Part V. The Effect of Affirmative Action on Law School Admissions Outcomes In this Part, we step back from our analysis of mismatch to examine the impact of
A. Simulating Race-Blind Admissions: Methodology
We assume that with race-blind admissions, black and white applicants with the same numeric, academic credentials would be admitted at the same rates. We recognize that law schools rely on other measures of ability beyond what we observe. There is substantial variation in admissions outcomes among students with the same observed credentials, presumably because some students have strong unobserved qualifications—essays, recommendations, etc.—while others do not. We do not assume that this heterogeneity would disappear with race-blind admissions. Instead, we assume that the distribution of admissions outcomes for black students would match that of white students with the same credentials – the same fractions of black and white applicants would be admitted to selective schools, admitted only to unselective schools, and not admitted at all.
Implicitly, then, we assume that black students’ unobserved credentials have the same distribution as those of white students, and that observed differences in their admissions outcomes reflect the availability of preferences for black students. This assumption is also implicit in our analysis of mismatch in Part V, as it is only reasonable to assume that black and white students would achieve the same law school outcomes if they have similar average
entering credentials. As before, we believe that it is most likely overly optimistic: Just as black students have lower average GPAs than white students with the same LSAT scores, it seems likely that when we match black students and white students on the basis of both LSATs and GPAs, the black students’ other qualifications will be worse, on average, than those of the white students.114 If our assumption is indeed incorrect, we will overstate black admissions outcomes under race-blind admissions by assuming that they would match those of whites with the same observed qualifications, and accordingly we will understate the role of affirmative action.
113 Our method is poorly suited to the analysis of historically black law schools. See the
discussion in footnote 118.
114 This arises from the well-known statistical phenomenon of “regression to the mean.” See Sir
Francis Galton, Regression Toward Mediocrity in Hereditary Stature, 15 J. OF THE
We make another important simplifying assumption: the admissions standards applied to white students’ applications will not change with the elimination of affirmative action. This is almost certainly incorrect. Without preferences, some admissions slots currently given to black students would become available to white students who currently fall just short of admission. Still, the number of admission slots freed through the elimination of preferences would be small and the change in the white admissions standard would therefore be negligible.115
An important issue is that applicants’ decisions whether to apply to law school, how many applications to submit (and to which schools), and, if admitted, whether to matriculate might all be affected by their perceived admissions probabilities. In a similar study of college admissions, Krueger et al. find that black high school students with moderately high SAT scores are much more likely to apply to the most selective colleges than are white students with the same SAT scores.116 They argue that this gap is probably due to the availability of affirmative action, and that black students’ application behavior would likely resemble that of whites if the two groups of students faced equal admissions probabilities.
We follow Krueger et al. in assuming that application and matriculation behavior would converge with race-blind admissions. We do not, however, assume convergence in every dimension: we assume that reduced admissions probabilities will not dissuade any black
students from applying to law school in the first place. This assumption is admittedly unrealistic. As we demonstrate below, many of today’s applicants would not be admitted anywhere under race-blind rules. Many potential applicants would anticipate low probabilities of acceptance and forego applying to law school altogether.117 Because a fraction of these discouraged applicants would be admitted if they did apply, our assumption will lead us to overstate the number of black students who would be admitted, particularly at low qualifications levels.
115 Tom Kane makes an analogy to handicapped parking spaces. “Suppose that one parking
space in front of a popular restaurant is reserved for disabled drivers. Many of the nondisabled drivers who pass by the space . . . may be tempted to think that they would have an easier time finding a space if the space had not been reserved . . . . [T]he cumulative cost perceived by each passing driver is likely to exceed the true cost. . .” Misconceptions, supra note 9, at 453.
116 See Krueger et al, supra note 75.
117 An additional factor is that the low black shares at many law schools would likely make the
environment appear more hostile to potential black applicants than it does today, perhaps dissuading many who would otherwise apply. See Chambers et al., supra note 8, at 1862.
Taken together, we believe that our assumptions are reasonable, appropriately cautious, and conservative. Likely violations would lead us to understate the impact of eliminating affirmative action on diversity.