Richard Sander is Professor of Law at UCLA School of Law.
Why did the Supreme Court decide to take Fisher v. University of Texas at Austin back for a second look? One intriguing possibility can be inferred from Joan Biskupic’s 2014 biography of Sonia Sotomayor. According to Biskupic, Justice Anthony Kennedy initially drafted a far-reaching opinion for Fisher I, supported by the Court’s four conservatives, that would have come close to eliminating the use of race in college and university admissions. During the spring of 2013, however, Justice Stephen Breyer approached Kennedy with word that Justice Sotomayor had prepared a blistering dissent, which Breyer thought would create national and intra-court wounds that might be difficult to heal. At Breyer’s urging, Kennedy took a different tack; instead of rolling back the scope of preferences permitted by Grutter v. Bollinger (the 2003 decision that had been the Court’s most recent word on racial admissions preferences), his revised Fisher I opinion merely emphasized the restrictive aspects of Grutter. Justice Sandra Day O’Connor’s opinion in Grutter has been notorious among commentators for the gap between its rhetoric (holding that preferences must be narrowly tailored, that a person’s race cannot trigger an automatic preference, and that the consideration of race should be phased out over time) and its favorable ruling for the University of Michigan Law School (which used very large racial preferences that were highly mechanical and showed no signs of ever ending). By emphasizing that standards like “narrow tailoring” should now be taken seriously, and that the Fifth Circuit needed to hold the University of Texas to these standards, Kennedy was able to maintain his perfect record of ruling against affirmative action, while still commanding a seven-to-one majority that included both Breyer and Sotomayor.
The story is plausible, since the mildness of the Court’s decision, and Sotomayor’s decision to join the majority, surprised most observers at the time. But if true, things did not work out as Kennedy hoped. A year later, in the spring of 2014, Justice Kennedy wrote another opinion about affirmative action, holding (for a six-to-two majority that again included Justice Breyer) that a voter initiative which prohibited the state of Michigan from using racial preferences did not violate the Fourteenth Amendment. Justice Sotomayor wrote a widely discussed, incendiary dissent – plausibly a reworked version of her unused Fisher I draft that harshly criticized Kennedy’s opinion. Then, a few months later, the Fifth Circuit, considering Fisher on remand, issued an opinion that was highly deferential to claims by the University of Texas (with little or no supporting evidence) that it had exhausted efforts to find a race-neutral alternative to racial admissions preferences. Under these circumstances, one should not have been surprised that the Court granted cert. for Fisher II.
What might Kennedy, and the Court, do in this round? It is possible, but very unlikely, that the Court’s four conservatives granted cert. without Kennedy’s buy-in, simply to put him on the hot seat again. If so, anything could happen – and it is a stretch to think that Chief Justice John Roberts and Justice Samuel Alito would take such a risk. So let us rephrase the question: if Kennedy was one of those voting for cert. (highly likely), and ends up writing the Court’s majority opinion (likely as well), what will that opinion look like? The university will lose, and the Fifth Circuit will be reversed again. But what, substantively, would the Court say?
The most likely outcome, unfortunately, is that the Court will simply reiterate Fisher I with somewhat more forceful language, explain why the Fifth Circuit did not follow its instructions (see the excellent dissent by Judge Garza) and tell it to try again. This would be the Court’s path of least resistance, but it would leave quite up for grabs just what, exactly, Fisher I meant. Alternatively, the Court might decide that since the university has not produced, in two rounds, evidence that comes close to meeting the university’s burden under Fisher I, its future use of racial preferences is enjoined. This would give Fisher I a harder edge, but it, too, would do little to clarify just what other universities are supposed to do.
A third alternative for the Court would be to engage in a serious effort to define the muddled terms at the heart of both Grutter and Fisher I. This would be a way for Kennedy to meaningfully restrict the use of racial preferences without eliminating them altogether, which seems to be where he is, jurisprudentially, on this issue. It would also enhance the Court’s legitimacy – many universities now openly flout Grutter and Fisher I because they are regarded as so mushy – and create more healthy and open dialogs in higher education about race (universities now try to hide what they do as much as possible). What might it mean to define these key terms? Let’s consider a few of the key concepts raised in Fisher I:
“Critical mass.” Universities are permitted to seek a “critical mass” of underrepresented minorities in pursuing the educational benefits of a diverse campus. But the University of Texas has been unwilling to specify just what this critical mass might be, because, justifiably, it is concerned that any number it specifies will be condemned as an illegal quota. Court doctrine has thus created an impossible bind for schools, but there is a way to cut this Gordian knot. Universities should be required to define the “educational benefits” they seek, and then conduct research to show how the benefits vary with different admissions policies and the resulting student compositions. Using actually admitted students, and not hypothetical research subjects, is important because one of the effects of using large racial preferences is to create an academic distance in the classroom (and on campus) that correlates with race; as many studies have shown, this academic distance directly undermines the goal of interaction across racial lines. If such studies are nonetheless able to document any benefits from using racial preferences to increase campus diversity, there will presumably be some point at which the marginal benefits from increasing diversity fall to zero (partly because a higher critical mass implies even larger preferences). “Critical mass” would be the point where the measured, demonstrable benefits of additional diversity are still rising at a substantial rate; universities that demonstrate this level would be permitted to aim for this goal – which leads us to the elements of “narrowly tailoring” their pursuit of that goal.
A “good-faith consideration of race-neutral alternatives” (perhaps the key phrase in Fisher I) could easily be made more concrete by explaining the sort of research a university needs to undertake. As the Court notes, no university should be required to demonstrate that it has exhausted every possible alternative to using race. But it does seem reasonable to require a serious evaluation of more than one, and the list should surely include socioeconomic alternatives. SES, unlike race, is an individual characteristic, not an immutable group characteristic; low-SES students are far more underrepresented at selective American universities than are racial minorities, and – partly for this reason – low-SES students should be able to make a substantial contribution to the “diversity” environment on campus. Several articles have explained the sort of simulations universities can undertake to compare the “diversity yield” from various combinations of SES and racial preferences. Universities should then tie these results into their analysis of the “benefits of diversity” by comparing the benefits generated by students who are more socioeconomically diverse *and* are admitted with smaller racial preferences.
The “consideration of the diversity contribution of each student” is another Court doctrine on racial preferences that badly requires deeper articulation. Its intent is to prevent universities from treating race mechanically. But the result has been a shift to “holistic” admissions which are often so opaque (even to the administrators who are supposed to oversee them) as to prevent any accountability. The Court should remedy this by stipulating that a university’s use of racial preferences will be granted deference only if its admissions process is sufficiently transparent so that applicants (and scholars) can meaningfully identify the marginal impact of race on admissions decisions. This requires that universities using preferences make available anonymized data on applicants, including the weights assigned to “subjective” characteristics of the applicants, so that one can use straightforward statistical tests to evaluate the marginal effect of race on decisions.
None of these steps are draconian: indeed, by articulating the sort of research and transparency standards universities need to follow to use racial preferences, the Court will make it easier for universities to compare “best practices” and reduce the current taboos and secrecy that pervade most current use and discussion in the field. If the Court chooses the route of detailing standards, rather than indulging in lofty rhetoric, it can make Fisher II a genuine advance over Fisher I.