Legal Affairs

Current Issue


printer friendly
email this article

space space space
Debate Club

Does Affirmative Action Hurt Black Law Students?

Richard Sander and William Henderson debate.

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

In the current issue of the Stanford Law Review, Richard Sander of UCLA Law School uses empirical analysis to mount a broad criticism of affirmative action. For black law students, Sander writes, the "costs of preferential admissions appear to substantially outweigh the benefits." Sander's core argument is that many black students would perform better and pass the bar at a higher rate if they attended less prestigious law schools.

Do preferential admissions policies hurt black law students?

Richard Sander is Professor of Law at UCLA. William Henderson is Associate Professor of Law at Indiana University.

Henderson: 1/10/05, 01:19 PM
Rick, I am really happy to have this forum to discuss your article. In the months prior to its publication, your study generated a huge volume of posts, exchanges, and "alerts" on various academic list-serves. Although it was not your intent, your study has been fodder for the two loudest participants in the affirmative action debate: those for and those against.

Unfortunately, I don't think the legal academy will reach any constructive conclusions on your study until we are capable of having exchanges that are driven primarily by data rather than ideology. Therefore, I would like to use this forum as an opportunity to push this dialogue in a new direction—and hopefully strike a more constructive tone.

So do preferential admissions policies hurt black law students? Rick, I think you would agree that a more precise formulation of this question is whether preferential admission policies produce a net harm for black students as a group. Certainly, your critics would have no problem assembling legions of black lawyers who could explain in great detail the opportunities opened to them by affirmative action. However, the net effect is measured in the aggregate rather than by individual success stories. The central finding of your study is that racial preferences appear to set in motion a pattern of "mismatched" credentials that (a) causes the majority of black students to cluster at the bottom of their law school class, (b) increases their attrition rate, (c) reduces their likelihood of passing the bar, and (d) handicaps them on the job market because, in most cases, the superior pedigree of their law school does not compensate for a poor law school record. Moreover, according to your study, the unintended consequences of this "mismatch effect" become more pronounced the further we move down the law school hierarchy.

Frankly, if this dialogue is going to be data-driven, I have to concede that there is, at present, no empirical evidence that directly rebuts your findings. However, I think some context is in order.

Prior to your study, there was limited empirical evidence suggesting that racial preferences improved the career prospects of black law students. Specifically, in 2000, a group of professors at University of Michigan Law School published a study in Law and Social Inquiry documenting the impressive career trajectories and accomplishments of its minority graduates, many of whom were admitted via racial preferences. This study was celebrated throughout the legal academy because it corroborated William Bowen's and Derek Bok's famous study, The Shape of the River: Long-Term Consequences of Considering Race in College and University Admissions, which documented the benefits of racial preferences at the undergraduate level. Nevertheless, it was unclear whether these findings could be generalized to all undergraduate institutions or law schools because both studies focused on minorities at elite colleges and law schools, where the relative costs and benefits of racial preferences may not be representative of higher education as a whole.

Your study cured this potential bias by using data that encompasses a much broader range of schools. Now, for the first time, we are directly confronted with a difficult question: Does our current system of affirmative action produce a net gain for African Americans students? You acknowledge the possibility that racial preferences may, on average, produce a "small plus ... for [black] students at top ten schools." Thus, your findings are at least partially consistent with the 2000 Michigan study. But for all black students throughout the law school hierarchy as a whole, your data suggests that racial preferences produce a net harm.
In November, David Chambers, Timothy Clydesdale, William Kidder, and Richard Lempert released a response paper that is very critical of your study. Their critique, however, focuses primarily on your subsidiary claim that a race-neutral admissions system would ultimately produce a net gain in the total number of blacks entering the legal profession. They argue that a race-neutral system would actually produce fewer black lawyers. Although this is certainly an important issue, it is separate and distinct from your broader empirical point that the mismatch effect is strongly associated with (a) lower law school GPAs, (b) higher law school attrition, (c) lower bar passage rates, and (d) diminished prospects on the job market.

To my mind, your study has evoked a profound fear, and that is—assuming we care that racial preferences may do more harm than good in terms of promoting the career prospects of young black lawyers—our only viable solution is to scuttle one of our most cherished and noble ideals: racial diversity in legal education. In fact, our options are numerous—they are just not as easy as applying racial preferences during the admissions process.

