she said, “there is simply no way of determining what classifications are ‘benign’ or ‘remedial’ and what classifications are in fact motivated by illegitimate notions of racial inferiority or simple racial politics.”

O’Connor took special pains to point out the features of the Richmond plan which, in her view, invited the suspicion that the plan was indeed motivated by “simple racial politics.”

In this case, blacks constitute approximately 50 percent of the population of the city of Richmond. Five of the nine seats on the city council are held by blacks. The concern that a political majority will more easily act to the disadvantage of a minority based on unwarranted assumptions or incomplete facts would seem to militate for, not against, the application of heightened judicial scrutiny in this case. 20

She rejected the city’s claim, that is, not through a blanket ruling that none of the interests it cited could ever be deemed compelling, in any circumstances, but because citing those interests was not enough to dispel all trace of the suspicion raised by other features of the actual circumstances. Richmond’s plan gave preference, for example, not only to local minority firms, for whose fate it might plausibly have taken some civic responsibility, but to firms controlled by “black, Spanish-speaking, Oriental, Indian, Eskimo, or Aleut” people anywhere in the nation. That alone left room for suspicion that Richmond was not pursuing a realistic civic purpose important enough to justify a significant deviation from the normally wise rule, intended to protect the city from illegitimate favoritism of all kinds, that contracts should be awarded to the lowest bidder. “The random inclusion of racial groups that, as a practical matter, may never have suffered from discrimination in the construction industry in Richmond suggests that perhaps the city’s purpose was not in fact to remedy past discrimination,” O’Connor said.

In her later opinion in the Adarand case21 (which held that the Small Business Administration’s regulations providing special benefits to business controlled by, among others, “black, Hispanic, Asian Pacific, Subcontinent Asian, and Native Americans,” are subject to the strict scrutiny test), O’Connor was even more explicit in disclaiming any mechanical understanding of that test. She reacted strongly to the suggestion that her approach could not discriminate between invidious and genuinely benign discrimination: strict scrutiny, she insisted, does “take ‘relevant differences’ into account—indeed, that is its fundamental purpose,” and

does not “treat dissimilar race-based decisions as though they were equally objectionable.”… To the contrary, it evaluates carefully all governmental race-based decisions in order to decide which are constitutionally objectionable and which are not. By requiring strict scrutiny of racial classifications, we require courts to make sure that a governmental classification based on race…is legitimate.

“Finally,” she added, “we wish to dispel the notion that strict scrutiny is ‘strict in theory, but fatal in fact.’ The unhappy persistence of both the practice and the lingering effects of racial discrimination against minority groups in this country is an unfortunate reality, and government is not disqualified from acting in response to it.”

Five other justices of the present Court have been even more explicit than O’Connor in rejecting any mechanical version of strict scrutiny. In his Croson concurring opinion, Justice Stevens said that racial classifications should be judged by examining their impact on the future, and he expressly rejected any implication that “a governmental decision that rests on a racial classification is never permissible except as a remedy for a past wrong.” Justice Kennedy, in his concurring opinion in the case, conceded that Scalia’s position, which “would strike down all preferences which are not necessary remedies to victims of unlawful discrimination, would serve important structural goals, as it would eliminate the necessity for courts to pass upon each racial preference that is enacted.” Nevertheless, Kennedy said, he believed that so rigid a policy was unnecessary, and preferred what he called O’Connor’s “less absolute rule” that racial preferences must face “the most rigorous scrutiny.”

Justices Souter, Ginsburg, and Breyer all dissented in the Adarand case, along with Stevens. Souter wrote that “the Court has long accepted the view that constitutional authority to remedy past discrimination is not limited to the power to forbid its continuation, but extends to eliminating those effects that would otherwise persist….” Ginsburg, in an opinion Breyer joined, emphasized her view that the Court should now be using strict scrutiny not mechanically, but as an aid to discovering actual legislative motives that are illegitimate because they offend the equal concern required by the equal protection clause. The strict scrutiny test, as defined in O’Connor’s majority opinion, Ginsburg said, is a device “to ferret out classifications in reality malign, but masquerading as benign.”


So the two-judge opinion in Hopwood was wrong in assuming that the Court has already adopted a mechanical strict scrutiny test that makes university affirmative action plans in the Bakke mold automatically unconstitutional.22 It does not follow, however, that the Court will not strike down race-sensitive admission standards, in the test case that many commentators now predict, even on a less mechanical, rebuttal reading of strict scrutiny.

So we must ask whether and how university affirmative action plans can meet a strict scrutiny test construed in that way. The River study suggests two main justifying purposes for race-sensitive admissions tests: the universities’ own need for racial diversity in their student bodies, and the community’s need for a greater presence of minority members in important political, business, and professional roles. Are either of these needs sufficiently “compelling” to justify the use of race as one factor among many in evaluating applicants? Does the record rebut any scintilla of reasonable suspicion that the schools surveyed in the River study have used race for illegitimate purposes?

