As I mentioned before, the data suggests when you look at that bottom 10% of a given law school class, and compare them to students at lower-tier schools who had the same incoming credentials, the students at the lower-tier school perform better on the bar.

This suggests one way of addressing the problem of affirmative action: get rid of it. Let URM students compete purely on their numbers. They'll still get into law school, just not the same ones. But theoretically, they'll be more likely to become lawyers. This is a proposal advanced by Prof Richard Sander, who has done much recent research in the area (and whom I've studied with this past year)

I buy into a lot of Sander's findings, but I disagree with him on this point. One major problem with suspending all racial preferences would be that URMs would largely disappear from top-tier law schools. This is exactly what happened to Boalt and UCLA law in the year after Prop 209 (which banned racial prefs in Calif) was passed. I think it's important, both for the sake of the student body & the profession, that elite schools continue to have the tools to admit URMs.

The other problem is the law of unintended consequences. The statistical prediction is just that – a prediction. Removing racial preferences would shift most URM students down to the next lower tier of school. Without knowing which individual schools offer the best success rates for URM students, we can't say with any certainty that this would improve the overall output of URM lawyers. It might decrease it.

I don't think eliminating preferences is the answer. Rather, I think better preferences (ie. those with risks disclosed) and better follow-up (e.g. academic support programs for the bottom 10%) is the way to go. Don't send those students down to a lower-tier school. Work on bringing them up to the standards of the school they're at, so they can enjoy the same graduation and bar passage rates of any other student. As Bill Clinton said of affirmative action 10 years ago – mend it, don't end it.

Risk disclosure is ultimately about accountability. Risk disclosure, if it were required by all law schools, would become a form of benefit disclosure to applicants. If you were a student entering in that bottom 10%, you would have the information you need to compare schools. Students would gravitate towards schools posting the best results, and more of them would eventually become lawyers.

What about the other schools, with worse results? They'll have the competitive pressure of the free market as an incentive to improve. In the short term, they'll probably have to provide tuition subsidies & scholarships to get URMs to attend. In the long term, they'll want to capitalize academic support & other programs that bring URM performance into parity.

Will there be students who, confronted with the risks, change their mind about going to law school? Sure, but I can't say I see that as a bad thing. Right now, there are plenty of low-credential students who are induced to go to law school in spite of the risks. They spend 3 yrs and $100K, but they can't pass the bar. If we deter these students from going to law school, we haven't changed the outcome – either way, the student has no career as a lawyer. At least their time & money haven't been wasted in a fruitless pursuit.

Further reading:
Richard Sander, A Systemic Analysis of Affirmative Action. Sander's 2004 article kicked off the current reheating of the AA debate. This link includes the original article, a set of critical responses by other law professors, and Sander's Reply to Critics, where he adds some important extra analysis to the original data.

Among other findings, Sander introduces his theory of the 'mismatch effect'. The idea is that students with poor entering credentials start at the bottom, but fall farther and farther behind as they struggle in classes that are aimed above their skill level. It's not a racial theory, nor is it specific to law school. Sander's point is that admissions preferences induce the mismatch effect for many students receiving those preferences.

MB, Addressing Risk Disclosure in Affirmative Action Through Litigation. Could AA recipients make a legal claim against law schools for failure to disclose risks? I expand on the points here. I also explore theories of legal liability in detail. The short answer is, common law fraud and tort claims probably wouldn't work. But consumer protection claims probably would. I think law schools are wide open right now to this kind of suit.

David Bernstein posts on Volokh.com about last week's civil rights commission hearings in Washington DC about affirmative action in law schools. Bernstein testified, as well as Sander. This issue has not escaped notice of the commission:

Several [civil rights] commissioners expressed grave concern about the extraordinarily high rate at which African American law students at non-elite law schools either fail out of law school or fail to pass the bar exam ... while the new [ABA accreditation] standard requires law schools to pursue diversity in admissions, it says nothing about the need to ensure that admittees actually succeed in becoming attorneys. [The ABA representative] acknowledged that the ABA committee that drafted the Standard has no data regarding the failure rate of African American matriculants at particular law schools, nor about how the new standard may worsen this rate.

The ABA can't get its hands on data about URM bar passage rates? Is there anyone who actually believes that? Anyone? Apparently, the Dept of Education is considering suspending the ABA's accreditation authority until some of these issues are worked out – that is a major administrative sanction right there.

Jerry Kang, Fair Measures: A Behavioral Realist Revision of Affirmative Action. This is a new article by Kang, who's a Critical Race Theory prof at UCLA law. Kang relies on 'stereotype threat' to explain differences in racial performance. I don't personally buy this explanation – among other issues, Kang & his psychologist co-author rely on psychological studies, and almost no academic performance data. But it's another view.

20 Jun 06