Assessing Affirmative Action
In affirmative-action cases, the Supreme Court never seems to learn the obvious lesson, or perhaps it is determined to finesse it. The lesson is that universities that are keen to implement race-based affirmative action (and it is hard to find a highly or even moderately selective school that isn't) will figure out a way to do so unless the Court emphatically and clearly prohibits it. The Court's doctrinal fuzziness, which its 2013 decision in Fisher v. University of Texas only further obscured, allows these institutions to maintain the preference systems dictated by their political, reputational, and ideological incentives.
In a number of states, the American public has been able to exercise some countervailing influence through the ballot box, and the Supreme Court recently affirmed its right to do so. In a case decided in April, the Court upheld Michigan voters' power to ban affirmative action — indeed, to ban the same affirmative-action program at the University of Michigan that the Supreme Court had upheld in 2003 — through a constitutional amendment adopted by voter referendum.
But while Americans consistently voice a firm opposition to affirmative action in university admissions, the public debate surrounding the issue has been clouded by both weakly reasoned Supreme Court jurisprudence and incoherent factual claims by supporters of race-conscious admissions.
Over the course of 35 years, the Court has upheld race-based affirmative-action programs based solely upon a so-called "diversity rationale." All of the current justices except Antonin Scalia and Clarence Thomas have accepted this rationale. But the premises underlying the diversity rationale for race-based affirmative action are empirically tenuous and theoretically implausible. Policies justified under that rationale thus could not survive if the "strict scrutiny" standard were seriously applied. The Court applies strict scrutiny when assessing race-based policies like affirmative action, whether in higher education, government hiring, or government contracts. The framers of the 14th Amendment may have countenanced affirmative action favoring former slaves and perhaps their descendants, but they would never have approved of today's affirmative-action programs, in which most of the potential beneficiaries are immigrants or descendants of immigrants. But regardless of whether such programs are constitutional or not, they are undesirable public policy, indeed perverse in practice. Their costs vastly exceed their benefits, and in ways that should cause universities and courts alike to change course.
By considering both the legal architecture of the Court's affirmative-action jurisprudence and the empirical evidence regarding the effects of affirmative action in higher education, we can begin to see the defects of today's affirmative-action regime and the powerful case for change.
LOOSE SCRUTINY
The Court's jurisprudence regarding affirmative action in university admissions has taken shape through a series of cases stretching back to Regents of University of California v. Bakke in 1978.
In that case, the Court allowed a university to consider race in medical-school admissions so long as it was only one of several factors, was used to advance student-body diversity, and there was no specific quota for admission. The Court reaffirmed these three principles a quarter-century later in the 2003 cases of Grutter v. Bollinger and Gratz v. Bollinger. In Grutter, involving the admissions policy of the University of Michigan's law school, the majority ruled that an admissions process giving some advantage to "underrepresented minority students," but also taking into consideration a variety of other factors (applied strictly on an individual basis for each applicant), did not amount to a quota system and so was constitutionally permissible. But in Gratz, which looked at that university's undergraduate admissions policy, the Court found that, because it granted a set number of admission points to any racial-minority applicant (rather than considering each applicant individually as the law school did), it amounted to an impermissible quota system.
The Court reaffirmed this standard in the 2013 Fisher case, sending the case back to the lower court on the ground that those judges had failed to correctly apply the strict-scrutiny standard. That standard requires "a judicial determination that the burden [the disappointed applicant] is asked to bear" in a system that takes race into account "is precisely tailored to serve a compelling governmental interest."
Strict scrutiny is supposed to be, well, strict. Its purpose is to force reviewing courts to be rigorous, skeptical, and demanding enough to challenge the government's premises, flush out its true motives, and ensure a tight congruence of evidence, legal categories, and policy justifications. Courts apply strict scrutiny when they have strong reasons to think that the state may be playing with fire around highly combustible materials. Racial classifications epitomize this kind of risk.
Many academic advocates of preferences, to be sure, maintain that the Court's strict-scrutiny standard, as elaborated in earlier decisions, was too strict, even procrustean, and that a "benign" preference adopted by self-abnegating ethnic majorities should be judged less rigorously. There is much to be said for this argument, although benignity is in the eye of the beholder. The large number of people disadvantaged by preferences — all whites; the 48% of Hispanics, similar proportions of Asians, and 80% of Native Americans who self-identify to the Census Bureau as white; anyone else (dark-skinned Middle Easterners, for example) who is not considered part of David Hollinger's "ethno-racial pentagon"; and the more than seven million people (many with black ancestry) who consider themselves multi-racial and wish to be identified as such (if they must be racially identified at all) — are unlikely to think of this disadvantage as benign. But rather than adopting this benign-preference argument, the Court has instead diluted strict scrutiny without actually saying so.