Powell himself insisted, in Bakke, that affirmative action plans were subject to strict scrutiny, and his decision that universities may seek racial diversity was therefore a ruling that diversity was a sufficiently compelling interest to survive that scrutiny. It is true that O’Connor has rejected a diversity justification in other contexts, not only in Croson, but in a dissenting opinion in the Metro Broadcasting case, in which the Supreme Court sustained policies of the Federal Communications Commission that gave preference to minority-owned firms in applications for licenses for new radio and television stations-the FCC claimed that such preferences were necessary in order to improve diversity of viewpoint in programming. But neither of these O’Connor opinions forecloses allowing universities to use racial classifications to produce racial diversity in the classroom.

The overriding question, for a rebuttal reading of strict scrutiny, is whether an institution’s appeal to diversity is sufficient to dispel any genuine trace of suspicion that it has acted out of constitutionally forbidden motives. Richmond’s appeal to diversity was compromised not only by the factors I mentioned earlier, but because diversity has not been a traditional goal of officials in charge of awarding municipal construction contracts. On the contrary, a city that claimed, say, geographical diversity as a reason for denying construction contracts to the lowest bidders would raise deep suspicion of corruption. The FCC regulations that O’Connor condemned in her Metro Broadcasting dissent were open, as she emphasized, to a grave though different kind of suspicion: the argument that diversity in ownership is necessary to achieve diversity in broadcasting, she said, relied on racial stereotypes because it assumed that people’s “race or ethnicity determines how they act or think.”

O’Connor argued that the asserted interest in diversity of programming is, in any case, “too amorphous, too insubstantial” to rule out any possibility of racial preferences or prejudices. The FCC, she said, might, under cover of this alleged interest, identify a “black” or “Asian” or “Arab” viewpoint, and then deny licenses to races or ethnic groups it deemed less likely to present the favored view. She particularly feared that recognizing a general interest in diversity would allow great and indiscriminate use of racial classifications not just for particular purposes and a limited time, but for all purposes and for all time. Because it is impossible to define a particular racial viewpoint, or to assess how diverse one viewpoint is from another, she said, “Members of any racial or ethnic group, whether now preferred under the FCC’s policy or not, may find themselves politically out of fashion and subject to disadvantageous but ‘benign’ discrimination.”

Universities are in a much stronger position than Richmond or the FCC was to dispel any suspicion that they seek racial diversity for improper underlying motives or on stereotypical assumptions. University admissions policies are not set by politicians, who might hope to court the votes of a racial bloc, but by faculty members, who are not running for office. Their interest in diversity is not novel or unusual, as Richmond’s was, but traditional and recognized: no one disputes that large, mainly white, universities have a social and educational responsibility to seek a student body that is diverse in many ways, and any such university that abandoned that aim altogether would be behaving irresponsibly. Elite universities believe that it would now be irrational to seek diversity in geographical origin, in social class, and in cultural orientation, and not also to seek racial diversity.

Indeed, their failure to seek the latter dimension of diversity as well would make their general concern with diversity seem arbitrary. They have decided, and the River study amply confirms their view, that they cannot achieve racial diversity indirectly by relying on economic class as a proxy for race, or by using otherwise less efficient means to the hoped-for end. Any such policy would be not only disingenuous but harmful.23 Nor do universities rely, as O’Connor said the FCC did, on any presumed connection between race and belief, conviction, taste, culture, or attitude.

They seek racial diversity, as I said in my article in the last issue, because race is itself important, unfortunately but inescapably, in contemporary America: it is vital that students of each race meet and work with, not just other students with other attitudes or culture, but students who are in fact of a different race.24 Nor would the courts be risking open-ended and indiscriminate racial preferences by continuing to permit affirmative action on the Bakke model. Universities have used such programs judiciously for a third of a century, with no tendency to expand them beyond sensible proportions.

These institutions have, moreover, a crucial stake in their academic reputations, both absolutely and relative to other comparable institutions, that would check any desire significantly to expand an admissions policy or curriculum that might threaten that reputation. Nor is there any genuine risk that race-sensitive admissions programs will be used as a pretense for disfavoring any other particular group of applicants. Any suspicion of that could be tested, as I said, using statistical means like those used in the River study, by analyzing the retrospectively rejected students to see whether they were disproportionately members of any suspect group.

There is ample evidence, moreover, that O’Connor, as well as several other members of the present Court, has already accepted that the search for racial diversity among students is a compelling interest that survives strict scrutiny. In 1986, in the Wygant case, the Supreme Court struck down a Michigan school board’s collective bargaining agreement that gave minority schoolteachers special protection against layoffs: it rejected the school board’s claim that its interest in correcting the effects of past discrimination in the community at large, or in providing black faculty “role models” with whom black students might identify, justified this racial classification.25 O’Connor wrote a separate opinion in which she noted that the board had not claimed that it had acted to protect racial diversity on its faculty, and that the Court was therefore not to be understood to have ruled out that interest as compelling.26 “Although its precise contours are uncertain,” she said, “a state interest in the promotion of racial diversity has been found sufficiently ‘compelling,’ at least in the context of higher education, to support the use of racial considerations in furthering that interest.”27

O’Connor has several times, moreover, cited Powell’s Bakke opinion, which declared diversity in higher education a compelling interest, as authority for her view that any racial classification must be subjected to strict scrutiny. She would hardly rely on that opinion with such force if she thought that Powell did not himself understand the implications of his strict scrutiny approach, or intended to lay down, under that name, a different doctrine from the one for which she cited his authority.

