Tuesday, May 07, 2013


Estimating Benefits from University-Level Diversity

Barbara Wolfe & Jason Fletcher
NBER Working Paper, February 2013

Abstract: One of the continuing areas of controversy surrounding higher education is affirmative action. The Supreme Court has agreed to hear Fisher v. Texas, and their ruling may well influence universities’ diversity initiatives, especially if they overturn Grutter v. Bollinger and rule that diversity is no longer a “compelling state interest.” But what lies behind a compelling state’s interest? One issue that continues to require more information is estimating and understanding the gains for those attending colleges and universities with greater diversity. Most existing studies are either based on evidence from one institution, which has issues of both selectivity and limited “treatments,” or focus on selective institutions, which also face issues of selection bias from college choice behaviors. In this research we use Wave 3 of Add Health, collected in 2001–02 of those then attending college. Add Health collected the IPEDS number of each college and matched these to the racial/ethnic composition of the student body. We convert these data into an index of diversity and then ask whether attending a college/university with a more diverse student body influences a variety of outcomes at Wave 4 (2007–08), including years of schooling completed, earnings, family income, composition of friends, and probability of voting. Our results provide evidence of a positive link between attending a college with greater diversity and higher earnings and family income, but not with more schooling or the probability of voting.


Do Racial Preferences Affect Minority Learning in Law Schools?

Doug Williams
Journal of Empirical Legal Studies, June 2013, Pages 171–195

Abstract: An analysis of the The Bar Passage Study (BPS) reveals that minorities are both less likely to graduate from law school and less likely to pass the bar compared to whites even after adjustments are made for group differences in academic credentials. To account for these adjusted racial gaps in performance, some researchers put forward the “mismatch hypothesis,” which proposes that students learn less when placed in learning environments where their academic skills are much lower than the typical student. This article presents new results from the BPS that account for both measurement-error bias and selection-on-unobservables bias that makes it more difficult to find a mismatch effect if in fact one exists. I find much more evidence for mismatch effects than previous research and report magnitudes from mismatch effects more than sufficient to explain racial gaps in performance.

Nod to Kevin Lewis


Anonymous said...

I've read that first abstract 3 times and can't see how it doesn't say: "Previous studies have a selection problem. Here's our results that also have a selection problem."

Anonymous said...

It's also some sick kind of irony that now we're evaluating whether affirmative action is a good policy based on whether it's good for white people.