The Washington PostDemocracy Dies in Darkness

Opinion Affirmative action in college admissions doesn’t work — but it could

The Harvard University campus in Cambridge, Mass. (Charles Krupa/AP)

Roland G. Fryer Jr. is an economics professor at Harvard University, a fellow at the Manhattan Institute and founder of Equal Opportunity Ventures.

I benefited greatly from affirmative action. I very much hope my children will not — because the current system contains significant weaknesses. The fate of the American Dream depends on reforming it to make it work.

On Monday, the Supreme Court will hear oral arguments in an affirmative action case involving Harvard, where I am a professor. Many people who are concerned about racial representation at elite institutions fear that the justices will end the practice as we know it. But if they do, they could provide an opportunity to create a new, data-based system that would truly help level the playing field for disadvantaged kids.

I was raised, in part, by my father, who was sentenced to eight years in prison when I was in my teens. He never emphasized education — he beat me up more times than he read to me. I didn’t meet my mother until I was in my 20s. While my test scores were stellar in the early grades, I became indifferent to school and angry at the world by the time I took the SAT in high school. My score was well below the national average, partly because my only goal was to score the 700 minimum for college athletic participation, and partly because I took the test while still drunk from a party the night before.

I needed help. I needed someone to see flickers of promise in me — even though I couldn’t recognize them myself. I needed an opportunity to show that were it not for the drama at home, my full-time job, biased teachers and the anger boiling inside me, I could compete with anyone.

But for my college professors’ willingness to look beyond my past performance — but for affirmative action — I would not have benefited from twice-weekly 7 a.m. meetings with the economics professor who showed me how science could be used to help people. Or the statistics professor who marveled at my stories of my favorite uncle — a wino with sophisticated strategies of betting on Greyhound races — and helped me use formal models to explain his behavior. Or a spot at the American Economic Association’s summer school for minority students.

But affirmative action is very often not targeted at individuals who, because of disadvantage, are achieving below their potential. Seventy-one percent of Harvard’s Black and Hispanic students come from wealthy backgrounds. A tiny fraction attended underperforming public high schools. First- and second-generation African immigrants, despite constituting only about 10 percent of the U.S. Black population, make up about 41 percent of all Black students in the Ivy League, and Black immigrants are wealthier and better educated than many native-born Black Americans.

How can affirmative action be made consistent with true meritocracy? How can the minority applicants who have lower scores but high potential be distinguished from those who just have low scores?

The answer is simple: Use data more rigorously. The problem is not affirmative action per se. The problem is lazy implementation of it as a set of blunt racial preferences.

If racial diversity is the only goal, then explicit racial preferences — which are already illegal — are the most efficient way to achieve it. Mathematically, the optimal policy amounts to treating each race as a distinct applicant pool, and then admitting the top students from each one.

Under this approach, students of all races will disproportionately come from wealthier backgrounds, because grades, test scores and other factors used in admissions are positively correlated with parental income.

What if, instead, admissions involved the use of sophisticated analytics to predict which applicants stand to become the future top performers?

Evidence is accumulating that machine-learning algorithms can make better decisions than humans, especially when bias is present. A Cornell study of data on judges’ bail decisions found that computer predictions could reduce crime as much as 25 percent with no change in jailing rates. A Stanford University study found that a machine-learning algorithm performed as well or better than trained radiologists on screening X-rays. Why not use this approach in college admissions?

A machine-learning model would be fed historical admissions data, including candidates’ family background and academic achievement, and noncognitive skills such as grit and resilience, along with outcomes of past admission decisions. It would use these data to predict new applicants’ performance — as defined by each institution, such as college grade-point average or income 10 years after graduation. The model could figure out which characteristics best predict performance for various subgroups — for example, how salient SAT scores are for public-school Black students raised in the South by single mothers vs. private-school White kids from the Northeast. If we use only unadjusted test scores, all that context is lost.

This approach would sidestep other thorny issues in college admissions as well. If, as some have argued, the SAT test is racially biased, the algorithm could place less weight on scores for Black students. If students who demonstrate resilience as children are more likely to be hidden gems, the data will show that, too.

The Supreme Court seems poised to strike down the explicit use of race in university admissions. My hope is that it will still leave room for data-driven approaches to affirmative action that ensure real meritocracy.

Loading...