Attacking the 'Mismatch' Critique of Affirmative Action

One of the more influential and controversial studies of affirmative action in recent years came from Richard H. Sander in 2004. The law professor at the University of California at Los Angeles analyzed statistics about black law students and argued that they show that affirmative action hurts them by helping many gain admission to institutions where they are unlikely to be top students. This "mismatch," he argued, led to academic performance at lower levels than the same students would have achieved at the less prestigious law schools to which they could have earned admission without the consideration of race.

Sander's theory set off political and academic fireworks. They may be reignited with the release of a new study by the National Bureau of Economic Research that examines much of the same data Sander used. The new analysis -- with the sort of detail normally used by critics of affirmative action -- backs Sander's claims on the substantially different qualifications, on average, between black and white law school applicants. And it also finds gaps in performance in law school.

But the new research -- using simulations of admissions without affirmative action -- finds that race-neutral policies wouldn't send black students to law schools where they would do better. Rather there would be a huge falloff in black law enrollments -- far more than might be counteracted by some black students doing better on bar exams. The elimination of race-based admissions policies, the authors write, would lead to a 63 percent decline in black matriculants at all law schools and a 90 percent decline at elite law schools, the paper says. Even if some positive impact took place in the experience of black students who did enroll, there would be at least a 50 percent reduction in the production of black lawyers, they write.

The study -- by Jesse Rothstein of Princeton University and Albert H. Yoon of the University of Toronto -- starts off by reviewing the Sander data and its significance. Rothstein and Yoon note that a key part of Sander's thesis is that "affirmative action influences which schools African American students attend, but has only small effects on whether these students attend law school at all."

Rothstein and Yoon then turn to data about students who applied to law school in 1991 -- similar to the data Sander used. (They authors of the new paper note that changes in the performance of black students have been very modest since then, suggesting that similar findings would be possible with today's data, although more recent data sets are less complete.) The authors also note that the Sander argument -- left unchallenged -- has the potential to undercut the political and legal arguments for affirmative action. From President Lyndon Johnson through recent Supreme Court arguments, defenders of affirmative action have assumed that its practice promoted black advancement rather than hindering it.

Notably, Rothstein and Yoon accept that the "mismatch" theory is "plausible," given that admitting a student to a law school above his or her abilities could frustrate and "demoralize" him or her. Further, they note that the requirement to pass the bar means that students who get through law school but can't practice are sure to feel particularly poorly served by the process.

Where Rothstein and Yoon depart from Sander is in their analysis of the differing admissions qualifications and academic performance of black and white law students. Sander uses the data to suggest "mismatch." For example, 92 percent of white students who enter law school graduate, while only 81 percent of black students do. And of those who graduate from law school, 87 percent of white students pass the bar within a year, compared to 64 percent of black students.

But Rothstein and Yoon focus more on admissions data to suggest that -- without affirmative action -- many black applicants to law schools would simply never get in.

For example, in the data studied, only 1.5 percent of black students taking the LSAT achieved a score of at least 38. Of all of those admitted to Yale University's law school, 89 percent had scores of at least 41. While Yale law is among the most competitive law schools in the country, the study notes that there isn't such a thing as a truly non-competitive law school and that admissions credentials gaps are present at all types of law schools. For example, at American University's law school for the years studied, 91 percent of admitted applicants had LSAT scores above 32 and college grade-point averages above 2.5. Only 2.2 of law school applicants nationally with those credentials are black. (The LSAT scale has since changed to once in which scores are reported from 120-180.)

The authors conduct a simulation -- based on admissions patterns of various types of institutions -- to show what would happen to black representation at law schools, and the results are dramatic. Students don't shift down a prestige level -- but disappear. Only at "third tier" law schools would there be any gains for the black enrollment share, and those are small.

Impact of Race-Neutral Admissions on Share of Black Law School Matriculants

The authors of the new study can hardly be described in their analysis as cheerleaders for affirmative action. "Given our findings, it is reasonable to ask whether law schools serve students' best interests by admitting applicants with low academic credentials, or whether these students would be better off not attending law school," they write, adding that their analysis "cannot answer this question."

But they argue that their work does definitively debunk the "mismatch" thesis. "We find that affimative action is pivotal in achieving racial diversity in law schools, and that any resulting mismatch effects are concentrated among students who would not be admitted to any law school without preferences," the authors write. "As a policy matter, reasonable people may disagree about whether the costs of 'taking a chance' on marginal black applicants outweigh the benefits, and we have little that is new to say about this. Our analysis suggests, however, that one cannot credibly invoke mismatch effects to argue that there are no benefits."

Sander, the mismatch proponent, could not be reached for comment. But his Web site contains many articles and data sets he has set up to back up his views.