A note on Fisher v. University of Texas as it goes back to the Fifth Circuit

The American Educational Research Association Amicus brief filed in the Supreme Court case tried to undermine my research with Mark Ramseyer and Jeff Standen.  Here is the relevant discussion from their brief.

It is true that only two universities were examined, but most of the rest of the statements are misleading or wrong.

1) While both schools tended to have relatively low total African-American enrollment, the point of the study was to examine how students did as they went from one class to another.  In some of their classes there were few black students, but in others they made up fairly high percentages of the students in those classes.  So it wasn't that the schools lacked data to study critical mass.  There is a very large range of percentage of black students in these classes.

2) It is simply false to claim "their model do not incorporate other potential explanatory variables that could affect grades, such as the imposition of a mandatory grade curve or the skills tested on examinations."  The estimates accounted for fixed effects for student, class, professor and semester.  Those fixed effects will pick up all the differences across how different teachers teach and across different classes.  The paper also explains out the grade curve issue biases the results against the results that we found.  

3) Neither the brief of Scholars of Economics and Statistics nor the paper with Ramseyer or Standen "neglect" or hide the one result for Asian Americans that there is a benefit to Asians from more Asians.  In the brief it is noted: "Only for Asians was there evidence that increasing their share of the class after they exceed their share of the applicant pool increased their grades, but the effect was exceedingly small."

4) There is no other evidence that directly tests the critical mass claim.  If one believes the American Educational Research Association Amicus brief, they don't even appear to think that the "critical mass hypothesis" is testable because they don't even think that there is an identifiable "critical mass."