SINCE I WRITE a lot about affirmative action, a number of my friends told me about Randall Kennedy’s new book even before it appeared, urging me to read it. Kennedy, a professor at Harvard Law School and an African American, has long been a leading moderate voice on a variety of racial issues, so some anticipated that he would have a valuable, balanced take on the various debates surrounding racial preferences. In the weeks after its publication, however, I was surprised at the wide range of reactions it provoked among my moderate friends. Some said it was honest and thoughtful, while others thought it was quite incompetent at critical junctures. Having now read the book, I think both camps were right. Kennedy set out to write a balanced defense of affirmative action, but his resulting work is often so thin and sloppy that he ends up confusing the debate more than advancing it.
Most of Kennedy’s book focuses on affirmative action in higher education admissions — that is to say, the use of racial admissions preferences to increase the number of underrepresented minorities — blacks, Latinos, and Native Americans — at selective universities. This may not be the most common form of affirmative action, but it is the one that seems to generate the most controversy and, in recent years, the most attention from the Supreme Court.
The honest and thoughtful parts of the book arise from Kennedy’s generally sensible, nonideological approach to race. Many defenders of affirmative action tend to minimize the actual scale of preferences, portraying them as essentially tiebreakers; Kennedy acknowledges that admissions preferences are often quite large (equivalent to many hundreds of SAT points). Defenders often claim or imply that “disadvantaged” students make up most of those benefited by admissions preferences; Kennedy candidly concedes that most blacks who receive preferences into elite schools come themselves from fairly comfortable backgrounds. But for Kennedy, discrimination in favor of minority elites is justified by the importance of the goal: integrating American leadership. According to Kennedy, leaders come from top schools, and having blacks and Latinos in leadership positions is vital to making sure that attention is paid to social inequalities and problems that disproportionately affect racial minorities. A leg up for the Talented Tenth enables them to effectively help the other 90 percent.
This is an interesting argument, but when it comes to really making his case, Kennedy becomes vague. Indeed, Kennedy provides no hard evidence whatsoever that helping black or Latino elites to become still more elite produces trickle-down benefits for the rest of the black and Latino populations. Kennedy here compares poorly with Eugene Robinson, an African-American writer and Pulitzer Prize–winning journalist, whose 2010 book, Disintegration: The Splintering of Black America, argued persuasively, and with significantly more evidence, that affirmative action for minority elites was distracting attention from the much broader and deeper problems faced by the disadvantaged in American society. Racial preferences in higher education, Robinson concluded, was costing blacks a lot of political capital needed to formulate broader, race-neutral programs to attack social inequality. Kennedy considers, but ultimately sets aside, this argument.
To be fair, much of Kennedy’s book discusses legal issues, such as whether the 14th Amendment really bars all racial classifications or just those that have an invidious purpose. On such questions, Kennedy brings impressive historical material and legal arguments to bear. Even here, however, his book is handicapped by the fact that Kennedy wrote it before the Supreme Court’s most recent case on this issue — Fisher v. University of Texas — since the book was in production before the Court handed down its decision. Kennedy’s book thus neither contributed to the broad public debate last year on Fisher, nor is it able to incorporate its lessons.
But the crucial parts of For Discrimination deal with social policy, and the driving theme of the book is Kennedy’s seemingly even-handed consideration of the good and bad sides of affirmative action. The reader is encouraged when Kennedy observes that affirmative action must be reformed where its costs exceed its benefits, but dismayed when Kennedy never gives an example of an actual program he thinks requires this reform. As Kennedy gingerly sifts through the evidence on crucial questions, he writes with shocking superficiality and, often, with blatant inaccuracy.
I can illustrate this best by looking closely at Kennedy’s discussion of a crucial alleged harm of affirmative action — the “mismatch” issue. “Mismatch,” or “peer effects” research, examines students who are placed in environments where they have weaker academic preparation than their classmates — for example, a student who has the credentials to get into the University of North Carolina on her credentials alone, but through an admissions preference (which could be racial or could be based on alumni connections or any number of other factors) is admitted to Williams or Colgate. The “mismatch hypothesis” is that such students are likely to learn less, or be more likely to experience other harms, than they would at a school where their preparation level or credentials were typical of their classmates. A particularly convincing example of this phenomenon is “science mismatch”; several studies have documented that when promising high school seniors who aspire to science careers receive a preference (often, but not necessarily, racial) into an elite school, they find themselves in freshman physics and chemistry courses that are being taught at a level unfamiliar to them, in competition with significantly better-prepared students. As a result, these students leave science and technology (the so-called “STEM” fields) in droves, and only about 10-12 percent of the mismatched students ever obtain a STEM degree — a rate that would double, triple, or even quadruple had they attended a less elite but still very good school where their academic preparation was closer to the class median.