My solution: close the performance gap between white and minority students. I have some ideas on how to accomplish this, but first I need to know whether you think the white-minority performance gap can be attributed to immutable, inherent differences between ethnic subgroups, or some combination of other factors that are amenable to change. In other words, does your study present us with an opportunity to solve a longstanding social problem, or is it merely the end of the road for law school diversity?

Sander: 1/10/05, 07:39 PM
I'm delighted to join you in this discussion. I've admired your intellectual honesty, and share the view of many others that your recent article on timed exams is the most thoughtful critique of the Law School Admissions Test (LSAT) to appear in many years. You and I agree that law school admissions systems need some serious rethinking.

I also agree with you that hidden fears about the immutability of the black/white test score gap have haunted the discussion of racial preferences for decades. If the gap is viewed as somehow "intrinsic," then a natural reaction is to either reject the legitimacy of the tests themselves (and perhaps all ranking systems in general), or to embrace some notion of racial proportionality as a necessary safeguard for minority interests. As my paper suggests, I think the second of these strategies is turning out to be counterproductive for blacks in the law school world.

The evidence for genetically-based racial differences in intelligence was always very weak, as Steven Jay Gould showed marvelously in The Mismeasure of Man. But I think it's only in the last ten to fifteen years that we've finally been able to put the spectre of genetic differences to rest with a battery of superb studies. Several studies of orphans have shown that whites and blacks raised in similar households have identical IQs (some studies, indeed, found blacks outperforming whites). Studies of American GIs in postwar Germany found that the children of both black and white servicemen, when raised in Germany, had nearly identical test scores as teenagers. And so on. Most psychometricians still agree that "intelligence" is a meaningful concept and has a genetic component, but there is an overwhelming consensus that racial differences are due entirely to environmental factors. Hence a subtle but important change in terminology—we now speak of tests measuring "cognitive skills", not "cognitive ability."

Accordingly, it's not surprising to me that the black/white gap in test scores has been declining gradually but steadily ever since the 1960s. On the LSAT, the gap in the median scores of blacks and whites has fallen from 1.6 standard deviations in 1976, to 1.2 standard deviations in 2001. And since the black/white gap in college attendance has fallen too, making the black pool larger, the number of black students with strong qualifications quadrupled over the same period.

So you and I are both environmentalists on the issue of racial gaps in test scores. How do you tie this into the affirmative action debate?

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

Henderson: 1/11/05, 12:21 PM
Rick, as your post suggests, over the last three decades, a reflexive "racial proportionality" has shielded us from uncomfortable questions on the white-minority performance gap. If racial preferences improve the life chances of black law students, perhaps this is good social policy. In Grutter, the Court upheld affirmative action because "the educational benefits that flow from student body diversity" comprise a compelling state interest. As a normative matter, I agree: the educational benefits of diversity are profound. I have experienced them firsthand with my students and colleagues.

Your study disrupted the status quo by demonstrating that racial preferences may have the unintended effect of actually harming the career prospects of aspiring black lawyers. Obviously, the stakes here are extremely high. In this instance, shallow or uninformed dismissals have the potential to perpetuate a serious social harm. Before we dismiss your analysis, we need to make sure that you are wrong. (Readers are encouraged to carefully read Sander's study—all 112 pages—and four forthcoming response essays in the Stanford Law Review.) Further, if you are right—and, though we still have some responses that need to be considered, you have a lot of evidence on your side—then our current system of affirmative action needs to be reexamined. We can argue about the form of change—we both agree that a rapid transition to a race neutral system would be a mistake. But it would be morally indefensible to do nothing.

I would like to return to my earlier point on the white-minority performance gap and the unspoken fear (refuted, as you note, by the best available social science) that it is caused by inherent differences between the races. In a recent discussion in the Buffalo Law Review, one of your staunchest critics, David Chambers, expressed his apprehension that "Grutter is going to make law schools complacent." Because law schools can continue to use racial preferences, Chambers observed, "we may devote too little attention to improving minority success at the stage before and after admission." I agree. If Grutter struck down racial preferences, it would have been a remarkable catalyst for change. Ironically, your study may have functionally the same effect.

