The Sad Irony of Affirmative Action


In 2003, the Supreme Court held that the University of Michigan's law school could substantially relax its admissions standards in order to admit a "critical mass" of African-American and Hispanic students. Many observers interpreted that decision — Grutter v. Bollinger — as an open-ended embrace of affirmative action.

The University of Texas was among the many universities emboldened to ramp up its use of race-preferential admissions policies. In 2003, the university already had in place an admissions policy designed to raise the number of under-represented minority students attending its flagship campus in Austin by admitting the "top 10%" of the graduates of each Texas high school without regard to SAT scores. Soon after the Grutter decision, however, the university announced that it was still dissatisfied with the diversity of the student body at Austin, 21% of which was composed of under-represented minorities (16.9% Hispanic and 4.5% African-American), and that the school would be implementing race preferences to boost that diversity. Under the new policy, the proportion of the student body composed of Hispanics and African-Americans rose to 25%.

The result was a lawsuit. The plaintiff — Abigail Fisher — is a young woman from Texas whose academic credentials were good, but not quite up to the standards that whites and Asians must meet in order to gain admission. They were, however, above those necessary for African-American and Hispanic students. Fisher, who is white, was rejected, and wound up attending the less prestigious and (for out-of-state students) more expensive Louisiana State University. Her case — Fisher v. University of Texas — was argued before the Supreme Court in October. It will be decided sometime in the coming months.

The Court may decide Fisher on narrow grounds. There are several dimensions along which the University of Texas's race-preferential admissions policies are more aggressive than those in Grutter. For example, Grutter permitted Michigan to use racially preferential admissions policies to admit a "critical mass" of African-Americans and Hispanics to its overall student body. Texas, however, takes the position that it needs "critical mass" not just in its student body as a whole, but in each classroom, program, and major. Under the "top 10%" policy, Texas had likely already achieved a "critical mass" of minorities across its student body. Classroom-level "critical mass," however, requires much more extensive preferences; it could conceivably justify racial discrimination in course registration and other more aggressive discriminatory practices.

Affirmative-action supporters worry, however, that the Court will take the opportunity to cut back severely on Grutter. They point to changes in the Court's personnel — most notably Justice Sandra Day O'Connor's replacement with Justice Samuel Alito — as cause for concern. Since Grutter was a 5-4 decision, it may not take much to swing the Court in the opposite direction.

The biggest change since Grutter, though, has nothing to do with Court membership. It is the mounting empirical evidence that race preferences are doing more harm than good — even for their supposed beneficiaries. If this evidence is correct, we now have fewer African-American physicians, scientists, and engineers than we would have had using race-neutral admissions policies. We have fewer college professors and lawyers, too. Put more bluntly, affirmative action has backfired.


How could such a miscalculation about the effects of affirmative action occur? As University of California, Los Angeles, law professor Richard Sander and legal journalist Stuart Taylor, Jr., describe in their important, recently released book, Mismatch: How Affirmative Action Hurts Students It's Intended to Help, and Why Universities Won't Admit It, one consequence of widespread race-preferential policies is that minority students tend to enroll in colleges and universities where their entering academic credentials put them toward the bottom of the class. While academically gifted under-represented minority students are hardly rare, there are not enough to satisfy the demand of top schools. When the most prestigious schools relax their admissions policies in order to admit more minority students, they start a chain reaction, resulting in a substantial credentials gap at nearly all selective schools.

For example, according to data released by the University of Texas in connection with Fisher, the mean SAT scores (out of 2400) and mean high-school grade-point averages (on a 4.0 scale) varied widely by race for the entering class of 2009. For Asians, the numbers were 1991 and 3.07; whites were at 1914 and 3.04; Hispanics at 1794 and 2.83; and African-Americans at 1524 and 2.57. The SAT scores for the Asian students placed them in the 93rd percentile of 2009 SAT-takers nationwide; the African-American students, meanwhile, were at the 52nd percentile.

 This has the predictable effect of lowering the college or professional-school grades the average minority student earns. And the reason is simple: While some students will outperform their entering credentials, just as some students will underperform theirs, most students perform in the range that their entering credentials suggest.

