Ask the author: Richard Sander and Stuart Taylor, Jr. on Mismatch
on Oct 16, 2012 at 9:39 am
On October 9, Basic Books released Mismatch: How Affirmative Action Hurts Students It’s Intended to Help, and Why Universities Won’t Admit It, a new book by Richard Sander, a law professor at UCLA, and Stuart Taylor, Jr., a legal journalist. In their book, Sander and Taylor argue that race-based admissions preferences for minority students lead to “mismatch” between students and universities: as a result of race-based preferences, students are admitted to more selective schools than they otherwise would be based on their academic credentials alone; once enrolled at these more prestigious schools, the students fall behind and are less likely to finish. This “mismatch” effect, the authors contend, then cascades down all tiers of higher education, with the effects becoming more pronounced at each successive tier. The authors have generously agreed to answer a few questions about the book for SCOTUSblog.
Question: It was once said of Keith Moon, the drummer for The Who, that although he did not have the technical prowess to be considered the greatest drummer of all time, he was the greatest Keith Moon-style drummer of all time. Moon’s somewhat sloppy drumming style “matched,” you might say, with The Who, which allowed him to be very successful. But if he had been a member of the Beatles instead, he might not have “matched” their style and been as successful. Intuitively, it makes sense that it’s better to be in a place where you can thrive, but that simply may not always be possible. If these kinds of “mismatches” exist in many aspects of life, why should there be a special burden on educational institutions to make sure that students match with their schools?
Answer: The special constitutional burden on state colleges and universities — imposed by the Supreme Court’s longstanding rule that state-sponsored racial classifications are presumptively unconstitutional — is that they may use racial preferences in admissions only if they can prove them to be “narrowly tailored” to serve a compelling interest. While the Court held in Grutter v. Bollinger that a diverse student body brings educational benefits enough to amount to a compelling interest, it also held that narrow tailoring means, among other things, that such preferences must not “unduly harm members of any racial group.” It is hard to overstate the relevance to that determination of the powerful evidence that large (though not necessarily small) preferences harm many or even most of the black and Hispanic supposed beneficiaries by subjecting them to the mismatch effect. Especially when combined with the fact that the larger the preferences, the more white and Asian applicants are harmed by race-based rejection, the mismatch evidence strongly suggests that large racial preferences unduly harm members of every racial group. They are thus unconstitutional as currently used by UT and virtually all other major universities.
To be sure, plaintiff Abigail Fisher did not put evidence of the size of UT’s preferences on the record in the case, and so far neither the litigants nor the courts have taken note of their size. But public university documents show that in 2009, for example, the mean SAT score of black UT freshman admitted outside the “top ten percent” law was a staggering 390 points (out of 2400) below the mean white score and an astonishing 467 points below the mean Asian score. The mean Hispanic score was 120 points below the mean white score, and there were similarly large racial gaps in high school GPAs. With gaps of this size, mismatch is bound to be a serious problem for the racially preferred at UT.
A related point is that there is absolutely no reason to think that qualities such as leadership, social skills, and even demonstrated ability to overcome adversity — which no doubt make some academically sub-par students valuable additions to student bodies as a “holistic” matter — correlate closely with race. So why do universities’ supposedly “holistic” admissions systems so consistently produce entering classes with large racial gaps in academic preparation, and so routinely admit preferred-minority students while rejecting Asians and whites who are not only less affluent but also better qualified? The only plausible answer is that “holistic” is mainly a cover for raw racial preferences, which are often given more weight than leadership skills, social graces, a history of overcoming adversity, and all other non-academic traits combined. It is also telling that in the “Personal Achievement Index” used in UT admissions, being born black or Hispanic is actually classified as an “achievement.” Being born Asian or white is, well, just being born.
Question: Part of your analysis depends upon making causal inferences from counterfactual claims. One of the examples you use to illustrate mismatch is that students who, as a result of race-based preferences, attend higher-tier law schools than they would otherwise have gotten into have lower bar-exam passage rates than students at lower-tier schools who did not receive any preferences and therefore had an easier time learning. You assume that if the mismatched students attended schools where they matched, they would have been more likely to pass the bar exam because they would have been better suited to learn there. One of the amicus briefs in Fisher, submitted by a group of empirical scholars, disagrees with your inference. The brief uses an example of two black female students with the same GPA, 3.8, and LSAT scores, 168: one student, after receiving race-based preferences, attends Yale (mismatch), while the other (who does not receive any preferences) attends a good but less elite school like the University of Alabama (match). The Yale student doesn’t pass the bar on her first try, but the Alabama student does. Although the failure to pass the bar could be attributed to mismatch, the amicus brief disputes the inference that the Yale student would have passed the bar on her first try if only she had gone to the University of Alabama. The brief argues that the mismatch inference fails to account for “unobserved differences” between the students, such as the Yale student possibly being predisposed to take the bar exam in a more difficult jurisdiction, like New York or California. What is wrong with the argument in that brief?
