Peter Arcidiacono on Harvard Admissions Bias

Welcome to the Policy of McCombs podcast, a data driven conversation on the economic

issues up today in this series. We invite guests into our studio to provide a highlight
of their work presented during a visit to the University of Texas at Austin Policy.
Emma Combs is produced by the Center for Enterprise and Policy Analytics at the McCombs School of Business.
I am your co-host, Carlos Carvalho, with my colleague Mario Villarreal.
Our guest today is Peter r-s.d Iacono, professor of economics at Duke University. Peter joined us
today to talk about his work in association with the Students for Fair Admissions versus Harvard. A recent high profile
case challenging Harvard’s admissions policies as potentially discriminatory against Asian-Americans.
Peter, welcome to Policy McCombs. Thanks, Evan. So let’s start by
describing generally, I guess, your role on the case as an expert witness in and how would you characterize
your general findings? Well, my role is to take
Harvard’s data and analyze it and look to see whether there was discrimination against Asian-Americans.
And also to sort of measure the size of of racial preferences.
So with that, we have access to incredibly rich data. You know, it’s six years
of Harvard’s Harvard’s admissions data. Was that a reason for six years
that I don’t know why they gave us that particular amount? There was harvest.
choice or the judge against the courts decided it was. I think Harvard’s case to give us
no. But beyond that. And so in
how would you describe that your gender, sort of your top line summary of your report
and that you presented in court? So I think both
myself and David kardon on the other side, you know, sought to measure the
degree of racial preferences. And that was really not the focus of the of the trial itself. I think
people had come to different conclusions as to how begue racial preferences
should be. But what is interesting about it in this case is this is your first
chance to really see how big they are, because typically universities
are not going to show this kind of data. So is really a unique opportunity there with the trial
primarily focused on was whether Asian-American
applicants were being discriminated against relative to white applicants. And for
me, the fiscal arguments for that are pretty over overwhelming.
You know, I find a penalty for Asian-American applicants who
are not in one of these special groups and as a glorified those groups who are
well, the shorthand is a l._d._c, which is athletes legacies,
children of donors and children and faculty and staff. That group represents
applicants. So for the vast,
vast majority of Asian-American applicants, I’m finding this up a penalty relative
to similarly situated white students. That’s sort of above and beyond
the fact that they also lose out because of things like legacy
and athlete preferences, preferences for those LDC groups. So just
to summarize that, if you put everybody together and if you were to run a regression,
let’s say controlling for lots of different, different characteristics
of the applicants, everybody together, it is clear that you see a negative coefficient and
impact something that suggest a negative impact for Asian-Americans regardless
over. But on top of it, if you separate now and look just at the students that are being considered
in this for the general population, not an athlete’s legacy, dean’s list, etc.,
that bounty is even larger relative to whites. Yes. Now there are ways,
as you know, the other side did. You can do some things to make
it appear as though the penalty is smaller. But
I think a correct treatment of the data shows that a clear penalty. So the data
side can ever show that the penalty goes away. For example, they can make it go in significant,
I say, by including everyone together,
including the athletes and legacies and so on, including the personal rating where I think
there’s clear evidence of bias against Asian-Americans. And
I think that part of what what’s happening here is that when you include athletes and legacies
and such and they actually show this in our paper,
academics just are not as important for legacies and athletes. Asian-Americans
do incredibly well on academics. And so if you do things that make it, if
you add people to the model, which shrinks the importance of academics, right.
Then it’s going to say, well, no, we’re not discriminate against Asian-Americans, academics to start as important. Well,
that may be true for a l._d._c applicants, but it’s definitely
not true forever. Else, could you spend a little bit on on on that,
because that’s one of the most common reactions I’ve seen among various people that say, look,
I mean, on in a ratings for our emissions process, we take
into account various dimensions and we weighed them in various ways. And it
just happens to be that order applicants are there are non Asians like whites,
namely rank very well in this disproportionate weighting, other
dimensions like extracurricular activities or sports
or things like that. Could you elaborate a little bit about what is your reaction to that position? I
guess maybe useful to talk a little bit about the four categories because they’re very clear here. So let’s fly the four categories that
Harvard has in their process. Well, Harvard has lots of categories, but, you know, they have
fought for what are called profile radio ratings or grades. And this profile ratings are academic,
extracurricular, athletic and personal. Asian-Americans
clean up on academic. It is amazing how well they do relative to white applicants,
let alone other groups. They are also the strongest on on the extra curricular
rating. So then you’re left to f. I mean, the the other side basically argued that
they’re not as multi-dimensional as white applicants. You get that from
the athletic grading and the personal rating on the personal rating side. I
think that there’s clear evidence of bias against Asian-Americans. And the argument for that
is that if you look at the observable characteristics, the things
that we can see that are associated, higher personal ratings. Asian-Americans are stronger on this.