Among experts in the field, there now seems to be a strong consensus that science mismatch is a serious problem. Four major studies, each one carefully done and published in (or, in one case, under final review at) highly respected peer-reviewed journals, and each one using different data sources and methodologies, have all found powerful evidence of science mismatch (Elliott, Strenta, et al, 1996; Smyth and McArdle, 2004; Arcidiacono, Aucejo, and Spenner, 2012; Arcidiacono, Aucejo, Coate, and Hotz, 2013). I am aware of no peer-reviewed study that has either critiqued any of these findings or, in comparing students with differing individual levels of mismatch, has demonstrated that mismatch does not lower science attrition rates. Science mismatch is a huge part of why, even though college-bound black high school seniors have a higher level of interest in STEM fields than their white counterparts, they are only one-seventh as likely to eventually achieve a PhD in a STEM field. In other words, the unthinking use of racial preferences is, in this context, doing substantial harm to the “talented tenth” — keeping them out of the elite rather than boosting them in. An increasing number of experts involved in programs on the minority pipeline to science are taking the problem very seriously.
Yet in his extended discussion of the mismatch hypothesis, Kennedy never mentions science mismatch and, indeed, seems entirely unaware of the literature on the subject. Nor does he seem to be aware of “academic” mismatch — a well documented phenomenon carefully established in a study by two distinguished sociologists, Stephen Cole and Elinor Barber, and funded by the Council of Ivy League Presidents. Cole and Barber found that students who aspire to academic careers, but receive large preferences and thus experience mismatch in college, tend to receive lower grades and are far more likely to abandon their plans for an academic career than are otherwise similar students who are less mismatched.
Kennedy also ignores the growing, unrefuted research on “social” mismatch, which finds that the academic preparation and performance of students powerfully shape the social networks they form at college. Because of social mismatch, students who receive large preferences (whatever their race) can find themselves having more difficulty forming lasting friendships on campus. If preferences are correlated with race (as they strongly are at most elite schools), then the academic effects of those preferences create more social isolation among black and Latino students. Thus, a study of students at Duke University found that black freshmen often initially form friendships with whites that wither as their academic paths diverge; these black students end up as seniors having no more interracial friendships than they did at their (often segregated) high schools.
Each of these problems go to the heart of the issues about which Kennedy purports to care but has not bothered to learn anything about. Kennedy’s discussion of mismatch is limited to just one topic: the phenomenon of “law school mismatch,” where I have been a central figure. I published an article in the Stanford Law Review in the winter of 2004-’05 that provided the first detailed portrait of how racial preferences operated in American law schools, and in which I argued that there were strong circumstantial reasons to believe that large preferences — through the mechanism of mismatch — hurt the learning of blacks in law school, and explained about half of the very large gap between white and black bar passage rates.
My article provoked intense ideological fury in legal academia. This was partly because I disclosed uncomfortable facts that had never before been aired in public (such as details on the poor grades most students admitted through affirmative action earned in law school), and partly because many legal academics have a poor understanding of empirical work and tend to see “facts” as simply another tool to use in adversarial battle. This is why, as Kennedy points out, many critics dismissed my claim, advanced at the time and often since, that I am in fact someone who cares deeply about social justice and someone quite sympathetic to the goals of affirmative action. As Kennedy notes, I was instead subjected to a fair amount of ad hominem attack, and a number of leaders in legal education went to considerable (and fairly successful) efforts to prevent other data about legal education and bar outcomes from becoming public or even available to researchers. (Readers can find a fuller account of these developments in chapters 5 and 15 of my book with Stuart Taylor, Jr., Mismatch.)