As you note, the black-white LSAT gap has narrowed over time. This is good news. However, the persistence of a large performance gap over a period of decades also runs the risk of subtly reinforcing negative stereotypes. In my opinion, this is the price we pay by failing to pursue ambitious social science that asks hard, uncomfortable questions. Ultimately, I believe that a sizeable portion of the performance gap is attributable to one of two correctable sources: (1) measurement error, in terms of standardized tests and their relationship to a criterion variable that they are designed to measure; and (2) educational, cultural, economic, and familial disparities during the pre-K-12 years that reduce a minority student's preparedness for higher education (the "pipeline" problem). I will elaborate in a later post.

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

Sander: 1/12/05, 09:07 AM
So, I think we've identified two spheres in which the new debate on law school preferences might proceed. One is the debate on the actual effects of preferences on the intended beneficiaries. As you say, the important thing here is to keep the debate focused on the data and the issues. Much of the initial reaction was from the two established camps, but in the coming months we will start hearing from new voices who are intrigued by the findings, checking out the data, and putting together their own research. I am going to try to post this research as it comes out on my website. (I've recently posted there the initial Chambers/Lempert critique of my article, and my reply to their critique.)

The second sphere is the debate about what to do if it's true that preferences are even somewhat injurious to blacks and other minorities. This is the question that particularly interests you. In my thinking about this so far, I've come up with four steps that seem to me modest and reasonable:

1) Law schools should start disclosing to their admitted applicants the average graduation and bar passage rate (first time and ultimate passage) for students who have credentials similar to the applicant. Currently, applicants can only get this sort of information for the student body as a whole, which conceals the fact that the weakest students at any given law school tend to have much more difficulty becoming lawyers. The data disclosures would help minority candidates judge offers from schools that will extend them a large admissions preference, but which may put them in greater academic jeopardy.

2) We should have a much more concerted effort to understand the potential benefits of academic support. My past research with Kris Knaplund suggests that academic support can help overcome—to a modest degree—the academic disadvantages of students having trouble in law school. But (a) many academic support programs don't work; (b) the ones that do can be oversold; and (c) there has been far too little effort to replicate the best ones across the country. This is the sort of complacency to which David Chambers refers, I think, and I join him in wishing for more commitment.

3) As the research progresses on the mismatch effect, we should look closely at the effects of reducing the size of preferences. My preliminary work suggests that cutting the size of preferences for blacks in half would reduce mismatch effects by 80%. I'll try to explain the logic behind that pattern in a future post.

4) We owe it to law students to seriously explore ways of broadening our admissions criteria. This is difficult; it's easy to be too glib about what can be done; but we ought to try. How would you go about this?

Henderson: 1/12/05, 01:12 PM
Rick, I think you have correctly framed two spheres of constructive debate: (1) the existence and extension of the mismatch effect, and (2) how we respond to it. You observe that I am particularly interested in the latter. In fact, I am extremely interested in both. I just think the sphere one discussion will be more principled and more reasonable in duration if the participants had some idea of the range of possibilities in sphere two. Let me explain.

Your study was the first salvo in high stakes debate in sphere one. In my opinion, the broader dialogue has gotten off to a bad start because many important constituencies have an unarticulated fear on its ultimate endpoint—abolition of affirmative action and racial stratification within the legal profession. However, my own academic research has made me very attuned to the imprecision and uncertainties that pervade our entire system of law school testing (the LSAT, law school testing methods, and the bar exam). There are myriad ways we can improve this system that simultaneously further our dual goals of meritocracy and social justice. If legal academics better appreciate our range of options, I believe the sphere one debate has a better chance of being data-driven and therefore more constructive.

As you know, there is huge academic acclaim waiting for the researcher who can demonstrate, on methodological or empirical grounds, that your study is flawed. Many top-notch academics are racing for this prize. However, as a matter of comparative advantage in this broader debate on racial preferences, I have focused on sphere two—what we do if the mismatch effect is injurious to blacks (and possibly other ethnic subgroups).

I generally agree with your four proposals. But I would go further.