Indeed, the Grutter majority did not even come close to applying the strict-scrutiny standard as traditionally understood. As Chief Justice William Rehnquist correctly noted in his dissent, the majority's review of Michigan's preference system was "unprecedented in its deference." The majority failed to explain convincingly why universities that sponsor preferences should receive more such deference than, say, the private employers or municipal procurement agencies whose plans the Court has struck down in the past under strict scrutiny. After all, universities that adopt and structure such programs are responding to the same kind of political, ideological, competitive, social, legal, and institutional pressures that affect employers and government agencies. If anything, the well-documented leftist bias and political correctness of such universities in these matters — on display in their recent backtracking on commencement speakers — should arouse the very suspicion about motives that strict scrutiny is intended to test.
There is no question of good faith: All the university administrators in these cases are clearly doing what they think best advances their institutional values and interests. But the Court has often held, most recently in Fisher, that strict scrutiny makes the actors' good faith and the purity of their intentions quite irrelevant as a legal matter.
Justice Sandra Day O'Connor's permissive approach to strict scrutiny in Grutter (a laxness that has defined courts' attitudes to affirmative-action programs ever since) deviated radically from her own more rigorous approach to that test in earlier cases. She finessed (or mischaracterized) five crucial questions that bear on affirmative-action policies: the nature of educational diversity; how educational diversity relates to both the "critical mass" idea at the core of Michigan's theory and the ethno-racial stereotypes that the school claimed to abhor; how the majority distinguished between valid and invalid preferences and how Michigan's program fared under that test; the existence of race-neutral alternatives; and the duration of preference policies.
The Court's superficial treatment of these questions reveals the profound weakness of its affirmative-action analysis.
DIVERSITY AND REMEDIATION
First, the Grutter majority reiterated Justice Lewis Powell's embrace in Bakke of student diversity as a compelling state interest sufficient to justify admissions preferences — even as it brushed past the conditions that the Bakke plurality had imposed for validating them. Of these conditions, a rigorous, individualized appraisal of an applicant's actual diversity value was the most important.
What the majority did not provide was a coherent account of the meaning of diversity value that went beyond general platitudes. Nor did it explain why the Constitution allowed the law school to define the desired, favored diversity in narrow ethno-racial terms that excluded even most minorities (other than African-Americans, Native Americans, and the Spanish-surnamed) while treating other kinds of diversity as either much less weighty or wholly irrelevant to satisfying the overriding diversity rationale. Indeed, as Justice Thomas pointed out in a footnote in his dissent, the school seemed not to value the additional diversity that black men, who are greatly under-represented relative to black women, would provide.
The only convincing explanation for Michigan's approach had little to do with the goal of educational diversity and everything to do with the desire to remedy the historic injustices suffered by African-Americans and Native Americans. (The moral claims of those with Spanish surnames are more debatable.) This motive, although admirable, is one that the Court has consistently held cannot support the constitutionality of a university's race-based preferences. Any sophisticated observer (surely including all members of the Court) who is not blinded by the rhetorical fog thrown up by the now-obligatory diversity talk understands that the law school's true purpose was not diversity but remediation.
If Michigan's program were really about educational diversity rather than remediation, it would not have limited the program's benefits to a few favored groups. It would instead have included religion, political ideology, and other demographics that directly represent the different worldviews with which educational diversity is supposedly concerned. Would an evangelical Christian or a conservative Republican create less diversity value for Michigan's students than an applicant whose only special claim to such value was his skin color or surname? The answer, obviously, is no — and this is true whether one defines diversity value in terms of disparate worldviews, interacting with people of unfamiliar backgrounds, encouraging dorm-room chit-chat, or even breaking down traditional stereotypes.
What distinguishes African-American students (at least those descended from American slaves) from these other groups is not their diverse, exotic views but the injustices visited on their ancestors. Those injustices should matter a great deal to our society in general, but the Court has always ruled that they cannot justify race-based university admissions.
CRITICAL MASS
The Grutter majority insisted that breaking down ethno-racial stereotypes was crucial to achieving the putative goal of educational diversity, and that the "critical mass" of favored minorities intentionally produced by the school's preferences would help to achieve this goal. But this notion of critical mass is incoherent.
First, critical mass by its very nature must be a matter of the number or proportion of students needed to produce it, yet even the Grutter majority had to concede that the Constitution prohibits numerical or proportional quotas and instead requires individualized assessments. Second, the critical-mass criterion is only intelligible if one specifies the level of university activity at which racial assignments are permissible in order to achieve the critical mass. Is the level campus-wide? Program-wide? Each major or only some? Seminars? Lecture courses? Dormitories? Sports teams? Grutter didn't say.