No serious supporter of race-preferential admissions denies this. In their highly influential defense of affirmative action, The Shape of the River: Long-Term Consequences of Considering Race in College and University Admissions (discussed later in more detail), former Ivy League university presidents William Bowen and Derek Bok candidly admitted that low college grades for affirmative-action beneficiaries present a "sobering picture." This is an understatement: The average African-American first-year law student has a grade-point average in the bottom 10% of his or her class. And while undergraduate GPAs for affirmative-action beneficiaries aren't quite as disappointing, that is in part because, as explained below, affirmative-action beneficiaries tend to shy away from subjects like science and engineering, which are graded on a tougher curve than other subjects.

One example that helps illustrate the consequences of mismatch — how lower entering academic credentials depress both academic performance and grades, and how lower-than-average academic performance and grades in turn harm professional ambitions — is the field of academia. In 2003, too late to be cited to the Court in Grutter, Stephen Cole and Elinor Barber published Increasing Faculty Diversity: The Occupational Choices of High-Achieving Minority Students. The authors' mission was to determine why more members of minority groups are not attracted to careers in the academy. The authors' conclusions, reached after extensively questioning 7,612 high-achieving undergraduates at 34 colleges and universities, pointed to race-preferential admissions as the culprit.

"It is a fact," Cole and Barber wrote, "that in virtually all selective schools...where racial preferences in admission is practiced, the majority of African American students end up in the lower quarter of their class." Lower grades sap the academic self-confidence of African-American students at elite schools, according to the authors, which in turn causes them to abandon their freshman interests in academic careers. Their counterparts at non-elite schools, on the other hand, are more likely to persist and to ultimately succeed. These counterparts enjoy school, in part because they correctly perceive that they are good at it, and they want to stay on campus to pursue careers in academia.

Cole and Barber found that the effect of grades on career ambitions was in fact substantial. The authors noted that among African-American students with GPAs at or near 2.6, only about 4% wanted to become college professors. Among those with GPAs at or near 4.0, however, the number was over 20%.

These findings build on long-established observations about the importance of grades and perceived achievement. Indeed, as early as 1966, University of Chicago sociologist James Davis published research demonstrating that a student who attends a school that is out of his academic league is often put at a professional disadvantage. In "The Campus as a Frog Pond: An Application of the Theory of Relative Deprivation to Career Decisions of College Men," Davis controlled for entering academic credentials and compared students at schools of different academic rank, examining their career choices to see which pursued "high performance" careers (in law, medicine, science, etc.). He found that college GPA correlated more strongly to career choice than did the academic rank of the school attended. He explained this finding in terms of the "theory of relative deprivation," under which students can be expected to measure their own potential in comparison to their immediate classmates, generally using one another's grades as "the accepted yardstick."

Davis put his conclusion in somewhat quaint terms. "Counselors and parents might well consider the drawbacks as well as the advantages of sending a boy to a ‘fine' college, if, when doing so, it is fairly certain he will end up in the bottom ranks of his graduating class," he wrote. Davis's research spawned a cottage industry in sociological studies on the hazards of being a "small frog" in a "big pond."

Further support for Cole and Barber's conclusion comes from an unexpected source: First Lady Michelle Obama's 1985 senior thesis at Princeton University, titled "Princeton-Educated Blacks and the Black Community." The future first lady mailed a questionnaire to 400 randomly selected black alumni; though the response rate was not overwhelming, the responses of the 89 black alumni who completed the questionnaire gave reason for concern. Black alumni were asked whether they felt "much more comfortable with Blacks," "much more comfortable with Whites," or "about equally comfortable with Blacks and Whites" in various contexts during three different periods in their lives — before attending Princeton, while students at Princeton, and after leaving Princeton.