For example, if the mismatch effect holds up, you recommend that law schools should disclose average graduation and bar passage rates for students with credentials similar to the applicant. Of course we should publicize the data. Our students are highly educated adults who are making an extremely expensive career decision. If they can't be trusted with consumer information, no one can. Moreover, this information should be compiled, standardized, and posted on a website so that students can see which law schools provide the greatest boost to bar passage after controlling for entering credentials. (The raw scores on the MBE and MPT would provide uniform independent variables for virtually all 50 states.)

This would cause law schools to (a) develop effective support programs (which dovetails with your option (2) above), (b) examine the relationship between curriculum, teaching methods, and bar passage, and (c) demand that the bar exam be a true measure of legal ability and competence. There has been virtually no research on this last point. It is a complete embarrassment. We should disclose that too!

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

Sander: 1/13/05, 09:06 AM
Excellent suggestions, Bill. Better disclosure by law schools (and the provision of actual bar scores to schools for statistical disclosure purposes) would create a number of very healthy incentives in the system, and would help push legal education away from the credentialing exercise it drifts toward, and more towards an active pursuit of educational excellence. It would also empower students to look at something beyond a school's U.S. News ranking.

I agree, too, that we should work on improving the bar. I like the California system, which adds to the usual multistate and essay parts a one day "legal practice" exam, giving the examinees a packet of documents and asking them to solve realistic legal problems for hypothetical clients. You seem to have a more ambitious agenda in mind—trying to measure what actually makes a good lawyer, and better fitting both the bar exams and law school admissions to an assessment of those qualities. I'm all for that, so long as the net effect is not simply to water down existing standards.

Let me return to another of my reform proposals—considering reductions in the size of preferences. My paper demonstrates that nearly all law schools currently follow a rule of rough racial proportionality, admitting students in proportion to the racial makeup of the applicant pool (for blacks, this means they average about 8% of each school's students). This creates the cascade effect, in which the size of admissions preferences doesn't fall at all as one moves down the hierarchy of schools. If schools adopted a different approach, using preferences for blacks only to the extent necessary to keep black enrollment from falling below 4%, a very different dynamic would emerge. Preferences at the top schools would be much smaller, of course. But more importantly, the size of preferences needed to reach the 4% goal would diminish rapidly as one moved down the law school hierarchy, because the cascade dynamic would be disrupted. Fairly high up in the hierarchy, preferences would effectively disappear, and black enrollments at most schools in the bottom half of the law school hierarchy would be well above 4% (and often above 8%) using purely race-neutral standards. In other words, moving from an 8% goal to a 4% goal would eliminate more than three-quarters of the black/white credentials gap. Moreover, the schools with the largest remaining gaps—those at the top of the hierarchy—are the schools at which the negative consequences of the mismatch effect are smallest.

Henderson: 1/13/05, 12:57 PM
Rick, your 4% proposal has two problems, one substantive and the other tactical.

First, it has risks you don't fully account for. Specifically, by your own admission, a 4% black enrollment at elite schools could, through the lessening of the cascade effect, produce "black enrollments in the bottom half of the law school hierarchy ... well above 4% (and often above 8%) using purely race-neutral standards."

So, in other words, as we move down the U.S. News rankings, the proportion of blacks would significantly increase, mirroring (at least in part) the underlying black-white test score gap. As someone who believes that this gap can be closed, I worry about any compromise that accepts the current gap as an appropriate starting point. To date, we have expended too little effort systematically examining the imperfections of our testing instruments or the impact of large cultural and economic disparities that persist during the pre-K-12 time period. I think your 4% solution may give too much credence—albeit inadvertantly—to the accuracy and fairness of our existing system of testing. Thus, I don't think it is a good solution, even in the short run. Disclosing the effects of the current system might be enough. Further, the tilted enrollments, with blacks underrepresented at the top of the hierarchy and overrepresented at the bottom, could spawn unintended "tipping" dynamics where blacks and whites self-segregate by school. I am attracted to the benefits of reducing the mismatch effect by 80%, which your 4% plan would purportedly achieve, but there could be serious unintended consequences.

My second criticism is tactical. Returning to the sphere one / sphere two framework, your 4% plan is clearly in sphere two—how law schools should respond. In my opinion, you would be wise to limit your public comments to sphere one, the empirics that document the existence and operation of the mismatch effect.

