After Affirmative Action
Yascha Mounk and Peter Arcidiacono discuss what the practice of affirmative action actually looked like, and how the recent Supreme Court ruling is likely to shape what comes next.
Peter Arcidiacono is an economist at Duke University and an expert on affirmative action. Arcidiacono served as an expert witness for Students for Fair Admissions, Inc. (SFFA) in SFFA v. Harvard.
In this week’s conversation, Yascha Mounk and Peter Arcidiacono discuss the role that racial preferences have played in the admissions processes of elite American universities in recent decades; the workarounds that universities are likely to use in the wake of the recent Supreme Court decision; and why real progress for less privileged students will require fundamental changes that look beyond the admissions practices of a few elite universities.
The views expressed are those of the speakers, not those of Persuasion. The transcript and conversation have been condensed and lightly edited for clarity.
Yascha Mounk: You're one of the most renowned scholars on affirmative action in the United States, and you come at it from the empirical perspective of an economist.
What did the system of affirmative action actually look like before the recent court ruling, and how much of a difference did it actually make in terms of who got admitted?
Peter Arcidiacono: We have holistic admissions. And that, in part, was put in initially to discriminate against the Jews because they were doing too well on the academic front. But it really got reinforced by the Supreme Court decisions, particularly in the two Michigan cases.
They had an explicit point system: you got so many points for your LSAT score, so many points for being of a particular race—the court said they can't do that. What you have to do, instead, is consider it as part of their “whole person review.”
Mounk: And one of the parts of the ruling, if I understand it, is also that they only allowed a particular kind of justification for affirmative action. From a normative perspective, in my mind, the strongest justification for some form of affirmative action in the United States is specifically tied to historical injustice, given slavery and Jim Crow and some desire to rectify it. But the Court sort of rejected that premise, and found permissible grounds for diversity under the idea that universities could try to create this diverse student experience to somewhat improve learning outcomes, which in my mind sort of makes minority students a means to an end, rather than them being there for their own gain. And the beginning of the American discourse of diversity comes in part from this ruling, as I understand it.
Arcidiacono: That's right. And it basically created a whole academic literature designed to show the benefits of diversity in order to justify affirmative action, when the reason people were supporting it was for reparations. And you're totally right, it disproportionately favors recent immigrants and those from wealthier families. It's disproportionately biracial. And even the way the preferences operate, it favors advantaged African Americans much more than disadvantaged ones. Part of me feels like with affirmative action, it was a Band-Aid that allowed you to cover it up. Because we're able to say, look, Harvard looks racially representative, everything's okay. But our school systems have fundamentally failed African Americans. Our society has fundamentally failed African Americans, to the point where 1% of African Americans get above a 1390 on the SAT, while that figure is 8% among whites and 25% among Asian Americans. They’re already at a big disadvantage.
Mounk: Let's get into what the system actually was.
How big a factor was race? Was it that sort of small thing that would tip an application over the edge, or did it make a more fundamental difference?
Arcidiacono: It makes a huge difference. And as universities get more and more selective over time, you see that the preferences have to be bigger and bigger. We actually saw that on the legacy side, legacies might initially be twice as likely to get in as non-legacies. Now, legacies are more like four to five times as likely to get in.
Racial preferences are largest for African Americans. They're about half the size for Hispanics. Two thirds to three quarters of African American admits at Harvard would not get in absent the racial preferences. So that's a stunning share. And what's remarkable about that is that's Harvard. What UNC faces is that their top black applicants who might have applied to UNC out-of-state, they get into Harvard. So for UNC, 9 out of 10 African American admits from out of state would not have been admitted absent the racial preferences. But this is at the elite schools. Most students actually go to not as selective schools, and affirmative action fundamentally affects where people go to college, not whether they go to college.
Everyone talks about affirmative action. They know that race plays a role, but they never knew how much. And for me, when we think about affirmative action in that regard, there ought to be an optimal amount. I think everybody actually believes that if you go too far, you're going to have some mismatch, at least in some areas. And the question is, where is that point? Everybody wants to go to the school that's a little bit better than they are, and that's going to benefit them. At some point, though, you can get pushed so far that you can have negative consequences. And those negative consequences will really depend on what you want to study. So it matters a lot more in STEM fields. And the reason for that is STEM fields build upon your previous preparation.
Mounk: And it's impossible to get anything below a B plus in the humanities or social science subjects at American elite schools.