And there are other groups who receive a massive bump on the personal rating who are weaker on this.
So that makes you think that racial preferences and penalties are playing a
role in in that personal rating. That leaves the athletic grading,
the athletic ratings. Very interesting. It’s not something that receives as much attention during the trial itself,
but all the information in the reports, it’s part of some papers I’ve written since. And
using the publicly available data, the people we do best on the athletic grading are
white legacies. And part of that’s due to
some readers gaging this based on how good you are at sports, at Harvard
offers. Well, Harvard offers sports like sailing and and so
on that are privileged, very privileged sports. I’m looking at a list
here that you have in the paper about the sports that Harvard introduced in the past since 1974
and is exactly women’s squash, women’s fencing, women’s lacrosse, women’s cross-country women’s sailing, women’s skiing,
indoor track and field, women’s soccer, women’s ice hockey. Those are all clearly very
privileged and they’re also very heavy on the female side. I assume just because for title nine.
Exactly. But that provides them again. I don’t know if you have something in mind,
the effect of gender within the categorization of the sports, that might be something that females
are benefiting dramatically as well. Right. We don’t have a lot of information on that in terms of
of what was what was reported. I will say that I think that
this athletic grading. It’s surprising that they they score them, score
them on that. I don’t believe that it does. I don’t believe that Stanford has an athletic rating.
I don’t know about other institutions. And just again, want one thing to to be very
clear front from the paper here that even if you control for the personal ratings,
you still see a negative impact for Asian-Americans
controlling the 994 non-healthy. Exactly. Once you remove the athlete’s legacy
and dean’s list students, which you have to do things that are one that is already discriminating
against Asian-Americans. And on top of it, you have an extra penalty. Exactly right.
So. OK, so let me actually use this to highlight one thing that you mentioned
here about agents cleaning up on the on the academic standards. One number
that that I’m just going to scroll here and find in the paper that you point
out is that you put students, applicants on deciles of of
academic achievement. And if you look at the top decile, how many of the Asian-American
students are the top decile of academic achieve the top 10 percent best academic students? All right.
general group of people applying to Harvard, 70 percent of the Asian-American that apply aren’t within
that group. That’s right. And that’s double what it is for white. Is double what it is for whites
and is like nine times what it is for for Hispanics and is 17
times essentially what it is for African-Americans. So as they’re getting credible quality level of students coming in
in the Asian-American pool. That’s right. And so this gives something. And this is where I think
Asian-Americans are unfairly stereotype based on that. They they do incredibly well
in the academics, but it’s not as though that they’re weak on these other than the other dimensions,
with the exception potentially of the athletic grading where they
they and Hispanics do not do very well on that. And I think that.
Yeah. I mean, what will we think about the athletic raiding thing? Well, the sports is there ready everybody
on soccer as Hispanics who do a little better. But no, that’s not what they’re radio on Thursday, the NCAA.
Look, I saw well on that. It’s also like whether or not you’re the captain of your sports, right?
Right. Show some leadership trait or something like that. That’s right. Right. So, you know, when I went to school,
there was no possibility I would make the soccer team. You know, I’m not particularly
athletically gifted. You know, my kids do go to a small private school. Everybody
makes a soccer team if they want to play. So it’s a really favors
richer schools. They’re smaller. You know, you can provide people with
lots and lots of opportunities. And that’s
really actually one of the surprising things which this paper, the legacy athlete preferences
at Harvard and illustrates is that when we think about holistic admissions,
you think about evening, the playing field, that fundamentally you think people can buy off
by higher test scores or test prep programs and such. It’s not clear
in, you know, at Harvard anyway. Some of these other ratings seem to be more influenced
by income than the academics, at least within racial
groups. So, you know, this holistic admissions, even things out across races,
but within races, things like the athletic grading, the personal rating and so on,
those are actually still favoring privilege, still favoring privileged, more so than the
things like the academic rating. Right. Right. Yeah. I think I find it I find it very
interesting about the papers and the ability to through your analysis, for us to see that changing probability
of a certain applicant, if you just were to, you know, start from a baseline and and think about change, a couple characteristics
that applicant, whether it’s the income, whether it’s the net income, but but whether it disadvantage or not,
the whether they’re an athlete or not or some degree about the ratings, you could even change change that in the ratings. So
a number that I heard and you know, we talked a little bit about this. That, let’s say an Asian-American
with a 20, 25 percent chance of admissions. That would be a very high achieving Asian-American,
probably from a middle class that currently has a 25 percent probability of getting in.
