The [NAS'] brief reiterated the critique of the Gurin study offered in the Wood and Sherman Report that the NAS had commissioned--i.e., that Gurin fails to link numeric diversity to positive outcomes, that diversity experiences are irrelevant to the cases, that Gurin uses subjective metrics to assess outcomes, and that her effect sizes are too small to be useful.
--Justin Pidot, "Intuition or proof: the social science justification for the diversity rationale in Grutter v. Bollinger and Gratz v. Bollinger," Stanford Law Review, Dec 1, 2006
My previous contribution to this series, "The marriage of affirmative action and transformative education," identified the pedagogical and ideological roots of Res Life programs like the one at the University of Delaware that recently became the center of a firestorm of negative publicity and criticism. That earlier contribution also pointed out that proponents of programs like the one at U Delaware are engaged in an ambitious research effort, lavishly funded by foundations like Hewlett and Ford, that is designed to provide social science support for such programs and for racial preferences in admissions.
In future postings, I will be looking at some of the more recent research in the field, much of which has been inspired by the IGR program at U Michigan-Ann Arbor. Before delving into more recent research, however, it will behoove us to review the state of such research at the time that the Supreme Court decided the Grutter and Gratz cases.
It will come as a surprise to many that the claim that campus racial diversity has educational benefits was hardly proven when these cases were decided. To put it this way, however, seriously understates how weak their case was. For it wasn’t just that to date, no evidence had been found to support their policies. The much more serious problem was that very good research had been conducted that met all the norms of solid social science research - and this research had actually reached the conclusion that the hypothesis that racial diversity has educational benefits must be rejected. That research was based on the data on which the University of Michigan principally relied in its social science defense of its preferential admissions policies. The study in question--the only one then, and so far as I know, the only one now--that had sufficient resources to address the question, was conducted by the Higher Education Research Institute at UCLA. The HERI project itself was a spin-off of a research project that was begun in 1966 by the American Council on Education.
I learned about the HERI study when I was cutting my teeth on the preferences issue shortly after I had started my work as the executive director of the California Association of Scholars. Beginning almost on the day I was hired, the association got embroiled in what was called the "WASC wars," so-called because the Western Association of Schools and Colleges had adopted an agenda of coercing member institutions into using racial preferences to create more racial diversity in their student bodies. Even then, that agenda was premised on the claim that racial diversity has educational benefits.
At the time, WASC thought it sufficient to simply quote an article by Alexander Astin on the matter in Change magazine, in which Astin had touted the benefits of multicultural education. No evidence supporting the claim of educational benefits was given in that article. Instead, reference was made to an earlier study by HERI that Astin had written about in a book that had been published by Jossey-Bass called What matters in college?
Intrigued--and skeptical--I bought the book. It was a good thing I did, because I was apparently the only person in the entire world who read the book and noted that the study did test for the hypothesis that campus racial diversity has educational benefits, and had found (at the 5% confidence level) that the hypothesis had to be rejected.
I then wrote to Eric Dey, who at the time (1991) was the associate director of HERI, with questions about the study and in particular about the CIRP database that Astin and HERI had been using. (Dey is presently an Associate Professor in the Center for the Study of Higher and Postsecondary Education at the University of Michigan School of Education, Ann Arbor). Dey answered my questions in a straightforward, perfectly civil letter. He also passed my letter along to Alexander Astin, who sent his own reply. It was not sent to me, but was written and sent instead to Seymour Martin Lipset, whose name was given on the CAS letterhead as a member of our board of advisers. Astin was offended, even incredulous, that Lipset, who was at the time a highly respected senior scholar of the Progressive Policy Institute of the Democratic Leadership Council and America’s most eminent political sociologist, would be affiliated with an organization that dared to attack higher education’s sacred cow. The tone of his letter was sharp and indignant, but what could Astin really say substantively? The evidence was there, and Astin himself had found it and had been intellectually honest enough to report it. So far as I know, Lipset never even bothered to respond to Astin’s letter.
I wrote at some length about the CIRP/HERI study in a policy paper the CAS issued during the WASC wars. I believe what I wrote had some effect in those battles, but no one, at HERI or elsewhere, ever challenged our claim (which was really unchallengeable) that their own research rejected the claim that campus racial diversity has educational benefits.