Arcidiacono: And what's remarkable about that is from an economist's point of view, that emerges because of supply and demand. Lots of people come in wanting to major in STEM courses, because that's where the money is. And universities effectively allow these other departments to bribe them to leave those fields by offering them higher grades and lower workloads. And the irony is that (I have another paper on this) that actually really hurts women, because women tend to value grades more than men. So you could actually do something to close the STEM gap just by saying, look, we need to equalize grades across fields. Why should it be the case that it's so much easier to get a good grade in a humanities course than a science course? I don't think it's that way in high school, where you’re captive and everyone has to take those courses.
Mounk: Take us through the empirical evidence here. I take it that you err on the side of debate that says that there has empirically been a significant mismatch in the United States for the past decades, and, therefore, negative consequences for many students who benefited from affirmative action.
Talk us through why we should believe that. What's the evidence on your side and how would you respond to the people within that debate who disagree with your conclusion?
Arcidiacono: I think that it depends, again, on what outcome you're looking at. At Duke, once you condition on gender, African Americans were more likely to want to major in STEM and economics than whites, by just a little bit. But they switched out at a way higher rate. Over 50% of African American men who started in STEM and economics switched out, while 8% of white men switched out. And once you conditioned on differences in academic background, that all went away. So it was not that the sciences were like, “Oh, you're black, get out of here.” Except that that was being done from the fact that the preparation matters, and they're coming in with much worse preparation, both because of affirmative action and because of what's happened prior to college.
There's one paper by Zachary Bleemer, who argues that minority students lost out as a result of removing affirmative action in California. Two comments about that paper. The first is, that's not dealing with the optimal amount of affirmative action—that is moving from not having it to having it. When he says that after racial preferences were removed, earnings for underrepresented minorities went down in California—that result is completely driven by Hispanics. You don't see any effect on the earnings for black students. Why does that matter? Black students are getting the biggest preferences. That shows that they'd crossed that point where they're now not getting those benefits that went that far. If black students had been given the preferences that Hispanic students had gotten, maybe it would have been a win. Some affirmative action has always got to be better than no affirmative action for the group that it's helping.
Mounk: One of the things very briefly alluded to is the effect that affirmative action is having on schools towards the middle of the distribution. As I understand it, part of what happens through affirmative action is that students who would have ended up at pretty good schools end up going to very, very good schools. Students who struggle academically, who just about make it to college, who make it to non-selective colleges, sort of stay in that range of schools.
So how does that affect the large number of schools across the United States that are somewhat selective, that are in the middle of the distribution?
Arcidiacono: UNC loses out because Harvard and Duke practice affirmative action. They lose those minority students who would have been a great fit for UNC. To the extent that there are benefits of diversity, I think it really matters if people are coming in with more similar academic backgrounds. So some of my work has been on how you would ever measure this, and we're looking at cross racial friendships. And if you end up so far apart on things like SAT score (and it's not like people are going around with SATs scores on their heads), but fundamentally, you know, you sort the study groups, you sort into classes, and then you end up in a more segregated environment, when you have very aggressive affirmative action policies.
Mounk: So the argument here, just to draw it out a little bit, is that there's homophily; this is probably the most stable research findings in sociology in the last 100 years—like attracts like. There are lots of dimensions of similarity, and so students of a similar strength might naturally gravitate towards each other in class, they might take more similar classes and be exposed to each other in that kind of way. And so if you have members of different ethnic groups more closely academically matched, there's going to be many more friendships between those groups. Whereas if you're in an environment where, because of affirmative action, those groups are less matched than they otherwise might be, it actually might be driving some amount of segregation on college campuses. Did I understand that argument correctly?
Arcidiacono: That's exactly right. And one of my studies actually showed, if you look at the share of same-race friends that black students have in high school, it looks about the same as the share of same-race friends in college. But it should have been a much lower share of same-race in college because the share of black students is much smaller in the colleges they're attending than in their high schools. And once you account for that, college is actually leading to less mixing than high school. That's stunning to me, because colleges advertise themselves as places where this mixing is going to occur.
Mounk: That is really interesting. One of the things I'm struck by in this debate is that, because nearly all journalists at the top outlets in the United States were undergraduates in top universities, the debate is nearly exclusively about what happens at the top universities. And because every parent is obsessed with getting their kid into a good college, undergraduate admissions loom much larger than postgraduate admissions. What does all of that end up looking like at the post undergraduate level before the recent decision by the Supreme Court?
Arcidiacono: Certainly for law school, we've had some good data on that. But even then, universities don't like you looking at their data. And it's clear that law school admissions, in my view, are actually very formulaic. If you look at a combination of your LSAT scores and your grades and form a kind of ranking based on that at the University of Michigan Law School, the median black admit (in terms of that combination of LSAT scores and grades) would fall at the second percentile of the white distribution. It is a feature that you would be at the bottom of the class.
Mounk: That would mean that 98% of white admits, according to what you're saying, would have stronger academic backgrounds than the median black admit?
