Reposted from Science  magazine, 24 October 2003
In June of 2003, a majority of the U.S. Supreme Court ruled that the University of Michigan Law School could consider an applicant's race in making admission decisions ( 1, 2). The court's decision was driven largely by the fact that, given the current distribution of academic performance among U.S. high school seniors, selective universities would admit very few African American or Latino children without taking race or ethnicity into account. The decision was tailored to accommodate universities' use of race in admission decisions, while limiting the impact outside of higher education. In this Policy Forum, I describe empirical realities underlying the debate and issues likely to arise in future legal challenges.
The debate over the use of race in admission decisions has been wrenching, because it demands a trade-off among three worthwhile goals: race-blindness, academic selectivity, and a semblance of racial diversity on selective campuses. A few justices did not find the trade-off sufficiently compelling to outweigh the equal protection clause in the Fourteenth Amendment. Rather than requiring an institution to reduce the number of African American and Latino students admitted, Justices C. Thomas and A. Scalia pointed out that a university could also reduce its academic selectivity to accommodate a race-neutral policy. Justice Thomas asked, if operating a public university law school is such a compelling state interest, why do a number of states including Alaska, Delaware, Massachusetts, New Hampshire, and Rhode Island--choose not to do so? Moreover, he noted, even fewer states choose to operate highly selective public law schools. Such concerns notwithstanding, a majority of the court found the public benefits generated by race-conscious policies sufficiently compelling to allow continued use of race in admissions.
Basing Admission on Class Rank
In what seems to have been a risky legal argument, the Bush Administration tried to deny the existence of the trade-off itself, arguing that even highly selective institutions could achieve racial diversity by race-neutral means, simply by granting automatic admission to students in the top of their high school class ( 3). They pointed to the experience of Texas, Florida, and California, which have substituted admissions based on high school class rank for race-conscious policies (although the extent of their success has been disputed) ( 4- 6), as evidence that there are workable alternatives.
But such policies rely on segregated schools, and not all states have highly segregated school systems. In 25 of the 48 states for which data were available, fewer than 10% of African American seniors attended high schools containing more than 90% African American or Latino youth. Latino students are typically less segregated: In 37 of 48 states, fewer than 10% of Latino youth attended high schools with more than 90% African American or Latino enrollment. Perhaps not surprisingly, the states that have substituted rank-based policies for race-conscious admissions--California, Texas, and Florida--are among the handful of states that have large numbers of both African American and Latino youth attending segregated schools ( 7, 8).
Basing Admission on Low-Income Status
As another way to avoid the trade-off between race-blind policies and student diversity, some have suggested race-neutral "class-based" admission policies--targeted at low-income and disadvantaged youth ( 9- 11). But, however worthwhile such policies may be, they will do little to produce racial diversity on selective college campuses. In 1992, among the high-scoring high school seniors (those with test scores in the top tenth of the class), black and Hispanic youth were three times as likely to be from families with incomes less than $20,000 than white and other non-Hispanic youth ( 12) (see figure, below). However, black and Latino youth still represented only one out of six high-scoring, low-income youth--17%. Because black and Hispanic youth represented only 7% of the top decile of test-takers, they represented a minority of most subgroups of applicants, even low-income applicants. As a result, selective colleges and universities would have to admit six times as many students under an income-based policy to yield the same number of black and Hispanic youth as would result from an explicitly race-based policy. Preferences based on economic disadvantage offer a very indirect means for achieving racial diversity ( 12, 13).
In a separate case involving undergraduate admissions at the University of Michigan, the court ruled that the college's mechanical point system, which granted a prespecified number of points on the basis of race, was not legitimate ( 14). In other words, although universities can consider race as part of a complete reading of an applicant's file, it cannot grant an automatic, prespecified number of points based on race. The distinction is somewhat elusive. The justices were clearly hoping that a more careful reading of each file would lead universities to consider a wider range of each individual's skills and to tailor the weights given to each characteristic for each individual. However, the decision provides no clear description of the factors universities must consider, the weights they can attach to them, nor how courts should determine whether institutions are indeed tailoring their decisions to individual files.
The inflexibility of the undergraduate point system led a majority of the court to rule against it. However, one strength of that system, emphasized by Justices R. B. Ginsburg and D. H. Souter, was its transparency. When a university's admission policy consists of the aggregation of subjective decisions, it is much more difficult to know precisely what the policy is and whether it passes the constitutional test.
The Concept of "Critical Mass"
In its previous decisions, the court had prohibited the use of race simply for the purpose of racial balancing ( 15- 17). In Grutter v. Bollinger, the court recognized the special pedagogical role of higher education institutions and allowed universities to use race when pursuing "the educational benefits that flow from a diverse student body." The primary beneficiaries are not supposed to be the minority students alone, but the whole class ( 18).