Those who argue that race-preferential admissions foster integration might be surprised by Obama's findings. In the category of "Intellectual Comfort," the number of black alumni who said that they felt "much more comfortable with Blacks" than with whites in an intellectual setting went up upon attending Princeton. In their pre-Princeton years, 26% of the respondents were at greater intellectual ease with fellow blacks than with whites; during their Princeton years, however, the number climbed to 37%. This sense of alienation from white students did not appear in other categories of interaction: For "Sporting Comfort," the change was in the opposite direction (26% felt more comfortable with fellow blacks prior to Princeton, compared with 25% who felt more comfortable with fellow blacks while at Princeton). In the categories of "Dating Comfort" and "Business Comfort," the proportions of respondents who felt "much more comfortable with Blacks" were unchanged.

It is difficult to see how reducing the "Intellectual Comfort" that black students feel with whites can lead to greater black achievement. Yet this is just one of the many perverse effects of affirmative action and the academic mismatch it causes.


Minority students' lack of interest in academic careers offers one example of the consequences of mismatch, but the strongest evidence comes from the fields of science and engineering. Contrary to what many might expect, college-bound African-American and Hispanic students are just as interested as white students in majoring in science and engineering. Indeed, empirical studies show that they tend to be a little more so. But these are difficult majors that many students abandon. Significantly, African-American and Hispanic students jump ship at much higher rates than whites.

It is not surprising that students with lower entering academic credentials give up on their ambitions to get degrees in science and engineering more often than students with higher academic credentials. What some do find surprising is this: Three in-depth studies have demonstrated that part of the effect is relative. An aspiring science or engineering major who attends a school where his entering academic credentials put him in the middle or the top of his class is more likely to persevere, and ultimately to succeed, than an otherwise identical student attending a more elite school where those same credentials place him nearer to the bottom of his class. Put differently, a student's chances of success in science or engineering are increased not only if his entering credentials are high, but also if those credentials compare favorably with his classmates'.

The earliest of these studies — titled "The Role of Ethnicity in Choosing and Leaving Science in Highly Selective Institutions" — was published in 1996 by a team of scholars led by Dartmouth psychologist Rogers Elliott. It found that the single most important cause for minority attrition from science at the selective institutions studied was the "relatively low preparation of black aspirants to science in these schools." The authors were careful to use the word "relatively." It wasn't just entering credentials demonstrating highly developed ability at science that mattered, but comparatively high credentials. A student who attended a school at which his math SAT score was in the top third of his class was much more likely to follow through with an ambition to earn a degree in science or engineering than was a student with the same score who attended a school at which that score was in the bottom third of the class. The problem for minority students was that, as a result of affirmative action, being in the top third of the class was relatively rare.

Elliott and his co-authors cited the extraordinary record of historically black colleges and universities, which graduate far more than their share of black engineering and science majors, as further support for their findings. Unlike at other colleges and universities, credentials gaps are not an issue at the historically black institutions. As one faculty member at a historically black school — North Carolina Central University's Walter Pattillo, Jr. — told Science magazine in 1992: "The way we see it, the majority schools are wasting large numbers of good students. They have black students with admissions statistics [that are] very high, tops. But these students wind up majoring in sociology or recreation or get wiped out altogether."

A more recent study by University of Virginia psychologists Frederick Smyth and John McArdle (now at the University of Southern California) confirmed Elliott's findings. And the effects were not subtle. In "Ethnic and Gender Differences in Science Graduation at Selective Colleges with Implications for Admissions Policy and College Choice," Smyth and McArdle found that, among a sample of under-represented minority students at 23 universities who intended to major in science, mathematics, or engineering, 45% more of the women and 35% more of the men would have succeeded in attaining their goals if they had attended schools where their entering credentials had been about average.

Another study — this one by Richard Sander, co-author of Mismatch, and UCLA statistician Roger Bolus — pulled data from nine University of California campuses. The authors came to a similar conclusion. "Minority attrition in science is a very real problem," they wrote, "and the evidence in this paper suggests that ‘negative mismatch' probably plays a role in it." Their multiple approaches to the data yielded consistent results: "[S]tudents with credentials more than one standard deviation below their science peers at college are about half as likely to end up with science bachelor degrees, compared with similar students attending schools where their credentials are much closer to, or above, the mean credentials of their peers."

