Merit, Not Race in College Admissions

Stephen Hsu is professor of theoretical physics at the University of Oregon and director of its Institute for Theoretical Science. He is the founder of two Silicon Valley information security startups and writes the blog Information Processing.

Updated February 22, 2012, 9:24 PM

In considering Fisher v. University of Texas, let’s acknowledge a key factual point about affirmative action: We have good tools for predicting college success, and those tools work about equally well across all ethnic groups and even for rich legacy candidates.

Race-based preference produces a population of students whose intellectual strength varies strongly according to race.

In comprehensive statistics compiled as part of Duke University’s Campus Life and Learning project, Asian-American students averaged 1457 out of 1600 on the math and reading portion of the SAT, compared to 1416 for whites, 1347 for Hispanics and 1275 for blacks. There is every reason to believe that a similar pattern holds at almost every elite university in America, with some notable exceptions like Caltech. Is this pattern justifiable, or even beneficial to the students with the lowest scores?

The data show that SAT score and high school grade point average are good predictors of success at Duke for all ethnic groups, as well as for wealthy legacy students. Those students admitted with weaker SAT scores and high school grades are more likely to drop out of challenging majors like science and engineering, and less likely to earn good grades in any major.

Again, this is true regardless of the race or legacy status of the student, and raises the important question: Would students admitted through race-based preference be better off at somewhat less elite universities where their abilities are better matched to those of their classroom peers?

Detailed, course by course, analysis of student records shows that the SAT is a meaningful measure of the ability to excel at a competitive university. For example, an analysis of five years of complete student records at the University of Oregon shows that students below about the 90th percentile in math SAT scores are highly unlikely to succeed in the physics or mathematics major. We can be confident that large differences in scores, as appear above, indicate significant differences in academic ability.

Race-based preference produces a population of students whose average intellectual strength varies strongly according to race. Surely this is opposite to the meritocratic ideal and highly corrosive to the atmosphere on campus. Furthermore, the evidence is strong that students of weaker ability who are admitted via preference do not close the gap during college. For these reasons, the Supreme Court would be wise to end the practice of race-based preference in college admissions