Under such a rationale, the challenge is to define how much diversity is "enough" to produce the educational benefits universities seek. The University of Michigan argued that to fully reap the pedagogical benefits of diversity, they needed a sufficient number of students from each group, to ensure that students felt comfortable expressing themselves honestly to their classmates. This concept of "critical mass" is understandably nebulous. To spell out a specific percentage would have invited charges that it was a surreptitious quota. However, the concept of critical mass will need to be clarified further to withstand future challenges. For instance, between 1995 and 2000, the University of Michigan Law School admitted a class that was 8, 4, and 1% African American, Hispanic, and Native American, respectively. If each of these groups constituted a critical mass, it is not clear why the critical mass required for African American youth was so much larger than the critical mass required for Latino or Native American youth. The question is important, because the university would have been able to achieve 1% African American and Latino enrollment without considering race.
The Handicapped Parking Analogy
As complicated as the legal issues may be, the political issues surrounding race-conscious admission policies are even more treacherous. Handicapped parking provides a useful analogy ( 12). Suppose that there were one parking space reserved for disabled drivers in front of a popular restaurant. Eliminating the reserved space would have only a minuscule effect on the parking options for nondisabled drivers. But the sight of the open space may frustrate many passing nondisabled motorists looking for someplace to park.
With the uncertainties surrounding university admissions, it is difficult to identify which individuals are paying the cost of race-conscious admissions ( 12, 19). In the Spring of 2003, Harvard College accepted only one applicant in 10 ( 20). Many of the rejected applicants (and, potentially, many more of those who did not bother applying) have better grades and SAT scores than many of the minority applicants who are admitted. A large fraction of these may well believe that they would have been accepted if Harvard had no racial preferences. Yet only about 18% of Harvard's undergraduates are black or Hispanic. Even in the unlikely scenario that ending racial preferences forced all these students to surrender their seats to white and Asian-American students, acceptance rates for the remaining students would only increase from 10 to 12%. If more than 2% of those who were originally denied admission believe that they were the "next in line" and that they would have been admitted in the absence of racial preferences, then the perceived costs will overstate the true costs.
Ironically, the more informal use of race used by the University of Michigan Law School could exacerbate such misperceptions. With a mechanical, point-based system, those who are harmed by race-based policies are more readily identifiable. In the less explicit system endorsed by the court, the perceived costs may be less intense, but more widespread, since it would be more unclear who was the next person in line.
Is 25 Years Likely to Be Long Enough?
Although not imposing an explicit time-limit, the majority in Grutter v. Bollinger expressed an aspiration that "25 years from now, the use of racial preferences will no longer be necessary to further the interest approved today." Given the legal rationale they used for endorsing race-based decisions, the court's desire for a deadline is somewhat puzzling. Deadlines are traditionally sought in discrimination cases, when the victims of past discrimination are no longer available to receive the remedy. But when racial considerations are based on the pedagogical value of diversity, a deadline makes less sense.
Is 25 years a realistic goal for closing the racial gap in test scores? The justices seem to have chosen the 25-year time period for no good reason other than that it has been 25 years since they last took up the issue ( 15). In 1978, the gap in math scores between African American and white 13-year-olds on the National Assessment of Educational Progress was about 1.1 standard deviations ( 21). By 1986, the gap had shrunk to about 0.6 standard deviations. However, by 1999, the gap in mean scores by race had expanded to about 0.8 standard deviations. The reversal of progress makes the court's 25-year goal look overoptimistic, although recent progress achieved in a few states provides hints that more rapid progress may be possible.
The federal No Child Left Behind Act is designed to encourage all states to close the racial gap in performance. Whether or not efforts to close the racial gap will succeed remains to be seen. However, school districts around the country have no time to spare. Those who will be applying to college 25 years from now will be entering school in 12 years.
References and Notes
U.S. Commission on Civil Rights, "The U.S. Department of Education's 'Race-neutral alternatives in postsecondary education: Innovative approaches to diversity' Are they viable substitutes for affirmative action?" (May 2003). J. R. Campbell, C. M. Hornbo, J. Mazzeo, "NAEP trends in academic progress: Three decades of student performance" (NCES 2000-469, U.S. Department of Education, Washington, DC, 2000).
U.S. Commission on Civil Rights, "The U.S. Department of Education's 'Race-neutral alternatives in postsecondary education: Innovative approaches to diversity' Are they viable substitutes for affirmative action?" (May 2003).
J. R. Campbell, C. M. Hornbo, J. Mazzeo, "NAEP trends in academic progress: Three decades of student performance" (NCES 2000-469, U.S. Department of Education, Washington, DC, 2000).
The author is at the School of Public Policy and Social Research and Department of Economics, UCLA, Los Angeles, CA 90095-1656, USA. E-mail: firstname.lastname@example.org