The evidence that mismatch has hurt African-American and Hispanic students' chances of having careers in science or engineering was highlighted in a report of the U.S. Commission on Civil Rights in 2010. The data and methodology of the research have not been challenged. The researchers' conclusions have not been rebutted. Nevertheless, the findings have been ignored by colleges and universities. Indeed, one of the arguments that the University of Texas makes before the Supreme Court in the Fisher case is that there are not enough minority students studying science and engineering to make those classrooms racially diverse. As a result, it claims, greater race preferences in admissions are needed. But Texas's race-preferential admissions will likely aggravate rather than alleviate this problem. The more colleges and universities engage in preferential treatment, the fewer the African-Americans and Hispanics who will graduate with degrees in science and engineering.

And the evidence keeps piling up. Recently, Duke University economists Peter Arcidiacono and Esteban Aucejo and Duke sociologist Ken Spenner found evidence supporting the mismatch thesis when researching the major choices of undergraduates enrolled at Duke. In their article in the IZA Journal of Labor Economics, "What Happens After Enrollment? An Analysis of the Time Path of Racial Differences in GPA and Major Choice," they found that black undergraduates were much less likely to persist with an entering goal of majoring in engineering, the natural sciences, or economics than white students were. Approximately 54% of black males switched out of these majors, while only 8% of white males did. Once again, the problem was not lack of interest in science and engineering among black students: Indeed, before starting at Duke, more black students than whites indicated an initial interest in majoring in these subjects. Instead, the differences in attrition were best explained by entering academic credentials.

These authors also dispelled the common belief that affirmative action beneficiaries "catch up" after their freshman years with their better-credentialed fellow students. What happens instead is that many transfer to majors where the academic competition is less intense and where students are graded on a more lenient curve. Their GPAs increase, but their standing relative to their peer groups does not.

This effect is by no means confined to affirmative-action beneficiaries. White children and grandchildren of alumni who receive legacy preferences have the same experience, earning lower grades than white non-legacies at the end of their first year. While the gap narrows over time, it is only because legacy students, too, shift away from the natural sciences, engineering, and economics and toward the humanities and social sciences. It is exceedingly unlikely that anti-legacy bias, lack of legacy role models on the faculty, or any other argument commonly advanced to explain racial disparities in science explains the legacies' collective drift toward softer majors. If it is the wrong explanation for legacies, it is overwhelmingly likely to be the wrong explanation for under-represented minorities, too.

The study created a firestorm at Duke. Unfortunately, the administration, instead of taking the research to heart, focused on pacifying indignant students, alumni, and faculty members who were insulted by the results. In an open letter to the campus responding to demands that the university condemn the study, provost Peter Lange and other administrators stated that they "understand how the conclusions of the research paper can be interpreted in ways that reinforce negative stereotypes." They assured students that there are no easy fields of study at Duke and took the position that, insofar as the mammoth problem identified in the study exists, it could easily be solved through student counseling and a few tweaks to the science curriculum.

Evidently, business will remain as usual at Duke. Potential affirmative-action recruits with an interest in science and engineering will continue to be told that Duke is the school for them. They will not be told that their chances of success in their chosen fields would be greater at the University of North Carolina at Chapel Hill. Nor will they be told that if they switch majors to disciplines like African and African-American Studies, Art History, English, Sociology, and Women's Studies, they are less likely to enjoy lucrative careers or indeed to get jobs at all. In securities law, this would qualify as actionable fraud. In higher education, it is considered forward thinking.


The problem of relative performance and credential mismatch does not end with college graduation. It extends to professional schools as well, and is particularly evident at America's law schools. Shortly after Cole and Barber's book was published, Mismatch co-author Richard Sander published a study of law schools titled "A Systemic Analysis of Affirmative Action in American Law Schools." His findings were similar. Outside of historically black colleges and universities, up and down the law-school hierarchy, the average African-American student had an academic index — a combination of GPA and LSAT score — more than two standard deviations below that of his average white classmate. Indeed, at some law schools, there was no overlap between the entering credentials of African-American students and those of white students (Sander did not study Hispanic students). These gaps in entering credentials affect student performance: Sander's research demonstrated that more than half of African-American law students had first-year GPAs in the bottom 10% of their classes. Even critics of Sander's ultimate conclusions agreed that these findings were both true and troubling.

