One of the contentions of the “Mismatch” critique of affirmative action is that, by moving students who benefit from racial preferences in admissions into more highly selective colleges than they would otherwise go, such students are less likely to complete STEM majors than if they had entered colleges where their skills were closer to the median. On the other hand, stronger black students enrolled in HBCUs or in, for example, the Meyerhoff scholars program at UMD-Baltimore , where their skills were above the median, were much more likely to complete STEM degrees. (For example, the students who attended UMD-Baltimore on a Meyerhoff scholarship were ten times as likely to get a STEM PhD and twice as likely to become a doctor as students who were offered the scholarship but turned down the program, most of whom went to more elite schools.)
A quick look at the students in the nationally representative High School Longitudinal Study’s 2009 cohort, followed up in 2016 when they would have been on track to be seniors in college, provides tentative support for this hypothesis. The probability of being a STEM major rises monotonically for both black and non-black students in colleges categorized as unselective or moderatively selective with students’ measured 9th and 10th grade math scores:
For students in colleges categorized as highly selective, however, this relationship disappears for higher achieving black students, who are no longer more likely to complete a STEM degree than lower achieving students.
Aside from whether grading standards are more or less stringent in the most selective colleges, students are very sensitive to their ability compared to their immediate peers. Even students who would probably benefit (intellectually or financially) from sticking it out in a quantitative major may drop into a non-quantitative major if they feel like they are lagging behind.
My sense is that selective colleges have little incentive to discourage this process. While students who completed quantitative majors may make more money (and contribute slightly more to the college endowment) in the long run, my view is that colleges favor affirmative action because it conveys moral legitimacy to intrinsically elitist institutions. Moving affirmative action’s beneficiaries into more highly politicized and ideological majors and away from quantitative majors has the side benefit of making them more effective advocates for the current system.
(For students with the highest expected long-term earnings, however, Asian and white males, the college has no incentive to make lower-earning humanities degrees more appealing, since that would draw them away from the technical degrees with the highest upside risk, and a portion of these highest earning alumni will be the bread-and-butter of donations going forward. My guess is this is part of the reason calls to depoliticize English or anthropology programs or otherwise make them less lopsidedly female fall so flat.)