Third, the critical-mass goal can be administered only by deferring to the university's judgment on these matters, a level of deference that Grutter countenanced but that Fisher (as we shall see) emphatically rejects. And finally, the Grutter majority saw a very close connection between critical mass and stereotype destruction: "[W]hen a critical mass of underrepresented minority students is present, racial stereotypes lose their force because nonminority students learn that there is no 'minority viewpoint' but rather a variety of viewpoints among minority students." But how can this possibly be? What alchemy enables the law school to prefer students on the basis of skin color or surname — albeit in the name of diversity — without at the same time strengthening the notions of ethno-racial essentialism and viewpoint determinism?
More to the point, what opaque prestidigitation allows the school to admit minority students with academic records that were (whatever the school's metric) substantially weaker than those of their (other minority or majority) competitors without thereby reinforcing stereotypes of academic inferiority? Did the Court majority think that the non-preferred students and faculty were somehow so clueless that they would not notice what was going on and draw the logical and stigmatizing inference that the preferred group's academic performance was inferior? Wouldn't these elite educational institutions by definition attract the most competitive students and faculty? And wouldn't such individuals place a very high (perhaps excessive) value on the coin of their particular realm — academic excellence — partly because they themselves were so well endowed with it?
To recognize the importance that elite schools give to academic factors is emphatically not to say that test scores and GPA are or should be decisive in determining admission. A sensible institution will consider a variety of factors in selecting its student body, although elite schools that hope to maintain their positions will weight academic potential or performance most heavily. But to assert, as admissions officers who defend affirmative action typically do, that people with manifestly inferior academic performance are somehow superior to others with respect to non-academic virtues — leadership, character, community commitments, grappling with obstacles — is to indulge in a patronizing and pernicious stereotype. There is simply no a priori reason to believe, for example, that black applicants as a group are more likely to exhibit these "soft" variables than are applicants of other groups. Nor am I aware of any evidence that they in fact do so.
By the same token, a critic of Grutter need not deny the obvious fact that preferences of other kinds exist and that some of them may be unfair or otherwise objectionable. Preferences favoring legacies and athletes, for example, are also widespread even at elite schools, and such preferences also engender stereotypes about their academic inferiority (e.g., "dumb jocks"). Why, then, don't preferences based on these variables trigger strict scrutiny, while ethno-racial ones do? Their different legal statuses reflect deep differences in the constitutional and historical meanings of these groups and their experiences. These differences in no way justify admission preferences for athletes and legacies, but they help to explain why their stereotypes are far less corrosive and stigmatizing than those that attach to racial minorities.
INVALID PREFERENCES
The majority in Grutter defined a theoretically defensible test for the constitutionality of ethno-racial preferences: "[E]ach applicant must be evaluated as an individual and not in a way that makes an applicant's race or ethnicity the defining feature of his or her application," which would place members of all groups on the same admissions track where they would compete "on the same footing." But having announced this test and repeated it in subsequent cases, the majority proceeded to dumb it down so that all could pass it while still insisting on its integrity.
Race and ethnicity, Justice O'Connor wrote in Grutter, could be a "'plus' factor" in a system of "individualized assessments" so long as it did not constitute either a "rigid quota" (as in Michigan's undergraduate program) or "racial balancing." But as the dissenting opinions of Chief Justice Rehnquist and Justice Anthony Kennedy showed statistically, the law school weighed the plus factor for race and ethnicity so heavily that it created — in effect though not in name — a two-track system tantamount to racial balancing in its admission offers, and did so in order to achieve its racially defined "critical mass." The majority, like the law school, wanted to have it both ways: On the one hand, ethno-racial factors must only be a "modest" plus factor in admissions, but on the other hand, few of the favored minorities would have been admitted without them. Again, the majority could only reconcile this inconsistency if — contrary to reason, experience, and egalitarian theory — the favored applicants were so superior to the others on the non-academic criteria that their academic deficits could be ignored.
The truth, as evident from the Grutter dissenters' analyses (and indeed from common sense), is that the law school's affirmative-action program could satisfy the majority's constitutional test for only one reason: The program's opacity allowed so much room for subjective, discretionary judgments in the undisclosed weighting of the "soft variables" for individual applicants that rejected students (like Barbara Grutter) could not possibly prove that they were rejected primarily because they were not members of the favored ethno-racial groups. This fact, of course, placed the burden of proof squarely on the shoulders of the individual challenging the preferences — which in turn made the decidedly un-strict nature of the majority's scrutiny that much more pivotal in protecting preferences in such cases.