At an early date, when the Grutter-Gratz litigation had already started working its way through the courts, I learned that Patricia Gurin at the University of Michigan was working on a research report to justify her university's use of racial preferences in admission. When I saw her report, I was astonished that the main part of her argument was based on the very same CIRP database, and for the very same years (1985-89) that Alexander Astin had used in his study, What matters in college? In a way, this was not surprising, because the CIRP database was the only database for higher education at the time that could adequately address the question whether racial diversity has educational benefits—as in fact I believe it continues to be to this day. (Two other studies that Gurin used in her report, which were confined to her own campus, failed to meet the norms and requirements for research design and statistical methodology in fundamental ways.) The basic problem for Gurin was, as I have pointed out, that the HERI research had reached the conclusion that campus racial diversity is not correlated with any of the 82 cognitive and non-cognitive variables used in the study to assess educational outcomes.
I was commissioned by the NAS to write a critique of the Gurin Report, which I did with Malcolm Sherman. Called Race and Higher Education: Why Justice Powell’s Diversity Rationale for Racial Preferences in Higher Education Must Be Rejected, it was issued in 2001, two years before the Supreme Court's decisions in the cases. Unfortunately, it fell on deaf ears, or rather, got no hearing at all. The NAS held a press conference in Washington, D.C., and made a major effort to get press coverage for the report, but failed.
Much later, Gurin and some of her allies did try to defend her research against our criticism, but their cause was hopeless. It was so hopeless, in fact, that Gurin herself did not even attempt to defend the policy that the University of Michigan was defending in the courts--that campus racial diversity has educational benefits. Gurin later denied ever having shared its claims. The claim that she and her allies defended instead was that "diversity experiences"--by which she meant such things as ethnic studies workshops and the facilitated peer group dialogues that are at the core of many Res Life programs--have educational benefits.
This outrageous effort to simply change the subject was, I regret to say, largely successful as a public relations strategy. It is one of the scandals of the debate over the use of racial preferences in higher education that Gurin and her allies were mostly successful in this gambit, despite the best efforts of the NAS and the Center for Equal Opportunity to squelch it.
As we pointed out over and over again, Gurin's entire approach was confused (or if one prefers, intellectually dishonest) for many reasons, of which I will mention very briefly only the following:
(1) The question she chose to defend was not even before the courts. The Fourteenth Amendment does not bar universities from holding ethnic studies workshops or peer-facilitated group sessions on the subject of race, and no one had ever suggested that it did. The question before the court was whether the Amendment bars the use of racial preferences in university admissions.
(2) As we showed in Race and Higher Education, Gurin's attempt to establish the educational benefits of diversity experiences was also fundamentally flawed, apart from its being irrelevant:
- She was unable to show consistent effects for her “diversity experiences” in direction or across groups. Over one-third of her results revealed significant negative effects. Whatever positive effects she reported were ludicrously weak.
- n the case of classroom-based diversity experiences (like ethnic studies workshops), Gurin failed to show that any of the outcomes--positive, negative, or neutral--were due to racial diversity in the classroom, because she was unable to distinguish in her study between curricular effects and whatever effects racial diversity in the classroom situation might have had. We do not know whether the effects of her classroom based experience variables--weak, inconsistent, and not invariant across groups as they were--would have been any different if there had been no racial diversity in the classrooms.
- Her research on her "diversity experiences" was weak--even fundamentally flawed—because it was not based on any controlled experiments. Students were not randomly assigned to her "diversity experience" courses, so she had no way of showing that her effects were not due to sampling bias.
I have already pointed out that the Gurin Report was unable to show consistent effects for the alleged educational benefits either in direction or across groups, and that many of her findings show significant negative effects. Anyone who takes the trouble to look at the tables Malcolm and I provided for her diversity experience results in our Race and Higher Education will be able to get an immediate and powerful visual impression of the truth of this claim. See, for example, Tables 1-5 that are on pages 91-96 of our report (pages 97-102 in your web browser).
Clearly, only a fanatic could think that anything of any real significance is going on here. Besides the large number of outcome variables that show no effect whatsoever, all of the positive and negative results are exceedingly feeble. It is really quite extraordinary that Gurin and her allies were audacious enough to cite such findings as evidence supporting the case for racial discrimination (a.k.a. racial preferences) in college admissions.
Malcolm and I hammered away at the weakness of Gurin's cited effects in the section entitled "The Relationships That Gurin Claims Are Statistically Significant Are Exceedingly Weak," which begins on p. 87 of Race and Higher Education (p. 93 of your web browser).
