Not Catching Up: Affirmative Action at Duke University

A working paper titled What Happens After Enrollment: An Analysis of the Time Paths of Racial Difference in GPA and Major Choice by Duke university economists Peter Arcidiacono and Esteban Aucejo and sociologist Ken Spenner is creating a stir. The authors track a sample of Duke students from admissions to graduation in order to determine the effects of affirmative action.

Under one theory of affirmative action the goal is to give minority students an opportunity to catch-up to their peers once everyone is given access to the same quality of schooling. On a first-pass through the data, the authors find some support for catch-up at Duke. In year one, for example, the median GPA of a white student is 3.38, significantly higher than the black median GPA of 2.88. By year four, however, the differences have shrunk to 3.64 and 3.31 respectively.

Further analysis of the data, however, reveal some troubling issues. Most importantly, the authors find that all of the shrinking of the black-white gap can be explained by a shrinking variance of GPA over time (so GPA scores compress but class rankings remain as wide as ever) and by a very large movement of blacks from the natural sciences, engineering and economics to the humanities and the social sciences. It’s well known that grade inflation is higher in the humanities and the social sciences so the shift in college major can easily explain the shrinking black-white gap in GPA. (The authors show that grades are higher in the humanities holding SAT scores constant and also that students themselves report that classes in the sci/eng/econ are harder than classes in the humanities and that they study more for these classes).

The shift of black students across majors is dramatic. Prior to entering Duke, for example, 76.7% of black males expect to major in the natural sciences, engineering or economics but only 35% of them actually do major in these fields (almost all Duke students do graduate so this result is due to a shift in major not dropping out). In comparison, 68.7% of white males expect to major in sci/eng/econ and 63.6% of them actually do graduate with a major in these fields (this is from Table 9 and is of those students who had an expected major). White and black females also exit sci/eng/econ majors at high rates, although the race gap for females is not as large as for males. The authors do not discuss the consequences of dashed expectations.

An important finding is that the shift in major appear to be driven almost entirely by incoming SAT scores and the strength of the student’s high school curriculum. In other words, blacks and whites with similar academic backgrounds shift away from science, engineering and economics and towards the easier courses at similar rates.

I have argued that the United States would benefit from more majors in STEM fields but that is not the point of this paper. The point is that there is no evidence for catch-up at Duke and thus to the extent that affirmative action can work in that way it may have to occur much earlier.

Hat tip: Newmark’s Door.