Answer: One of us (Sander) has been arguing since 2005 that the greatest challenge in comparing university outcomes is the difficulty of controlling for “unobserved differences.” The leading work in the field – Bowen & Bok’s 1998 book, The Shape of the River – completely ignored this problem, which severely biased most of their findings towards the conclusion that affirmative action programs had benign effects. It is telling that the “empirical scholars brief” never mentions that Bowen, Bok, and others whose research suggests an absence of mismatch tend to be those who ignore the problem of unobserved differences.
There are several ways of dealing with this problem; in the literature on law school mismatch, perhaps the best solution is what’s known as the “first choice/second choice” comparison. The largest available dataset on law school students and their outcomes includes survey data asking students about where they applied to law school, which schools accepted them, which school they chose, and why they chose it. About half of the black students in the study got into their “first-choice” school, and about eighty percent of those enrolled in that school; the other twenty percent enrolled in their “second-” or “third-choice” school, usually for financial or geographic reasons. “First-choice” law schools, for any given student, tended to be more elite than the second- or third-choice schools, and for black students, generally involved larger racial preferences.
This, then, provides a way of controlling for unobserved credentials. Black students who got into their first-choice law school, but enroll in their “second” or “third-choice,” are generally students who had strong enough academic characteristics to get into a more elite school, but ended up going to a less elite school – and one where they received a smaller racial preference – for reasons unrelated to school eliteness. Comparing these students with those who go to their “first-choice” school thus compares people who had similar skills, but attended schools with differing levels of mismatch. It’s worth pointing out that this exact method was first put forward by two of the signatories of the empirical scholars’ brief as a particularly good way to deal with the problem of unobservables.
The first-choice/second-choice results for law students are stunning: The students receiving smaller preferences (via attendance at second- and third-choice schools) have dramatically better outcomes – grades, graduation, and bar passage – which almost perfectly match the predictions of the mismatch hypothesis. Yet this evidence is never mentioned in the empirical scholars’ brief. Their brief came out after we finished our book, but it is another great example of the sort of intellectual chicanery by the defenders of the status quo that we document throughout the book.
Question: The “preference-based” affirmative action that you credit with creating mismatch situations for minorities was regarded as quite successful at bringing women into our education systems. Why is mismatch more of an issue for minorities than it was for women?
Answer: The premise of the question seems wrong: We have seen no evidence that universities ever used large gender preferences to admit more women, or that mismatch has ever been a problem for women. It was not gender-based preferences that brought large numbers of women into once all-male or mostly male colleges — where women have so thrived that they now substantially outnumber men — and professional schools. It was, rather, the ban on gender discrimination in the 1964 Civil Rights Act and other laws, plus the non-preferential recruitment and outreach programs that were what “affirmative action” originally meant. Equal opportunity was all that women needed, because they have long been at least as strong academically on average as men.
By contrast, the reason that after more than forty years of racial preferences universities cannot meet their racial enrollment targets without continuing to use large preferences — and thus creating mismatch problems — is the large racial gaps in academic achievement of students coming out of high school. Black/white gaps in K-12 testing narrowed greatly in the 1970s and early 1980s, but have been largely stuck in place for the past twenty years. In the most recent National Assessment of Educational Progress, the average black twelfth grader scored about as well as the average white eighth grader. Until the racial gaps in K-12 education are closed, admissions preferences will be at best a band-aid that diverts reformist energy and political capital from the far more important task of radically reforming K-12 schools. That’s the implicit message of some striking observations by Thomas Espenshade, a leading pro-preference scholar, in an October 4 New York Times op-ed. “Race-based affirmative action has been a woefully inadequate weapon in the arsenal against inequality,” wrote Espenshade, and a “Supreme Court ruling against the university might put ethnic and racial diversity on college campuses on a firmer footing for the long term. It would spur Americans who care about racial inequality to seek alternatives to affirmative action by addressing the deeply entrenched disadvantages that lower-income and minority children face from the beginning of life.”
Question: Putting the question of mismatch aside, you also argue that if the goal of our education system is equality of opportunity for all students, it would be much more effective to redirect the resources used to recruit minority students, many of whom come from affluent families, towards recruiting more economically disadvantaged students who are more academically prepared to thrive, and would therefore be a closer “match,” than some of the students who currently receive preferences. Rick, you were successful at UCLA in shifting the admissions focus from actively recruiting and funding minority students to recruiting and funding economically disadvantaged students of all races. This ended up working out very well for UCLA in terms of building a diverse class that also had a higher bar passage rate than previous classes that were admitted using “traditional” race-based preferences. What were some of the institutional hurdles that you had to overcome to get UCLA to shift the recruitment paradigm from race to class?