RACE-NEUTRAL ALTERNATIVES
The Grutter majority's abject deference to the university was nowhere more apparent than in how it treated the race-neutral alternatives to which the law school must give "serious, good faith consideration." The Court maintained that the law school had weighed the alternatives but chose to continue using racial criteria because none of the alternatives would have given it the number of preferred minorities that the institution felt it needed to achieve "critical mass." This was true enough, but it was an answer to the wrong question. The right question was: Given a constitutional presumption against ethno-racial preferences — a presumption so strong that strict scrutiny is required to enforce it — how imperfect must a race-neutral alternative be before the Court will allow the state to reject it in favor of a race-conscious (indeed, race-determined) program? There is no clear answer to this question, of course, but the majority did not even ask it, and so it did not consider any alternative approaches.
There are, of course, several alternatives, some of which are essentially race-neutral. None would be perfect, but they would be less problematic than race-conscious admissions. One often-discussed approach, familiar from the many institutional programs that grant need-based tuition assistance, seeks to determine disadvantage or need directly. Extending such programs to schools' initial decisions to admit would be more difficult, of course. Determining economic need directly for a very large number of applicants would be at least as challenging as it has proved to be in the administration of state need-based social-welfare programs. Michael Kinsley, who supports some forms of affirmative action, described the consequences in the New Yorker: "Is it worse to be a cleaning lady's son or a coal miner's daughter? Two points if your father didn't go to college, minus one if he finished high school, plus three if you have no father? (or will that reward illegitimacy which we're all trying hard these days not to do?)...Officially sanctioned affirmative action by 'disadvantage' would turn today's festival of competitive victimization into an orgy."
Difficult, surely, but not impossible. Richard Sander, a law professor at the University of California, Los Angeles, and co-author (with Stuart Taylor, Jr.) of the important book Mismatch, reports that he actually devised and implemented a sophisticated system of preferences for UCLA based on economic need, and that the system worked "exceedingly well. Audits against financial aid statements showed little abuse; the preferences substantially changed the social makeup of the class and never, to our knowledge, prompted complaints of unfairness." Such approaches would have to be assessed and attempted more broadly, of course, but they may offer one race-neutral alternative to affirmative action.
Another possible alternative would be a program that automatically admitted students in the upper echelons (say, the top 5% or 10%) of their high-school classes. Texas, Florida, and California have adopted such percentage programs (although Texas, unsatisfied with the number of minorities that its percentage plan yielded, added to it the race-based program challenged in Fisher). Percentage programs seem to increase racial diversity on college campuses, though they presumably bring to those campuses many students whose academic preparation is relatively poor (given differences among high schools in different communities).
Another race-neutral alternative would not only increase the number of minority students attending selective institutions but also ameliorate a different, more tractable, and even more socially wasteful kind of problem: A substantial pool of high-school students who are perfectly capable of performing well at selective colleges do not even apply to them, or indeed to any college at all. Caroline Hoxby and her colleagues have shown that applications by these students, many of whom are minorities, can be increased through better information about how to apply, financial-aid opportunities, and other assistance available on campus. Moreover, increasing applications from this group can be accomplished at trivial cost — as little as $6 per student.
Finally, there are ways to reduce the size of ethno-racial preferences without wholly abolishing them. In Mismatch, for instance, Sander and Taylor propose that if ethno-racial preferences are retained, their magnitude should not be permitted to exceed in size the preferences (if any) that the same school uses for socioeconomically disadvantaged students of all races.
Whatever the imperfections in these (or other) race-neutral alternatives may be, they pale before the legal and policy defects of ethno-racial preferences. Most important, the Court has demanded precisely this kind of comparison, most recently and clearly in Fisher.
THE DURATION OF PREFERENCES
Much has been made of the Grutter majority's expectation that "25 years from now, the use of racial preferences will no longer be necessary." Justice Thomas's recitation, in his dissent, of the grim statistics on comparative academic performance makes such a hope seem unrealistic. And the studies of ethno-racial preferences in other societies provide no support for it either, as the economist Thomas Sowell has shown. To the contrary, they indicate that such preferences, once established, tend to endure and perhaps even expand to new groups and new programmatic benefits.
It is true that six politically diverse states (Arizona, California, Michigan, Nebraska, Oklahoma, and Washington) have banned these preferences by voter referenda, while New Hampshire has done so through statute and Florida through executive order. But California's experience after its voters banned ethno-racial preferences suggests that such bans on affirmative action do not end it; they simply drive it underground. The California system engaged in a series of stratagems in the early 2000s expressly designed to circumvent the state's ban. Some of the more egregious ones involved channeling minority students to new "critical race studies" programs with lower admissions standards; awarding special admissions credit for foreign-language fluency to minority students who were already native speakers; adopting "percentage" plans that relied for their efficacy on the continuation of segregated schooling patterns; and using unspecified (and unspecifiable) "holistic" criteria as well as winks and nods by admissions officials.