But the sorriest part of the “law school mismatch” battle was the decision of several leading empirical scholars to write purportedly “empirical” but in fact highly misleading critiques. My article had been aimed at a broad audience, and I tried to use relatively simple methods to illustrate my points. Critics like Richard Lempert, Ian Ayres, Richard Brooks, and Katherine Barnes argued that my analyses were methodologically simple (this was true), and that better, more sophisticated analyses produced opposite results (this was false). None of these critiques held up. When I sent Lempert a detailed, point-by-point rebuttal of a critique he released in November 2004, he withdrew it from circulation. When some colleagues and I contacted the editors of the journal where Barnes published her results, she confessed that she could not replicate her own results and published an apology. The results published by Ayres and Brooks, it turned out, were incomplete and included inappropriate assumptions; when these problems were corrected, their results fit closely the predictions of the mismatch hypothesis. There are now four published peer-reviewed studies corroborating various aspects of the law school mismatch argument, and no peer-reviewed studies refuting it (see Williams, 2013; Williams, Sander, et al, 2011; Sander and Bambauer, 2012; Sander and Taylor, 2012).
Kennedy chides the tone and tactics of the critics; he writes, for example, “The adoption of an excessively pugnacious attitude by affirmative action defenders is a big mistake. It facilitates the denial of important facts that warrant attention and discourages the valuable habit of reconsidering established practices that need reform.” This is commendable and fair-minded. But Kennedy has not himself paid close attention to the important facts — he seems not to have read any of the literature closely, and he credulously accepts claims made by the critics that have been decisively refuted. Let me list some specific examples:
- Kennedy writes that “Ian Ayres and Richard Brooks note an aspect of the situation that Sander largely ignores — that, at least sometimes, ‘when students are overmatched by their classmates, they appear to be carried along to more success.’” Kennedy is apparently unaware that both I and other writers have reanalyzed in detail the evidence Ayres and Brooks rely upon for this claim, find their analysis fatally flawed, have invited them (without response) to rebut these corrected findings, and have published our results (Sander, 2011; Williams, 2013).
- Kennedy writes that “Professor David Wilkins complains that Sander underestimates the full panoply of benefits that come with attending a higher-tier [law] school even at the cost of lower grades,” and quotes part of Wilkins’s argument, “The network effects of elite schools are so obvious that one would think that they would have to be a large part of any examination of the benefits of attending such institutions. Yet Sander ignores them entirely.” But as Wilkins knew, and Kennedy should know, a significant part of my original Stanford article was devoted to testing the career effects of racial preferences on black lawyers, and I found no evidence of a net benefit. Wilkins himself provided only anecdotal evidence and crude comparisons to support his thesis, and in a detailed reply I published in the same issue in which Wilkins’s critique appeared, I offered substantial further evidence that the lower grades that come with large law school preferences more than outweigh the eliteness benefits (Sander, 2005). And last fall, I published (with Professor Jane Bambauer, an empiricist at the University of Arizona), in the Journal of Empirical Legal Studies, an even more detailed study, drawing on a half-dozen independent data sources, and again strongly refuting the claim that lawyers with elite degrees, but poor performance, are better off than comparable peers at less elite schools (Sander and Bambauer, 2012).
- Kennedy endorses the concern that, without racial preferences, the decline in black enrollment at elite schools might be so dramatic that it would discourage blacks from applying to law school at all. He cites a study which itself had zero hard evidence of this “discouragement” thesis. But Kennedy completely ignores the real-world data about what happens when racial preferences end, such as the evidence from California, where voters passed a ban on racial preferences in 1996. This issue was studied by two eminent economists, David Card (a labor economist at Berkeley) and Alan Krueger (a Princeton economist who has just finished serving two years as chairman of President Obama’s Council of Economic Advisers). Card and Krueger used data from the College Board to show that, at the University of California, strong black applicants who were likely to qualify for admission to elite public universities (e.g., Berkeley and UCLA) even without racial preferences applied in the same or larger numbers after California’s preference ban went into effect as before (Card & Krueger, 2005). More recently, UC San Diego labor economist Kate Antonovics led a study (which I co-authored) based on data on all applicants to the University of California in the three years before preferences became illegal, and the first three years of the “no racial preferences” regime. We found powerful evidence that when blacks and Latinos received offers of admission at the University of California, they were far more likely to enroll in the period immediately after race preferences ended than they were when race preferences were used. The evidence strongly suggested that minority students preferred to attend a campus where they believed racial preferences were not being used. This was particularly true at UC’s most elite campuses — Berkeley and UCLA — where, incidentally, racial preferences had been most aggressive before the ban (Antonovics & Sander, 2013).