To get a better idea of just how weak the reported effects are, let us take a particular example: listening ability. (As it turns out, Gurin found a positive effect for this variable, but the point about weakness would be equally valid for any of her effects, either positive or negative.) Here is what we said about the alleged (and also entirely self-reported) listening ability effects of Gurin's "diversity experiences" in our report (pp. 97-98 of the text, pp. 103-04 in your browser):
Suppose that a university decides to stop taking race into consideration in its admission practices, and that the fraction of non-white-Asian students falls by 50 percent from 10 percent to 5 percent. What impact would this have on the (self reported) white student listening ability outcome? Gurin’s results predict that white student listening ability would decline at this institution by 5 x 0.145 = 0.725 percent, or about seven-tenths of one percent.
Thus, about 88 percent of Gurin’s models show that there is either no relationship, or a negative relationship, between structural diversity and both learning and democracy outcomes. In those few cases where structural diversity does have a positive impact on learning or democracy outcome variables, its impact is extremely small, as illustrated by the example above.
It might be helpful at this point to stand back from the details of social science research and simply appeal to our intuition. Ask yourself: Why would anyone expect that racial diversity in itself would have any significant impact on educational outcomes, all other things being held equal?
It is a mystery to me why anyone would believe the two would be significantly correlated. Take Sandra Day O'Connor, who seems to have had a fairly strong intuition about this, as an example. Suppose someone had said to Justice O'Connor when she was a student: "Okay, here's the deal, Sandy. You have a choice between going to law school at the University of Arizona and Stanford Law." For the purposes of argument only (I have no idea whether this was true then or is true now), let us suppose that UA Law was more racially diverse than Stanford. Suppose also that it was true then, as it is now, that Stanford Law has a higher ranking and a better reputation as a law school than UA Law. How likely is that O’Connor would have chosen UA over Stanford? Not very likely, I should think. Her likely preference for Stanford over UA might have had something to do with the question of prestige, and the belief that a law degree from Stanford would advance her career more than one from UA. Even so, the point would still go to show that racial diversity, even though she might have regarded it as a plus factor, wouldn’t have mattered all that much in her choice. Furthermore, prestige need not have been the only consideration. Because Stanford was more selective then, as it is now, she might reasonably have expected that her legal education at Stanford would have been better. The faculty was more distinguished, her fellow students would have been brighter, and she could very well have concluded that she would have been stretched and challenged at Stanford in ways that she would not have been at UA Law.
We should not allow ourselves to be transfixed by the details of regression coefficients, T-statistics, and all the other details of social science research design and statistical methodology. If we do that, we run the risk of losing sight of the forest for the trees. To counter this risk, we should ask ourselves the same kind of question as the one above. When I ask myself that sort of question, I find that racial diversity is positive, but only given some very important provisos and to a very limited degree. That is, in choosing between university A and B for myself, I would likely favor the one with more racial and ethnic diversity (let’s say that it’s A), primarily because I think that life might be more interesting on a campus like that. However, that would be true ceteris paribus--e.g., only if A were as selective and as good as B. I would also be disinclined to go to a more racially diverse campus like A if I knew the racial diversity had been engineered there by the use of racial preferences—partly because I am opposed to racial preferences, and partly because I would suspect that other things were not being held equal at a university that was using them.
Additionally, I would only have a preference for A if I thought that A’s racial diversity was also associated with significant differences in student backgrounds and experiences. I would see no advantage whatever in choosing A over B if the black, Asian, or Hispanic students at A were the same as the ones at B in all respects other than their skin color or race or ethnicity.
I am aware that the provisos I have included here are very hard to fulfill, and may not apply to any universities in the real world. Astin's HERI study, for example, found no correlation between racial diversity and student satisfaction with the colleges they attended. I am also aware that there is a difference between a good college education and one that is "interesting." I do not suppose that racial diversity in the classroom has anything to do with learning science or math, or any other subject. So even on a campus where there is greater real diversity (correlated somehow or other with the otherwise irrelevant racial and ethic diversity), I might not be getting as good an education even though in some sense the experience was more "interesting." Hence the importance of the proviso of holding all other things (like selectivity) equal when making the choice.
It is particularly important to appeal to common sense and our intuitions when the issue of diversity is raised in the constitutional context. Here it is not just a matter of asking whether campus racial diversity has some educational benefits, however marginal, but whether it has enough benefits to justify the use of racial preferences in admission despite the Fourteenth Amendment's bar against racial discrimination. I find it incredible that anyone's intuitions on the matter are strong enough to meet the strict scrutiny test of the Constitution.