Only slightly more controversial was Sander's finding that this effect was almost entirely the result of affirmative action. When African-American and white law students with similar entering credentials competed against one another, they performed very close to the same. Race-based admissions were thus creating the illusion that African-Americans are somehow destined to be poor law students. The truth is that, if they were attending schools where their credentials matched the average student's, they would be just as likely to do well.

Strangely, however, African-American and white students with identical entering credentials were not performing similarly on the bar exam. Sander showed that the likely reason is that they are not attending the same schools. The African-American students were more likely to be at law schools that are more theoretical in their approach and where "teaching to the bar exam" is considered déclassé. Rather than benefiting from the more competitive learning environment these schools offer, African-American students were falling behind their white academic counterparts who were attending somewhat less competitive schools. Sander's critics, on the other hand, had no explanation for why white students perform better on the bar exam than African-American students with identical credentials.

Under Sander's calculations, if law schools were to use race-neutral admissions policies, fewer African-American students would be admitted to law schools. But since those who were admitted would be attending schools where they were very likely to do well, fewer would fail or drop out. In the end, more would pass the bar exam on their first try (1,896 versus 1,567 successful African-American first-time test takers among the graduating class of 2004) and more would eventually pass the bar (2,150 versus 1,981 among that same class) than under current admissions practices.

Sander's research was criticized by proponents of race-preferential admissions on the ground that it was just one study, and Sander agreed that more research would be desirable. He used the best and most recent data available at the time, and his calculations have been verified by others, but surely confirming the results with a different and more recent database would have been useful. In a report issued in 2007, the U.S. Commission on Civil Rights urged grant-making agencies to fund research into this issue and requested that state bar associations cooperate with this research.

Unfortunately, something closer to the opposite has happened. In order to confirm his initial findings, Sander assembled an ideologically diverse team of investigators and sought data from the State Bar of California. Urged not to cooperate by some of the very same people who had previously complained that Sander needed more evidence, the state bar denied the team access. It didn't matter that Gerald Reynolds, chairman of the U.S. Commission on Civil Rights, flew to San Francisco to ask personally for the state bar's cooperation. It didn't matter that the data had been cheerfully shared with other researchers. The California bar wanted no part of this important research. A court battle is now underway.

Meanwhile, Sander and University of Arizona law professor Jane Yakowitz Bambauer have taken to examining one the most dearly held beliefs of affirmative-action advocates — that enrolling in the most prestigious school one can get into is the key to success. This premise, central to affirmative action, turns out to be false: In predicting future income, getting good grades in law school matters more than getting into a top law school. And as Sander and Bambauer demonstrate in "The Secret of My Success: How Status, Eliteness and School Performance Shape Legal Careers," this is true for law students generally, not just under-represented minorities.

Put differently, aspiring lawyers who tear their hair out to get into the most prestigious law school possible — figuring they can just cruise to a law degree once they get to campus — are making a mistake. They need to be putting at least as much effort into excelling once they are in school. If students at Harvard don't work hard, their professional stars may be eclipsed by lawyers with similar entering credentials who attended lesser law schools and made better grades.

Again and again, the results are the same, no matter what the area of study: Attending a highly competitive school is a good thing. But so is getting good grades. Indeed, getting good grades is somewhat more important than attending a prestigious school. A public policy that ensures that African-American and Hispanic students will disproportionately attend schools where their grades are likely to be worse than their classmates' thus works to the minority students' disadvantage.


To be sure, those who wish to ignore the mismatch literature have been given a convenient excuse to do so: the influential 1998 book defending affirmative action, The Shape of the River: Long-Term Consequences of Considering Race in College and University Admissions, written by Bowen and Bok. Their book calculated that even black men with combined SAT scores of less than 1000 — low for elite schools — who attend top-tier schools like Princeton out-earn similarly credentialed students who attend schools like Pennsylvania State.