Moreover, the Court's evident desire to end the debate about preferences — by allowing them to continue but insisting that they will not go on forever — is more or less destined to fail. Viewed most charitably, Justice O'Connor's opinion in Grutter attempted to craft a kind of compromise that might resolve this bitter debate once and for all, enabling American society to "move on." But if that was her strategy, Fisher, Schuette, and other such challenges show that it has failed miserably.
Indeed, O'Connor's hope to use a Court decision to achieve closure — if that was her intention — was probably doomed from the start. Our national experience suggests that divisive public issues in American life, of which affirmative action is certainly one, are not resolved by Court decision or by any official fiat. History has condemned the Court's attempts to settle such issues prematurely, peremptorily, or ex cathedra. Dred Scott v. Sandford, Plessy v. Ferguson, Lochner v. New York, Roe v. Wade, and Bowers v. Hardwick are among the many decisions that illustrate the point.
In Fisher, its most recent decision on the substance of university affirmative-action policies, the Court made a different attempt to foreshorten such policies. The Fisher majority (which consisted of seven justices) did not explicitly criticize Grutter (or Bakke) in any way. Nonetheless, it offered what might fairly be called a stern implicit rebuke. Fisher required the states (and reviewing courts) to apply those decisions' precepts and tests more rigorously than the lower court had done when it upheld the Texas program in question.
Specifically, Fisher focused attention on the narrow-tailoring prong of the "most rigid" strict-scrutiny standard and on a couple of other requirements that Grutter had affirmed. To address strict scrutiny, the decision focused on a few important aspects. First, the desired diversity must be achieved with no more use of race or ethnicity than necessary and only after considering all race-neutral alternatives. Second, even in the higher-education context, the state is entitled to "no deference" on the key factual narrow-tailoring determinations of whether the fit between its programmatic means and its diversity-enhancing end is tight enough, and whether there are workable race-neutral alternatives. And third, the state bears the burden of proof on these issues, proof that will probably require a trial rather than being resolved on summary judgment. Looking beyond strict scrutiny, a school must not use diversity as a pretext for "racial balancing, which is patently unconstitutional," and as a result a program must assess applicants as individuals. Perhaps most telling in Fisher's analysis was the fact that it ignored the idea of "critical mass," a central pillar of Grutter's defense of preferences.
In other words, the Court upheld but seemed to tighten its permissive attitude toward affirmative-action admissions policies. Rather than suggest wistfully that affirmative action might end someday, the majority indicated that it must be constrained. But this warning by the Court about the constitutional infirmities of ethno-racial preferences seems only to have emboldened their advocates, who took the ruling as an unqualified victory. And despite the Court's ruling, very few admissions offices plan to alter their policies.
THE BENEFITS OF DIVERSITY
Whatever the constitutional status of race-based affirmative action may be, it is even weaker as public policy, for it fails the essential test that any policy must pass: Its diversity-based benefits are lower than its costs.
I shall make no effort here to formally quantify either the benefits or the costs. Instead, my approach is to take the diversity rationale seriously by clarifying the elements that it must entail, and then to show that the social benefits claimed by the programs' proponents are modest at best and plainly swamped by their costs to society and even to many of their supposed beneficiaries.
What, then, are the benefits in terms of diversity that should count in favor of using preferences in university admissions? To answer this essential question, we must address three other closely related ones. What does diversity mean in this context? What is it about any group that accounts for whatever social benefits are conferred by its presence rather than the presence of some other group? And what diversity value is actually generated by the specific groups that affirmative-action programs favor?
Unfortunately, few discussions of diversity and the diversity rationale for affirmative action even address what diversity actually means, much less explain which groups and which kinds of attributes create diversity value. Nevertheless, the ways that affirmative-action programs are designed and defended leave no doubt that program advocates almost always mean ethno-racial diversity. This is true despite the many anomalies, evasions, and confusions that pervade most ethno-racial discourse. For one thing, race is a spurious category. People of different races have married and had children together in America since colonial times, producing a very large number of mixed-race individuals. Furthermore, recent immigrants are included in preference programs. "Hispanic" is a language designation, not a racial one. "Asian" refers to many ethnic, religious, linguistic, and national-origin groups with little or nothing in common with one another, and indeed with histories of deep conflict. The category of Native Americans is equally artificial and absurdly broad.