Defendants in court have the burden of showing that there is more than just some level of benefit for racial diversity at, say, the 5% confidence level. Their burden is to show that the benefits are large enough to justify otherwise constitutionally impermissible racial preferences. Any failure to hold them to this exacting constitutional standard permits them to reason fallaciously as follows:
(1) Higher education is itself a compelling state interest--right?
(2) Campus racial diversity has at least marginal educational benefits.
(3) Therefore preferences to achieve campus racial diversity, if necessary to achieve sufficient levels of racial diversity (critical mass), also meet the compelling state interest test.
Of course, proponents never really felt comfortable making this argument, and in fact tried to argue for something much stronger, as in
(2)' Campus racial diversity is essential if colleges and universities are to offer quality education to their students.
despite the clear evidence marshaled in Race and Higher Education and other critiques that even (2) is false.
Sometimes diversiphiles appear to be arguing that any program X that makes a university better, however marginally, meets the compelling state interest test itself. But this cannot be true if strict scrutiny is to be truly strict--and it must be kept strict, in this area as in all other areas of the law. If courts allowed the proponents of racial preferences to reach the conclusion (3) above using (2) rather than (2)', the result would entail a serious weakening of the compelling state interest test.
It is really very tedious to have to go over all this territory again, since it should have been thoroughly aired and established in the public discourse at the time of the Grutter-Gratz litigation, when the NAS report was disseminated. So I was very gratified the other day to come across an article published in 2006 (i.e., 3-1/2 years after the Grutter-Gratz decisions) in the Stanford Law Review. Entitled "Intuition or proof: the social science justification for the diversity rationale in Grutter v. Bollinger and Gratz v. Bollinger," it was authored by Justin Pidot, a law student at Stanford at the time who is now a practicing attorney at a major law firm. Pidot argues that whatever persuasiveness U Michigan’s research had in court necessarily relied on an appeal to our intuition rather than proof--in other words, that the social science research that U Michigan and others cited in court simply failed to make the case. Pidot cites the NAS’ Race and Higher Education study and the NAS' amicus brief that was based on it approvingly, and at great length. Pidot's paper is particularly good at placing the social science arguments in the proper legal context.
Fortunately, it has been made available by The Gale Group at http://www.accessmylibrary.com/coms2/summary_0286-29469076_ITM, and registration is free. (I accessed it easily through the use of my local library card.) Pidot’s paper is worth reading in its entirety, keeping in mind the major points made above:
(1) The University of Michigan simply dismissed as unimportant or irrelevant the question before the courts--whether campus racial diversity has educational benefits.
(2) The results it reported for the quite unrelated question it did address--the alleged educational benefits of "diversity experiences"--were extremely weak and not consistent in direction or across groups. On this entirely unrelated matter, Michigan had nothing of any significance to report.
(3) On the question that was before the court, the only available social science research that had addressed the question found no correlation between "structural diversity" and any cognitive or non-cognitive educational outcomes, and this finding was based on the very database on which Michigan principally relied in its own study. Thus, the HERI/CIRP study did ask the relevant question, and instructed us to reject the hypothesis that campus racial diversity is correlated with educational benefits.
Sloppy reasoning and tendentious argument do not defeat themselves in court, as the example of the U.S. Supreme Court’s decisions in Grutter and Gratz shows. Hence the importance of doing a full court press on the research, much of it inspired by IGR at the University of Michigan, that advocates of racial preferences are marshalling in preparation for the next legal showdown. A careful examination of that research will be the subject of some forthcoming installments to the "How Many Delawares?" series.
Racial preferences have become an addiction to those who continue to advocate their use. A case can be made for them, but the temptation to use them needs to be resisted. It turns out that the case for racial preferences is weak, while the case against them is overwhelming.
The courts have rejected every rationale for preferences except the argument based on their alleged educational benefits. There are strong a priori reasons for thinking that research will never be able to support the claim that there are such educational benefits--to the extent, anyway, that would be required to overcome the legal and constitutional objections to their use.
Nor does existing research support the claim that there are educational benefits to racial diversity on campus. In fact, the very database on which the University of Michigan principally relied in Grutter and Gratz tells us to just say No to racial preferences.
Intuition or proof: the social science justification for the diversity rationale in Grutter v. Bollinger and Gratz v. Bollinger.
Publication: Stanford Law Review
Publication Date: 01-DEC-06
Author: Pidot, Justin
COPYRIGHT 2006 Stanford Law School