Answer: (from Rick) I worked closely on this with two UCLA colleagues, David Sklansky and Alison Anderson. The biggest hurdle we faced was that most of our colleagues were relatively apathetic about socioeconomic (“SES”) diversity. Despite the success of our experiment in bringing lots of disadvantaged students to the school who performed extremely well academically, my colleagues (as well as university officials and the media) focused almost entirely on whether SES preferences produced large racial dividends. They certainly produced some racial dividend – black and Hispanic applicants were two to four times more likely than whites with the same academic credentials to be selected – but since our SES preferences were smaller than the old racial preferences, black and Hispanic enrollment dropped significantly. And this was enough to persuade the faculty that it should scale back the SES program (from a target of two-fifths to one-fifth of the class) and experiment with other indirect methods of admitting minority students.
The moral for our broader national debate is that SES preferences work best if we value socioeconomic diversity for its own sake. Our fear is that colleges prevented from using racial preferences may use extremely large SES preferences just to increase their racial dividend; but that would produce much the same (and possibly even worse) mismatch problems as the current system. Part of the appeal of SES preferences is that relatively small preferences – especially if paired with better outreach programs to weak K-12 schools – can produce quite a bit of diversity. If colleges simultaneously improve academic support for entering students, and help low-SES students navigate the cultural transition to university environments, we can create a lot of social mobility and enrich higher education at the same time.
Question: In both the book and in your contribution to our Fisher symposium, you assert that some of the Court’s past decisions have put schools in a compromised position: they are expected to have diverse student bodies, but they are also expected not to use race overtly, which causes them to consider race but to hide the fact that they are doing so. This, you argue, has in turn had the unintended effect of creating a lack of transparency at two levels. At the student level, it comes from the failure to disclose to students who receive race-based preferences that they are less prepared than the majority of the entering class. At the institutional level, there is a lack of transparency in fairly communicating to schools how much campus diversity they are required to have.I have two questions about transparency. First, is there really a lack of transparency at the student level? That is, if most schools already make the average GPA and test scores for previous incoming classes publicly available, shouldn’t a student whose scores are lower than that realize that he is not as strong as the average student at the school and that he may struggle at the school, even though he was accepted? And, second, do you see any room in the Fisher case for the Court to address the transparency issue at the institutional level?
Answers: Preliminarily, schools are not “required to have” any particular level of diversity, as far as the law is concerned. The legal question is how aggressively schools can use racial preferences to reach whatever level of diversity they wish to have.
As to the first question: There is a gross and purposeful lack of transparency because the universities, many pro-preference scholars, the media, and others who know how things work systematically mislead minority students and almost everyone else into believing at least four big myths that just aren’t true: (1) that racial preferences in college admissions and racial gaps in SAT scores and high school grades among students entering individual colleges are not very large; (2) that SAT scores and other test scores don’t mean much anyway; (3) that preferred-minority admittees do fine academically; and (4) that they gain ground on their white and Asian classmates by the time they graduate. Each of these myths is contradicted by much empirical evidence that is undisputed among experts but widely denied and minimized by academics, journalists, and others with pro-preference agendas. The response of the universities in particular is sometimes to attack first-rate scholars who point to awkward facts, as the president of Duke University attacked distinguished members of his own faculty a few months ago in a revealing episode that we detail in our book.
As to the second question: We do have some small hope that the Justices may find in Fisher that narrow tailoring requires transparency, even though that question has not yet been raised directly by either the parties or the judges and justices. Chief Justice Roberts, among others, seemed especially frustrated during the October 10 argument about the amorphous idea that schools need an indispensable but unquantifiable “critical mass” of preferred-minority students to reach the promised land of diversity; this could imply a similar frustration about the lack of information as to how the racially preferential admissions process at UT (and everywhere else) really works. Indeed, standing back from the details of Fisher, it seems obvious that as long as the size, operation, and effects of racial-preference systems are secret, meaningful judicial policing of the principles that the Court has articulated to prevent overly aggressive use of racial preferences in Grutter and other decisions will be almost impossible. Transparency would also help preferred-minority students make more informed choices in choosing colleges and would help inform researchers, voters, and policymakers. It’s hard for us to imagine a good argument against it.
The other major reforms that we propose will almost surely have to await another case. We would require that a school’s racial preferences be no larger than its socioeconomic preferences, which we hope would both hold down the size of the preferences and give universities healthy incentives to end their shameful neglect of working class and poor kids. We would also ban race-based scholarships, which are mostly used in bidding wars among schools that neither increase diversity (in the aggregate) nor serve any other good purpose. But it’s hard to see the Court moving in those directions in a case that has gotten this far without any party, judge, or Justice mentioning either the size of the preferences or the mismatch problem.