The preoccupation with race by proponents of the diversity rationale is also misplaced because other attributes are at least as predictive of a person's unique experiences, outlooks, and ideas — and hence at least as relevant to the kind of diversity that universities claim to want. According to a study by Northwestern University's James Lindgren of the demographic correlates of viewpoint differences, political affiliation accounts for the largest cleavages, while religion produces roughly the same cleavages as race. The failure of affirmative-action programs putatively aimed at diversity to base preferences on religion is among the most revealing facts about the programs' true objectives. Indeed, the programs' lack of interest in exploiting our rapidly growing religious heterogeneity for the purpose of campus diversity casts great doubt on the coherence of the diversity rationale as now implemented through these programs. This lack of interest may in turn reflect the remarkably unrepresentative religious composition of university faculties. As University of Texas scholar Sanford Levinson observes: "One sometimes gets the feeling that ostensible defenders of 'diversity' and 'multiculturalism' have no real idea of how truly diverse and multicultural the United States has become, fixated as they are on the 'traditional' racial and ethnic cleavages within this country." A priori (which is how programs selected the groups to be preferred), doesn't the perspective of a Muslim or conservative Christian applicant have at least as much diversity value as that of a middle-class black or Hispanic?
So just how many African-Americans or members of other favored groups must be present in order to establish the requisite diversity value? Because diversity value surely depends on various factors, any sensible answer must be context specific. Unfortunately, the law and practice of affirmative-action programs instead offer a wholly reductionist answer to this question. They simply count the number of group members in the relevant community (or their percentage of the community total) and seek proportional representation, at least as a default but often effectively as the final answer.
Defining the relevant community — that which will be used in making the proportionality assessment on which the admission decisions will turn — almost always entails highly controversial judgments, if not arbitrary empirical and normative ones. The relevant baseline for judging proportionality, for example, can only be defined in terms of a number of elusive, hard-to-measure, and internally competing parameters, including group definition, geography, qualifications, attitudes, applicant pool, and others. Rhetoric aside, the task of actually administering affirmative action requires, ironically, that a program first combine many complex determinations that as a practical matter it can only make through almost comically arbitrary judgments, and then coming up with a bottom-line number that is certain to be breathtaking in its simplicity and lack of context. It would be surprising indeed if institutions that must process thousands of applications for relatively few slots in a very limited period of time did not pursue a spurious or formalistic kind of diversity: one that means color coding and color counting in service of a predetermined color targeting.
How, then, does a favored group in fact confer diversity value on an academic community? This is the second question we must confront when assessing affirmative action. A group can only confer such value if it possesses certain desired qualities as a group. It follows that a group can only do this if those qualities inhere in all members of the group (or else the group should be redefined to exclude those who lack them). But to affirm that a quality inheres in a racial group is to "essentialize" race in a way that utterly contradicts the bulk of our liberal, egalitarian, legal, scientific, and religious values. Together, these values hold that all individuals are unique and formally equal regardless of genetic heritage, and that their race causally determines little or nothing about their character, intelligence, experience, or anything else that is relevant to their diversity value. Indeed, for employers to use racial stereotypes in this way would be flatly illegal — even if the assumptions underlying the stereotypes were true.
The best that can be said of affirmative-action programs' use of diversity in this way is, as David Hollinger has put it, that they have reproduced "the most gross and invidious of popular images of what makes human beings different from one another" for a putatively benign purpose. They are propagating socially inflammatory stereotypes that, even if they were accurate, invite decision-makers to violate people's claim (and constitutional right, according to Bakke, Grutter, and Fisher) to be judged as individuals, not as members of ascribed groups. By parity of reasoning, legitimating the use of this proxy might equally justify racial profiling by police if it were intended to fight crime and were no less accurate than is the crude stereotyping in affirmative-action programs.
This brings us to the third crucial question about the diversity rationale for affirmative action: What diversity value does a favored group actually confer? Affirmative-action programs attempt to finesse the essentialism difficulty discussed above by assuming certain facts that might make the use of race as a proxy more defensible. They assume, first, that African-American students bring to campus histories and viewpoints that are unique to, and nearly universal among, African-Americans — even though those histories and viewpoints are not racially or genetically hardwired into them. Educational institutions and their African-American members, the programs further assume, should help people of other races to comprehend this experience, and campus diversity can strengthen the foundations of good citizenship in a pluralist democracy. Finally, they assume that race can serve under these circumstances as a rough but serviceable proxy for both diversity value and value diversity.
In their widely discussed 1998 book, The Shape of the River, William Bowen and Derek Bok strongly defend these three assumptions. They conclude that race-neutral admissions would substantially reduce the number of interracial social interactions and hence the socialization skills that both white and black students value and contribute to by attending racially diverse institutions.
But this conclusion, which is also supported by a study done to bolster the University of Michigan's defense of its affirmative-action program in the courts, has been challenged from several directions. The first is empirical. A review of survey data shows that most students and faculty place little weight on ethnic diversity as a source of positive educational outcomes, and its regression analysis of peer-group racial-composition effects finds no positive effect on any of the 82 outcome variables used by the American Council on Education.