Arcidiacono: Exactly. And there was that case at Georgetown, where that instructor was caught on video lamenting the poor performance of her black students, and the video is a little cringeworthy. But the university decried it as racist. And everyone said, “Look, this is horrible. We need to reevaluate everything.” But the reality, and we actually have data on this, is that this is a feature of affirmative action: affirmative action may still be better for your long-run outcomes, but it cannot be better for you in terms of your class rank. By definition, you're moving yourself up, with stronger peers, so your class rank has to fall.
And that's part of my issue with affirmative action. Those students at Georgetown, they were getting big bumps for their race, and they came out of Georgetown thinking the system is racist. And that’s because universities are so dishonest with how the policies work. And you can see that in the Supreme Court debate—Justice Jackson talking about race being a factor, like it’s a tiebreaker. But it's not a tiebreaker. And universities just don't want to say that because they don't want to hurt people's feelings. But that has negative consequences.
Mounk: Let me play devil's advocate for a moment here. If these students had this ambition to become lawyers and doctors end up less likely to be able to fulfill the dream because they end up being overselected into a university where it's more competitive, but if law schools and medical schools also have these very aggressive affirmative action procedures then, actually, these students might not be harmed in terms of their career path. If you also see some elements of that and in hiring practices and residencies for medical school or hiring for law firms, then this consistent chain of strong forms of affirmative action might, in fact, create the kind of thriving middle and upper middle class of people from ethnic minority backgrounds that you might wish for on the kind of reparations justification, which I think is a stronger justification than the diversity justification.
Now, obviously, that might also come with all kinds of negative effects. But in a way, if you just keep doing it consistently, some of these problems might, paradoxically, be remedied.
Arcidiacono: And I think you're right on that. Research has shown that affirmative action definitely happens on the lawyer side, but it happens right out of law school. When you become a partner, they're not going to do that as much. Law firms are always looking for the cheapest way to satisfy diversity pressures. And so the easiest way to do that is to hire the rookies but then not make them partner.
Mounk: One element of all of this that we haven't talked about is what's at the heart of the lawsuit in which you served as an expert witness. We've talked mostly so far about white applicants, African American applicants and, to some extent, Hispanic applicants. But of course, one of the paradoxical features of the system has been that it has really made it harder for Asian Americans to be admitted to colleges. Perhaps you can tell us a little bit through the evidence, specifically, of those forms of discrimination against Asian Americans.
Arcidiacono: That was a pretty big shock to me, because we just don't have very much data on what's happening. But with the Harvard data, you can. And what you see is that Asian Americans lose out across the board—they lose out to whites through legacy and athlete preferences; they lose out to Hispanics and African Americans because of racial preferences. And that operates in a few ways. They rank them lower on the personal score. Harvard gives you a ranking on a personal rating. But there's actually another personal rating in the data as well, because most applicants get interviewed by alumni, and the alumni give you a personal rating, too. And one of the big differences is that the alumni interviewer actually meets the applicant, and, through that meeting, you can overcome those stereotypes. If you had asked me before looking at the data where the discrimination would occur, I would say it would be among the alumni interviewers. But it's nothing in comparison to what you see from the admissions officers.
One case in particular haunted me throughout, because I actually read a bunch of these files. And this was an Asian American woman who spent time in foster care, whose dad had severe mental illness, who got hit by a drunk driver the summer before her applications. And she talked about it in her essays, and the letter writers talked about it. The only comment that the admissions officer wrote on her file was “SS.” And SS means “standard-strong”—good, but not good enough. And this candidate got the normal score on the personal rating, the one that everybody gets, unless they really distinguish themselves. But the alumni interviewer said, “This is one of the best interviews I've ever had.” They were so impressed with all that this person has overcome.
Mounk: Before we go into details here, something that I'm always struck by in politics is our ability to ignore the suffering or the injustice that is politically inconvenient. I remember some figure from something I read a while ago, and I may be getting it wrong, but I remember the personality rating for Asian students being about a standard deviation lower than the personality ratings for non-Asian students. Basically, what Harvard University is implicitly saying, in a very non-public way, is that the average Asian applicant to our school has a way, way shittier personality than applicants of any other race. Am I getting this broadly right, Peter?
Arcidiacono: Yes. There is one other rating that they don't do as well on—it’s stunning that Harvard even has this, but they have an athletic rating, and Asian Americans do the worst on that as well. The people who do the best on that are white legacies, and that's because they might be able to walk on the sailing team, that type of thing.