The argument is therefore strong that the Bakke principle, in force for over twenty years, remains good constitutional law, and that American colleges and universities may continue to rely on that principle to justify using race-sensitive admissions policies to secure a diverse student body. If I were defending such schemes in the courts, I would certainly emphasize that interest in student diversity, which seems enough, on its own, to ensure that the programs survive strict scrutiny. I must add, however, that I believe that the other institutional interest I mentioned-helping to redress the still-deplorable absence of blacks from key positions in government, politics, business, and the professions-is at least an equally important one that should also be recognized as sufficiently compelling to sustain race-sensitive admissions policies. One of the gravest problems of American society is the de facto racial stratification that has largely excluded blacks and other minorities from the highest ranks of power, wealth, and prestige; and past racial discrimination, as well as the vicious circle that robs black children of successful black leaders to emulate, has contributed substantially to that stratification.

Nevertheless, many statements sprinkled throughout the various Supreme Court opinions I have been discussing might well be read as hostile to that further, and different, justification of race-sensitive admissions policies, including Powell’s statement, in Bakke, that medical schools may not use affirmative action just in order to increase the number of black doctors.

Several of the justices have declared that racial classifications cannot be justified as helping to cure the lingering effects of past “societal discrimination,” and the Wygant decision rejected the claim that they can be justified as providing “role models” for black children. It might be, however, that these statements have not paid sufficient attention to the distinction that Justice Stevens has several times made-between backward-looking justifications of racial classifications as compensatory and forward-looking justifications that argue that such classifications may, in some circumstances, be in the general interest of the community as a whole.

Compensatory justifications suppose that affirmative action is necessary, as Scalia put it, to “make up” to minorities for damage done to their race or class in the past, and he was right to point out the mistake in supposing that one race “owes” another compensation. But universities do not use race-sensitive admission standards to compensate either individuals or groups: affirmative action is a forward-looking, not a backward-looking, enterprise, and the minority students whom it benefits have not necessarily been victims, as individuals, of any distinct injustice in the past. Great universities hope to train more blacks and other minority students not to repay them for past injustice, but to make the future better for everyone by helping to lift a curse that the past laid on us all.

O’Connor and other justices have worried that any broad and general remedial justification for affirmative action is too “amorphous” and “open-ended” because it would license racial preferences until every industry or social or professional stratum had the same racial and ethnic composition as the nation as a whole. But however genuine or inflated that concern might be as a worry about the consequences of government-imposed hiring or contracting regulations, it is distinctly out of place as an objection to university affirmative action plans. If any branch of government-whether Congress or a local city council-requires employers or contractors to hire a quota of black employees, or to set aside a quota of contracts for black firms, its decision ensures a particular racial representation in some segment of employment or industry. No more natural process of decision-making can alter or shape that racial structure so long as the government’s program is in place. In such cases, government, and only government, decides how many members of each of the racial or ethnic groups it designates will fill which jobs in which sectors or roles or offices. Judges who are particularly sensitive to the danger that some of these decisions might be made out of improper motives will be reluctant to accept so broad a justification for them as the claim that they are necessary to prevent excluding one or another race from power, wealth, and prestige.

But colleges, universities, and professional schools use race-sensitive standards not in response to any central government mandate but through individual decisions by individual schools. They act, not to fix how many members of which races will occupy what roles in the overall economy and polity, which is in any case beyond their power, but only to increase the number of blacks and other minorities who are in the pool from which other citizens-employers, partners, patients, clients, voters, and colleagues acting in their own interests and for their own purposes-will choose employees, doctors, lawyers, and public officials in the normal way.

The distribution of position and power that affirmative action helps to achieve, that is, flows and changes naturally in accordance with millions of choices that people make for themselves. If the policy works to improve the overall position of any minority—as the River study suggests it has helped to improve the position of blacks-it does so only because other people have chosen to exploit the results of that policy: the greater range and variety of graduates with the motive, self-respect, and training to contribute effectively to their lives. Affirmative action in universities, in that way, makes the eventual economic and social structure of the community not more artificial but less so; it produces no balkanization, but helps to dissolve the balkanization now sadly in place.

If the justices recognize this aspect of what our best universities aim to do, as well as their academic need for educational diversity, then they will have served us particularly well. They will have acted not just as judges allowing a crucial educational initiative to continue, but as teachers helping to explain to the nation the true and continuing costs to everyone of our racist past, and the distinct promise of an educational policy that can help us all to achieve, if we really want it, a more perfect union.

(This is the second of two articles.)

This Issue

November 5, 1998