The second challenge is comparative. Stephan and Abigail Thernstrom, prominent critics of the Bowen-Bok study and authors of an earlier analysis of American race relations that strongly opposed affirmative action, point to a 1997 national survey in which 86% of white adults reported having black friends, and to a 1994 survey in which 73% said that they had "good friends" who were black. To the Thernstroms, the Bowen-Bok study findings justify a different inference: "By these standards, the elite schools are hardly in the proud vanguard of progress. To the contrary, they are lagging woefully behind." Interestingly, both the Bowen-Bok and Thernstrom data on college-age and adult friendships overlook the increasing prevalence of interracial friendships among young Americans even before they reach college, which would suggest that the college experience may be less central to engendering such friendships than either camp supposes.
A third challenge to Bowen-Bok examines how the education process actually works on campuses. Terrance Sandalow, in an important review of The Shape of the River in the Michigan Law Review, maintains that any experiential differences between white and black students "are simply irrelevant to most of what students study in the course of their undergraduate careers. The irrelevance of those differences is perhaps most obvious in the study of mathematics and the natural sciences, but it is no less true of most of the humanities and the social sciences." Sandalow goes on to consider the argument, so crucial to the diversity rationale, that African-Americans are likely to advance different ideas unfamiliar to whites:
[E]ven though the subjects I teach deal extensively with racial issues, I cannot recall an instance in which, for example, ideas were expressed by a black student that have not also been expressed by white students. Black students do, at times, call attention to the racial implications of issues that are not facially concerned with race, but white and Asian-American students are in my experience no less likely to do so.
For what it's worth, my 35 years of experience in the classroom confirm this account. Recall, as well, that admissions officers almost never ask applicants about their ideas or viewpoints, much less how salient their race is to them. Outside the classroom, of course, race-based differences might encourage empathy and tolerance, but they might also promote greater conflict. Indeed, the simplistic version of the diversity rationale masks a deeper confusion about the diversity value arising out of social interactions. In this view, diversity demonstrates to people that despite our superficial differences, we are really all alike under the skin. The proposition is clearly true in many respects, but the diversity value that the diversity rationale invokes is supposed to grow out of the decidedly different viewpoints that diverse people are said to bring to these interactions. If we take the rationale seriously, then our similarity under the skin may confer negative, not positive, diversity value. The very logic of this rationale, after all, dictates that we seek differences under the skin, since it is those differences that constitute the payoff to diversity.
These doubts about the "socialization skills" premise of affirmative action are fortified by the dismaying evidence of persistent racial self-isolation on campuses. Orlando Patterson, a leading sociologist of race and an affirmative-action supporter, ruefully notes that "no group of people now seems more committed to segregation than Afro-American students and young professionals."
After carefully interrogating the diversity rationale, then, one is left with serious doubts about its coherence and persuasiveness. There is something to it, surely, but not much. Some advocates, as if recognizing this problem, seek to reconceptualize the diversity rationale into something else. In philosopher Elizabeth Anderson's view, for example, diversity is really "another way of talking about integration," a way that can link diversity to the advocates' "core social justice and democratic concerns." In the same spirit, legal scholar Robert Post sees diversity as the seedbed of "a democratic public culture."
This discursive move, however, is really an effort to change the subject; it defends racial preferences not as a way to enrich the experiences of students and teachers but as a remedy for social inequalities and generalized discrimination. That the Supreme Court has repeatedly prohibited this general remedial justification for racial preferences, of course, is not a conclusive argument against it; the Court, after all, is notoriously fallible. But this diversity rationale, as we have seen, is weak even in its own terms.
THE COSTS OF AFFIRMATIVE ACTION
If the diversity benefits of affirmative action are rather dubious, its costs seem very steep indeed.
One measure of the costs of affirmative action is the longstanding opposition to it by a substantial majority of Americans. A Gallup poll published in July 2013 — after four decades of experience with such programs — found that 67% of all adults, 59% of Hispanics, and 44% of blacks agreed that "applicants should be admitted solely on the basis of merit, even if that results in few minority students being admitted." Only 28% of all adults, 31% of Hispanics, and 48% of blacks agreed that "an applicant's racial and ethnic background should be considered to help promote diversity on college campuses, even if that means admitting some minority students who otherwise would not be admitted."
One leading study of such attitudes found, consistent with other studies, that "the most fundamental factor behind opposition to affirmative action is one of principle." That is, the opponents view preferences, rightly or wrongly, as inconsistent with the ideals of equal opportunity and merit that almost all Americans strongly endorse. Researchers of public attitudes toward affirmative action understand that the phrasing of questions, as well as other contextual factors, can affect survey results and that multiple interpretations of the data are possible. But no researcher in this field doubts that the public remains decidedly and intensely opposed, pretty much regardless of how the questions are framed, the state of the economy, or the age, sex, financial conditions, or general political inclinations of those polled. Indeed, one must presume that if the public actually knew how immense the weighting of ethno-racial preferences in selective college admissions must be in order to implement these programs (as discussed below), its opposition to preferences would be that much greater. The real question about affirmative action, then, is not why so many Americans oppose it but why it has managed to survive and even expand in some cases.