In your article, your 4% solution follows the results of a simulation of the number of black students who eventually pass the bar under a race neutral system; you claim this number would increase by nearly 9%, even though law schools would enroll 14% fewer blacks. This is the area of the debate where all your critics are piling on, primarily because your simulation depends upon contestable assumptions—a point you explicitly acknowledge! Notwithstanding the fact that you offer reasonable explanations for all your assumptions, I fear the most important issues raised by your study are being obscured.

Here is the nut of the problem: Racial preferences benefit me, a white male, because diversity enriches the educational experience for professors and students. Scuttling racial preferences in favor of a race-neutral system may or may not produce more black lawyers; more black lawyers is a good that benefits society. But it is also possible that our current system of racial preferences, through the mismatch and cascade effects, may harm the career prospects of aspiring black lawyers. This is the central issue raised by your study. It is a sphere one debate. Force your critics to confront it.

In the meantime, I'll do my best to convince people that the world doesn't end in sphere two.

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

Sander: 1/14/05, 12:28 PM
With our week nearly over, Bill, it's high time we found something on which we seriously disagree. If there really is a mismatch effect—and I would guess that by the end of 2005, most of those involved in the debate will concede that there is—then it doesn't seem reasonable to "take off the table" any strategy that actually improves the match of students with the credentials of their fellow students. I'm not advocating immediate implementation of the "4% solution" or other reallocation plans; I think we are still at the stage of measuring the problem and evaluating possible solutions. But it is vital that approaches which directly reduce the mismatch be part of the catalog of possible solutions we are comparing. The 4% solution has much more potential to reduce the harmful effects of the mismatch quickly and dramatically than any of the other approaches.

I also disagree that something like the 4% solution accepts or locks in the current credential disparities. On the contrary, it's our current system of proportional admissions that has fostered utter complacency. The discipline of being able to use preferences on only half the current scale would create constant, significant pressure for schools to keep their eye on the ball and pursue the other strategies we have discussed. As the black/white gap continued to narrow, as academic support improved, and as our admissions indicators broadened, the black presence at top schools would visibly rise, and the mismatch effect would wither away.

I note in passing that the theory of "racial isolation" is an unproven hypothesis. There's some important evidence to the contrary. Economists Card & Kruger found that after Hopwood and Prop 209, black high school graduates in California and Texas with strong academic credentials continued to apply to Berkeley, UCLA, and University of Texas in exactly the same net numbers—even though their chances of admission were lower and the black presence at those schools had dropped significantly. This is consistent with the idea that blacks find academic environments with much smaller credential gaps along racial lines very appealing—even if it means somewhat fewer black classmates.

You're right that I need to try to keep debate focused on the existence and size of the mismatch effect. But I want to keep reminding participants that one can make enormous progress in addressing this problem short of abolishing preferences. And it's also worth keeping in mind that we should not view the current system as "innocent until proven guilty." We already knew that preferences carried enormous costs. The burden of proof should be on those seeking to show that its net benefits are real.

Henderson: 1/14/05, 03:20 PM
Rick, at the end of the day, you and I and the majority of legal academy are on the same side. We value diversity, and we expect racial preferences to help (or at least not harm) our students. And, I think, the burden of proof has indeed shifted.

That said, I want readers to have at least one concrete example of the imprecision and uncertainty in our current system of testing. In an earlier project, I examined student oral and brief writing scores for an appellate advocacy class at a mid-ranked law school. Remarkably, even though the substantive material was identical, brief writing scores were uncorrelated with oral argument scores. Further, the oral arguments scores were uncorrelated with LSAT and undergraduate GPA. It is important to realize that neither the bar exam nor the majority of law school courses test knowledge using oral communication skills, a highly prized ability within the legal profession. Granted, this raises difficult test-design problems, but more and better research needs to be done—and quickly.

This Week's Entries: Monday | Tuesday | Wednesday | Thursday | Friday

printer friendly email this article letter to the editor
space space space

<& /legalaffairscomp/ads_dc.comp &>

More on Affirmative Action
WSJ: Critics Assail Study

Affirmative Action Timeline

Weekly Standard: The Mismatch Game

G. Liu: Sander is wrong

<& /legalaffairscomp/rss.comp &>
Contact Us