On the personal rating, the gap is especially big between Asian Americans and African Americans. And that's where you can sort of see it being abused as a way of also putting in racial preferences. What's interesting about the personal rating is if you're black, you get a big bump on it. If you're poor, you get a bump but not as big of a bump. But if you're black and poor, you don't get the poor bump. And that holds true in admissions as well, which to me shows this pattern of using it to get the types of students that you want: affirmative action fundamentally benefits non-poor, black applicants. And the thing is, to the extent that the personal rating is supposed to be a measure of overcoming hardship, African Americans are going to score better. But when you compare them to Asian Americans, you can't make that argument. It is true that Asian Americans now come from higher incomes than whites on average—it's not that different, but it is higher. But not among Harvard applicants. And that’s because poor Asian families prepare their kids to apply to Harvard at a much higher rate than poor white families. It's stunning to me how well that group is doing academically and on these other measures, because on extracurriculars they were beating whites as well.
But I think that there are windows of opportunity here, both because I think there is some pushback in recognizing that the non-academic factors favor the rich even more. We know that the SAT favors the rich, that's why we have this movement to ban the test because of coaching and resources and all that—that's fine. The test-based system is, I think, the fairer one. That's actually recognized in Korea. In Korea, my understanding is there are actually two ways to get into college: the test track and the non-test track. And the non-test track is known for favoring the rich. In Korea, you can point to some other issues (it’s a pressure cooker). But the US is such an outlier. In almost every other country, it's a test. And the reason I think that that's so important is trust in the system. To me, it's a really bad look that every university wrote the same email to their students with platitude upon platitude about how the ruling was not good, but they’re still committed and so on. But the public approves of the decision. The trust in higher education has been falling, and steps need to be taken to correct that. I think that's why you're gonna see things like legacy admissions fall.
Mounk: When you zoom out a little bit, what's striking is how extremely similar all these different universities are. By and large, in every dimension, the top 25-30 American schools are incredibly similar to each other. Why is that? Is there any hope for going in a different direction?
If we have some university presidents listening to this, what would you encourage them to do? Consulting with their values and best stakeholders, which direction should they go in, if they actually have the guts to engage in an experiment of self-government here?
Arcidiacono: What universities fail to do continually is to use their data. What's remarkable is that they have the experts on this stuff, and they don't use them, because they're afraid of lawsuits and such. And that really is damaging to students.
Randomized roommates is a great example of this. Randomized roommates is great if you want to study how roommate characteristics affect future outcomes. Well, most of the time they don't study it, sometimes they do. But they never implement any policies based on it. If you were actually using your data, you could figure out how to help your students. Competition for those top-scoring black students is already fierce. Now, it's gonna get fiercer. If a university could say, “Look, with our data, we can show you we’re actually really good for you in terms of your later life outcomes, in terms of your satisfaction. We know what professors are especially helpful for you, and we're gonna make an effort to improve that experience.” I think that could be huge. I feel like part of my reservations about affirmative action are that the universities care about getting them in the door, so that they look diverse, but then they stop. And I think that's because they defer to the faculty. That's why you end up with these majors being able to bribe these students into the low-paying fields, which I don't think is good for the black students.
Mounk: I realize, as we're nearing the end of this conversation, that I'm guilty of something that I often complain about, which is that we obsess about the world of extremely selective colleges and universities. But when you're actually thinking about socioeconomic opportunity in the country and lifting the prospects of minority students and black students, in particular, what's probably more important is the quality of community colleges and less selective colleges and universities where the huge bulk of students of any race in the United States go.
Zooming out beyond the world of elite universities, and perhaps even beyond the world of higher education as a whole, what are a few of the things we can do to increase the pool of highly academically prepared students who are available for admission at selective universities?
Arcidiacono: My philosophy, I think supported by the literature, is the earlier the better. It's actually remarkable, when we were having our kids, how much time was spent on preparing for the birth and not so much time spent on what are you going to do with this kid afterwards. Those first four years actually are incredibly important. When we get to the elementary school stage, we actually know what's fairly successful. These no-excuse charter schools—Roland Fryer has got some great work on this. He was actually one of the few who shows how to close these achievement gaps. He's now been read out of the profession, but he would be an amazing person to talk to on that front. Fundamentally, those schools by putting in the extra time and such are able to make up for the disparities that happen early on. And I think that's actually the path so that we don't need affirmative action in the first place. We have to address the skills gap up front. On the higher education side, states could invest more money into colleges where those students are actually going. YYou could see a liberal state shift their funding model in response to the ruling so that we better target the college resources for the schools that minorities are actually attending.
Please do listen and spread the word about The Good Fight.
If you have not yet signed up for our podcast, please do so now by following this link on your phone.
Podcast production by John T. Williams and Brendan Ruberry.
LinkedIn: Persuasion Community