Far more important, however, are the costs borne by affirmative action's own supposed beneficiaries. Sander and Taylor have provided the most exhaustive account of this tragic irony. They conclude that a key "mismatch" between students and schools largely explains "why, even though blacks are more likely to enter college than are whites with similar backgrounds, they will usually get much lower grades, rank toward the bottom of the class, and far more often drop out; why there are so few blacks and Hispanics with science and engineering degrees or with doctorates in any field; and why black law graduates fail bar exams at four times the white rate."
The most disturbing fact about these human costs, Sander and Taylor emphasize, is that they are wholly unnecessary: "[N]early all of these students do have what it takes to succeed" in the absence of affirmative-action preferences; they would just gain this success at the less demanding schools to which they could gain admission without preferences. They continue:
[T]he main victims of large racial preferences...are not the many whites and Asians who get passed over but rather the many blacks and Hispanics who receive preferences and do badly....Their intellectual confidence has been undermined, their career aspirations have in many cases been derailed, and they must deal with the stigma of being "affirmative-action admits." In the words of liberal scholar Christopher Jencks, "A policy that encourages the nation's future leaders to believe that blacks are slow learners will...do incalculable harm over the long run.
In fact, any declines in minority enrollments in the flagship campuses in state systems subject to bans of affirmative-action admissions tend to be offset by their attendance at less selective schools in the system (or elsewhere) — what Sander and Taylor call a "cascade effect." This outcome is hardly surprising, and a recent econometric study by Georgetown University's Peter Hinrichs and his colleagues confirms it empirically.
Indeed, Sander and Taylor's data show that the academic performance of minority students in the University of California system improved after Proposition 209 banned explicit affirmative-action policies in the state. Their four-year graduation rate rose 55%; the number who earned degrees in STEM fields rose 51%; the number who had GPAs of 3.5 or higher rose by 63%; and the number who earned doctorates and STEM graduate degrees rose by 25%.
Under any reasonable accounting, these immense gains under race-neutral admissions systems vastly outweigh whatever loss of diversity value has occurred at the most selective campuses, Berkeley and UCLA. Indeed, according to Sander, the overall level of integration of both blacks and Hispanics across the eight UC undergraduate campuses increased significantly after Proposition 209; that is, the distribution of both groups more closely approximated the distribution of whites across those campuses. Before the referendum passed, Berkeley and UCLA used racial preferences to attract blacks and Hispanics, and as a result half the blacks in the UC system had been concentrated on those two campuses. After the ban, when those schools could no longer officially employ such policies, this skewing largely disappeared.
A number of prominent scholars have strongly challenged the mismatch hypothesis, particularly as applied to legal education (the focus of Sander's original study), on a variety of methodological grounds. Sander has replied to his critics, and he and Taylor devoted a chapter of their book to reviewing these objections and defending their mismatch claims. It is difficult for a non-specialist to parse these arguments, much less assess them. Certain important facts, however, are perfectly clear even to a non-specialist.
First, the size of the preferences is enormous by any standard. Writing in 2009, Thomas Espenshade and Alexandra Radford reported that the admission "bonus" for being black was equivalent to 310 SAT points relative to whites and even more relative to Asians. The GPA differences are even greater than for SAT scores. Thomas Kane, a researcher in this field, found that black applicants to selective schools "enjoy an advantage equivalent to an increase of two-thirds of a point in [GPA] — on a four point scale — or [the equivalent of] 400 points on the SAT." Second, at every SAT score level, the test, which has long been criticized as being culturally biased against blacks, in fact over-predicts their actual academic performance in college.
In light of these two enduring facts, a substantial mismatch effect is more than a hypothesis. It would be little short of astonishing if no such effect existed.
AFFIRMATIVE HARM
The public opposition to race-based affirmative-action programs on campus is amply justified. Affirmative action defies — indeed flouts — equal protection and other liberal values. It rests upon a diversity rationale that is theoretically incoherent and in fact produces little if any of the diversity value that alone might justify it (and then only under a dubious rationale). It cannot satisfy the constitutional tests that the Court has laid down and reaffirmed as recently as last year. It has failed to increase its political support in the nation after four decades of energetic advocacy. It fosters corrosive racial stereotypes, poisons race relations, and encourages opacity, dissimulation, and even evasion by its administrators and advocates.
And if that were not enough, affirmative action seems to grievously harm many of its supposed beneficiaries — not to mention the non-preferred groups who are disadvantaged by the practice.
We are far from putting America's history of racial intolerance and injustice behind us, but affirmative action fails to rectify these evils and instead harms both our students and our society as a whole.