The book received an astonishing level of attention when it was published. Fawning editorials appeared in many newspapers. The New York Times announced that it "flatly refute[d]" the arguments of critics of race-preferential admissions. Newsweek's Ellis Cose commented that the book was the "most ambitious study to date of the effects of affirmative action in higher education" and "an important corrective to conservative propaganda." Some of the commentary specifically addressed the issue of mismatch: Harvard University sociologist Nathan Glazer argued in the Washington Post that it was now "clear" that worries over mismatch were misplaced. The Pittsburgh Post-Gazette editorialized that the notion that race-based admissions policies have hurt African-American students "is one that can be dismissed."

For many reasons, however, the methodology used in The Shape of the River is seriously flawed. For example, Bowen and Bok took account only of SAT scores, overlooking other academic credentials like high-school rank. One cannot assume that a student at Princeton with a given SAT score is the equivalent of a student with the same score at Penn State. There is an excellent chance that the first student has a substantially better high-school GPA or other distinctions in his favor; that is one reason he is at Princeton instead of Penn State. Indeed, Ivy League presidents like Bowen and Bok are constantly making this point themselves: Their schools reject many applicants with stratospheric SAT scores in favor of applicants they believe show greater academic promise in other ways.

Even using flawed methodology, however, it is difficult to avoid the evidence of affirmative action's failure. The Shape of the River's own figures show that black men with SAT scores between 1000 and 1099 and black women with SAT scores between 1100 and 1199 are likely to earn more if they stay away from the most elite schools.

Why might that be? Buried in book's appendices is a more sophisticated analysis that attempts to explain how various factors influence the subsequent earnings of black graduates of selective colleges or universities. Each such factor's effect was measured, including (to a limited extent) a student's high-school rank and whether his college grades put him at the top, middle, or bottom third of his class. The authors purport to show that attending a school like Princeton rather than a school like Penn State adds to the income of black students. They appear oblivious, however, to the stunning point made by their own figures, throughout the different permutations of their analysis: College grades generally contribute more.

Imagine two black males with identical SAT scores, both in the top 10% of their high-school classes, and both from middle-class families. Only their colleges are different. Bowen and Bok convincingly demonstrate that if the two have the same college major and similar grades, the one who attends Princeton will earn considerably more than the one who attends Penn State.

But what if they don't have similar grades? By the authors' own calculations, it is better to be a black male at Penn State in the top third of the class than in the bottom third at Princeton. The increased earnings the Penn State student gets from high grades are worth almost twice the increased earnings from attending Princeton. And the boost in earnings he would get from majoring in natural science rather than the humanities — a more achievable goal at Penn State — is a whopping $49,537 per year.

If one's class rank and major were unrelated to the selectivity level of one's college, then it would be perfectly sensible for the authors to celebrate the finding that, all other things being equal, black males get an earnings boost from attending Princeton rather than Penn State. But they are not unrelated. For students who would not have been admitted but for racial preferences, the chances of being in the top third of the class are remote.

The only question is whether a student who attends Princeton and winds up in the bottom third of the class would likely have been in the top third of Penn State. And the answer to that question, at least in many cases, is yes. Consider, for example, a black male with combined math and verbal SAT scores of 1300 (out of a possible 1600) who just missed being in the top 10% of his high-school class. If he attends Penn State, his SAT scores will put him exactly at the 75th percentile in the 2011 entering class (using figures from U.S. News & World Report). That would give him an excellent shot at earning grades in the top third of his class, or graduating with a natural-science degree, or both. If this student instead enrolls at Princeton, however, his SAT scores will put him 110 points below the 25th percentile for that school, likely making his academic standing very tenuous. If he wants to maximize his earnings upon graduation, the choice is obvious.

How could Bowen and Bok have missed the import of their own research? The answer may lie partly in the fact that the book was rushed to press in 1998 just two months before Election Day. On the ballot that year was Washington State's Initiative 200, a clone of California's Proposition 209, which prohibited race-preferential admissions policies in state colleges and universities. Supporters of race-preferential admissions hoped that The Shape of the River would change voters' minds about the desirability of such prohibitions.

