Systemic Analysis findings

Summary of the Critical Response

The admissions systems used by nearly all law schools race-norm black applicants in ways that violate the law

Not discussed by most critics.  Ian Ayres & Sidney Foster have written a piece [link] which claims to improve on my analysis, but shows that Michigan Law School used larger and more mechanical racial preferences than those used by the College and invalidated in Gratz.

Out of a 1000-point academic index scale, most law schools in the 1990s had a 170-point gap between white and black students

These three points are generally conceded; I'm aware of no critique.

Preferences are not limited to elite law schools, but due to a “cascade effect”, ripple down the law school hierarchy intact

Blacks perform badly in law school; median black student is at 7^th percentile of white students

The reason for low black performance has little to do with race; it’s a function of the large admissions preferences, which put blacks at a big academic disadvantage.  Blacks and students of all other races perform roughly as their academic credentials predict they will.

This is contested.  (1) Some analysts (e.g., Clydesdale) have used BPS data to argue blacks greatly underperform their credentials; but these analyses regress unstandardized credentials on standardized grades, producing meaningless results.  (2) Several critics pointed out that in some formulations, my data shows some modest black underperformance.  This is true, but several different sources and methods show the possible range of black underperformance is "insignificantly small" to "modest".  (3) Other critics infer underperformance by blacks from various cohort comparisons in the BPS, but these critiques consistently suffer from serious selection bias.

Students with very low grades are much less likely to graduate from law school.

Both of these claims are apparently conceded; generally not much discussed.

Students with low grades are much less likely to pass the bar

Without preferences, blacks would have go to lower prestige schools but would have higher grades, thus dramatically improving their graduation and bar passage rates

This is contested.  One response (e.g,, Ayres & Brooks) is to compare blacks and whites with similar credentials at schools in the same tier, and show that blacks still have worse outcomes than whites.  This is true, but suffers from serious selection bias and the wide range of schools within single tiers in the BPS data.  Another response (e.g., Rothstein & Yoon) is to concede this might be true (for bar passage), but argue that at the high end of black students,  the differences are not large enough to be statistically significant   I believe Rothstein & Yoon are viewing the evidence very selectively.

Legal employers generally give more weight to performance in law school than to law school eliteness, so black earnings would go up if mismatch effects went down.

Dauber suggested my analysis might not apply to blacks separately, but I showed in "Reply to Critics" that it does.  Wilkins contends that elite schools confer many benefits other than initial job placement; the evidence is mixed but this is unresolved.  Dinovitzer & Sterling, who have access to the AJD data I used in this analysis, presented findings arguing the grade effects are smaller than I suggested, but I haven't seen this paper and so far as I know it is not published or widely circulated.

Consequently, pushing blacks into higher-ranked schools where they do badly hurts their job prospects

Critics rightly point out that at top 10 schools, the earnings boost of an elite degree measures at a virtual wash with the grade penalty, so this argument might not apply to the very top.  (My initial analysis noted this caveat.)

Because of the mismatch effect, the attrition effects of the current system threaten to swamp the pool-expanding effects of affirmative action.  If black applications remained constant in a race-neutral regime, it’s likely that the number of new black lawyers produced by the system would go up.

The most debated (and mischaracterized) claim of "Systemic Analysis."  Many treat my discussion as a definite prediction rather than as a qualified thought experiment.  Some responses:  (1) If there is no mismatch effect, then ending preferences would lower the number of black attorneys (true, by about 10% under the rest of the model's assumptions).  (2) The year used in my simulation -- 2001 -- was unrepresentative (false, but it is true that more recent data shows a somewhat worse outcome on black enrollment if preferences ended).  (3) If preferences ended, black applications might fall sharply (certainly possible, though at least one study (Dale & Krueger, 2004) suggests otherwise); (4) If preferences ended, blacks would have no choice but law schools in Montana and other god-forsaken places (silly and demonstrably false in several respects).