- Kennedy suggests that “racially poisonous environments” at law schools might account for poor black performance on bar exams. But a major purpose of my original essay on affirmative action at law schools was to demonstrate that, when one sets aside the direct effect of racial preferences, blacks perform in law school, and on the bar exam, just as well as whites do.
- Finally, Kennedy argues that my “goal” is to create as many black attorneys as possible — that is, I would be willing to sacrifice the benefits that the “talented tenth” derive from going to elite schools if the elimination of mismatch produced, overall, more attorneys. But I have never argued for such a goal or such a tradeoff. My work, and that of other scholars in the mismatch field, has instead looked rigorously at a whole range of costs and benefits, and we find the current system of preferences to be perverse and miscalibrated to achieve any of a whole range of plausible goals. Regardless of what one’s affirmative action goals might be, our current system does not do a good job of achieving them.
In short, Kennedy completely ignores the undisputed literature on mismatch for science, academic aspirations, and interracial friendship. He addresses the law school mismatch issue in depth, but he seems to have only read critiques of mismatch, most of which have been thoroughly discredited. It is a sad performance.
And mismatch is far from the only situation where, when Kennedy encounters work that does not fit his story, he trivializes it rather than seriously engaging with it. A good example is his discussion of Richard Kahlenberg, a scholar and policy analyst at the Century Foundation, who has been, for nearly a generation, the nation’s leading advocate for “class-based” affirmative action that takes into account the socioeconomic (“SES”) origins of students rather than their race. Kahlenberg deserves enormous credit on several fronts, such as helping to nudge some elite universities toward policies that guarantee “tuition-free” education for students coming from families with incomes below $60,000, documenting ways in which schools can identify low-SES students who are likely to outperform their SAT scores in college, and making the case against legacy preferences (which, as Kahlenberg points out, often constitute “affirmative action for the rich”).
Kennedy portrays Kahlenberg as someone who wants to replace racial preferences with socioeconomic ones simply to avoid making decisions explicitly turn on skin color. According to Kennedy, Kahlenberg
never answers the charge that there is little or no substantive difference between racially selective policies that openly seek to enlarge the numbers of racial minorities in key institutions and policies that seek to accomplish that purpose surreptitiously with no open reference to race […]. Packaging matters. But surely it should not be determinative. More important are the aims behind the packaging and the consequences that ensue.
In fact, Kahlenberg has documented at some length that socioeconomic preferences benefit an almost completely different set of students than racial preferences. It is not only true that conventional affirmative action preferences, as practiced by elite schools, tend to benefit students from very affluent backgrounds. It is also the case that when admissions officers are focused simply on achieving target goals for racial minorities, rather than seeking out disadvantaged students of all races, they often admit wealthy students from Africa or the Caribbean, or students who are multiracial, simply to raise the school’s diversity body count. SES preferences tend to use many criteria of disadvantage, from parental income and education to neighborhood poverty levels, to identify students with genuine disadvantage; they thus tend to be less cynical and manipulable than racial preferences. Even better, good SES programs very disproportionately benefit low- and moderate-income racial minorities often overlooked by conventional affirmative action. And, if executed thoughtfully, SES preferences can greatly reduce the mismatch problem (as happened when my own law school experimented with SES preferences in the late 1990s).
Kennedy’s patronizing dismissal of Kahlenberg’s work is dismaying, and it underlines once again that while Kennedy seeks to position himself as a moderate voice of reason, that simply isn’t possible if one doesn’t read and think about the relevant evidence. Kennedy, I think, poses some core questions about affirmative action correctly. I agree with him that a departure from the nondiscrimination norm is justifiable for a policy that effectively corrects even broad societal effects of past discrimination. But such policies should only be countenanced if they are demonstrably effective: if the costs are easily and measurably swamped by the benefits. Because Kennedy is uninterested in closely studying the relevant evidence on the costs and benefits of affirmative action, he cannot give helpful answers to the questions he asks.