Initiative 200 passed anyway, but The Shape of the River slowed the momentum of state popular initiatives in this area. Perhaps more important, The Shape of the River was cited by and seems to have heavily influenced Justice Sandra Day O'Connor in her opinion for the majority in Grutter. The book and its influence thus point to the troubling implications of using social-science research in constitutional analysis, particularly on the subject of race.


Brown v. Board of Education may be the most important Supreme Court decision in the area of race in the past century. In arriving at its conclusion in 1954 that "separate but equal" school systems are inherently unequal, the Court relied in part on the now-famous doll experiments of Kenneth and Mamie Clark, intended to test the self-perception of young African-American children brought up in the Jim Crow South. The Clarks showed the children two dolls that were identical except for skin and hair color: One doll represented a blonde white person and the other a black person. When asked which of the two dolls was the nice one, which looked bad, which was the more attractive color, and which was more appealing to play with, the African-American children showed a consistent preference for the white doll.

Constitutional scholars look back at the doll experiments and ask, "What if the children had preferred the black doll?" What if it turned out that the children's preference for the white doll had nothing to do with low self-esteem caused by Jim Crow segregation? Would that have made the case for Brown v. Board of Education weaker? Should the constitutional right to equal protection turn on the latest social-science research?

The answer to these latter questions should be, "Of course not." The Constitution demands equal protection for all persons regardless of whether they can demonstrate through social-science research that they have been harmed by some law or policy or social practice. The Clarks' doll experiments were certainly interesting; given the uncertainties of litigation, the attorneys for the Brown plaintiffs were wise to bring them to the attention of the Court. But the Court probably should not have given the impression that its constitutional analysis might be shaped by the results of such an experiment.

If Brown should not have relied on the doll experiments, does that mean the Court should not take social-science research into account in rendering decisions in litigation over race-preferential admissions policies? Some have suggested as much, arguing that the research showing the harm done by race-preferential admissions should be off-limits.

As it happens, though, the Court has already taken social-science research into account — and in Grutter, it almost certainly took bad social-science research into account. In concluding that race-preferential admissions policies were beneficial to minority students, and that the Court should therefore make an exception to the otherwise overwhelming presumption against racially discriminatory laws and policies, Justice O'Connor's citation to The Shape of the River was explicit. But even without such a citation, it is clear that the Court's decision was premised on a belief that race-preferential admissions were helping, or at least not hurting, African-American and Hispanic students.

Of course, under Grutter, increased campus diversity was said to benefit all students, not just under-represented minorities. Consequently, racial discrimination to obtain that benefit was deemed permissible. But minority students are not public utilities; their futures should not be sacrificed to serve broader goals of social engineering. And it is difficult to imagine a college or university knowingly employing race-preferential admissions to give white and Asian students an advantage at the expense of African-American and Hispanic students. The Grutter decision thus would have been unthinkable in the absence of a strong conviction by the Court that affirmative action was providing minority students with a substantial advantage, not a disadvantage.

Now it is becoming evident that it was all a mistake. The strong constitutional presumption against race discrimination in all its forms, which must be firm and unchanging to be effective, was laid aside for no good reason.

To compare this to Brown and the doll experiments, one would have to imagine that Brown had come out the other way — in favor of racially segregated schools — because the Court had some reason to believe that Jim Crow was benefiting all students. If later, more sophisticated research had exposed that belief as erroneous, it would be incumbent upon the Supreme Court to return to the principle that race discrimination should not be tolerated.

It remains to be seen what the Court will do in Fisher. It seems unlikely that its decision will cite or discuss the mismatch literature, and that is as it should be. But that does not mean that this body of research will not, or should not, affect the Court's thinking. The mismatch literature is showing Grutter to be a well-meaning but ultimately misguided deviation from what otherwise had become accepted principle — that race discrimination should not be tolerated. Perhaps in the future, the Court will not be so flexible with its principles.

Gail Heriot is a professor of law at the University of San Diego and a member of the U.S. Commission on Civil Rights.