I close with one further example of the curiously disengaged character of On Discrimination. While discussing the origins of affirmative action, on page 34, Kennedy writes:
The idea that merely ending invidious discrimination [through such laws as the1964 Civil Rights Act] would be inadequate for the purpose of attaining a satisfactory sort of racial equality was held not only by dissident outsiders; it was voided as well by the ultimate American insider. In June 1964, at a commencement address at Howard University, President Lyndon B. Johnson declared: “Freedom is not enough….You do not take a person, who, for years, has been hobbled by chains and liberate him, bring him up to the starting line of a race and then say, ‘you are free to compete with the others,’ and still justly believe that you have been completely fair.”
Only eight pages later, Kennedy writes:
President Lyndon Baines Johnson also helped prepare the way for affirmative action. […] An example is the speech LBJ delivered at Howard University on June 4, 1965. [After discussing a part of the speech in which Johnson indicted discrimination, Kennedy continues:] The other pertinent feature [of the speech] is LBJ’s declaration what while freedom and equal opportunity are essential, even combined they are insufficient to overcome the obstacles erected by racial oppression in the past. According to Johnson: “You do not take a person, who, for years, has been hobbled by chains and liberate him, bring him up to the starting line of a race and then say ‘you are free to compete with the others,’ and still justly believe that you have been completely fair.”
There is no acknowledgement in the text that Kennedy is repeating himself, and the duplication seems to be unintentional. One is left with the conclusion that neither Kennedy nor his editor ever carefully read On Discrimination. And if they haven’t, why should anybody else?
Richard Sander is a Professor of Law at UCLA and the co-author of Mismatch: How Affirmative Action Hurts Students It’s Intended to Help, and Why Universities Won’t Admit It
 Barnes never explicitly disavowed her claim that her results argue against mismatch; but her numerical results match closely what mismatch theory predicts. In her model, she compares a regime with large racial preferences for black applicants with a regime with no racial preferences. Under the “no-preference regime,” black law students are 55 percent less likely to fail to become lawyers! See Williams, Sander et al (2011).
Antonovics, Kate, and Richard Sander, “Affirmative Action Bans and the ‘Chilling Effect,’” 15 American Law and Economics Review 252 (2013).
Arcidiacono, Peter, Esteban Aucejo, Patrick Coate, and V. Joseph Hotz, “University Differences in the Graduation of Minorities in STEM Fields: Evidence from California,” (revise and resubmit from American Economic Review, Fall 2013).
Arcidiacono, Peter, Esteban Aucejo, and Ken Spenner, “What Happens After Enrollment? An Analysis of the time Path of Racial Differences in GPA and Major Choice,” 1 IZA Journal of Labor Economics (2012).
Card, David, and Alan Krueger, “Would the Elimination of Affirmative Action Affect Highly Qualified Minority Applicants? Evidence from California and Texas,” 58 Industrial and Labor Relations Review 416 (2005).
Elliott, Rogers, A. Christopher Strenta, Russell Adair, Michael Matier, and Jannah Scott, “The Role of Ethnicity in Choosing and Leaving Science in Highly Selective Institutions,” 37 Research in Higher Education 681 (1996).
Sander, Richard, “A Reply to Critics,” 57 Stanford Law Review 1963 (2005).
Sander, Richard, “Listening to the Debate on Reforming Law School Admissions Preferences,” 88 Denver University Law Review 889 (2011).
Sander, Richard, and Jane Bambauer, “The Secret of My Success: How Status, Prestige and School Performance Shape Legal Careers,” 9 Journal of Empirical Legal Studies 893 (2012).
Sander, Richard, and Stuart Taylor, Jr., Mismatch: Why Affirmative Action Hurts Students It’s Intended to Help, and Why Universities Won’t Admit It, (Basic Books, 2012).
Smyth, Frederick, and John McArdle, “Ethnic and Gender Differences in Science Graduationat Selective Colleges with Implications for Admission Policy and College Choice,” 45 Research in Higher Education 353 (2004).
Williams, Doug, “Do Racial Preferences Affect Minority Learning in Schools?” 10 Journal of Empirical Legal Studies 171 (2013).
Williams, Doug, Richard Sander, Marc Luppino, and Roger Bolus, “Revisiting Law School Mismatch: A Comment on Barnes,” 105 Northwestern Law Review 813 (2011).