If we were to toggled that, let’s say to a white student, that probability is going to go up.
Yes. We have to look at the exact number. But to an African-American, we talked about it. What would that probability go? Over 90
percent. Over 90 percent. So I should say over 90 percent if they were both
not disadvantaged. The disadvantage, African-Americans.
They don’t get as large of a bump for being African-American as they do if their advantage,
which is sort of one of the goals of reverse where you expect. Right. Exactly. Yeah. Yeah.
And that’s that’s the key point here, is that having an admission it’s holistic to level
the playing field is one thing. But when the numbers are that big. That’s I guess, what’s being litigated. That’s what the
holistic we need to decide or figure out what holistic means, right? That’s right.
One of the main. Opposing views, too, to the point
of view that you just fleshed out. Here is David Cards at UC
Berkeley. And I believe that his main objection is that use left
out of your model, a group that if you include it, it will change the results
of these deaths. How I see the claim like so you meet it recruited athlete athletes,
children of alumni, children of Harvard faculty and staff members and students on a special list
that includes the children of donors. Those are, of course, kept
at a higher rate. So he argued that by removing them, yours,
security, the results. What is your answer to that? So
this is something we talked about briefly before, but the idea that including those groups,
it changes the relationships at the model shows. So fundamentally something like academics,
we are just not as important for legacies, legacies and athletes and such.
There. There’s another group that neither of us included which are foreign applicants. So
typically, you know, if if you thought it was important, include everybody,
then you should also have to include the the foreign applicant applicants as well.
In economics, you know, we’re really big on worrying about selection and
selection. You know, we have papers on twins. That’s not slicing and
dicing the data to get to analyzing Swinton’s. What you’re really worried about.
We’re looking at twins and the returns to college that
you’re trying to account for selection by only looking at those at those pairs and by
selection. Just be clear, you’re trying to toot, toot, toot to account for the fact that that you’re not getting a random pool
of applicants here. You’re getting a pool of applicants that are already that that
they’re different. They’re different. It’s right there. And there’s garley treated in different ways. So,
you know, if you look at the bottom 10 percent of applicants in terms of their academics.
Berg, I think virtually no one gets in if they’re a non.
l._d._c. But if you’re in a legacy and you’re white
during l._d._c applicants and you’re white, your admission rate is over 6 percent
when you’re in that bottom 10 percent. That’s higher than the average admission rate for whites and no
whites in that non ABC whites in that bottom decile. Got it.
So, I mean, that tells you the academics operate very differently for these
l._d._c applicants than they do for the non LTC applicants. And it’s even
more so for athletes. So for athletes,
if you get there, there is one academic rating where only athletes get it.
And then if you go up to a four in the academic grading, which is actually very,
very bad, their admission rate is almost 80 percent for athletes.
So we’re talking about substantial differences in
how academics matter for the different groups. I have a general
question that I don’t have one to close it up, October, go ahead. Well,
by reading your work, I couldn’t stop thinking about
you. I mean, the admission process is hard nowadays. Right.
And that relates to a conversation that Cardless and I and other colleagues have had about how
the admissions process. Maybe if it’s not broken. It may be
at a difficult point. So what could you tell us about
what you learn regarding that general aspects of
the admissions process in not only elite universities, but
in general lightly? Do you sympathize with the notion that is necessary and desirable
to have a well-rounded approach to it? Or would you say, look, I mean, it’s
about academics. Who cares if you play a sport or not?
We grabbed the best students, educate them. You’ve the scales and send them to the
labor market. And that’s our job, right? Like, oh, none at all. We need to to
maintain a level of diversity and then well-rounded citizens. So therefore, the admissions
process should take that into account because that’s where raw material. And in your idea. What will
I get that I’m asking you if you could reformed admissions process at Harvard and other places,
why would you do it? It’s funny because I think I did get criticized for putting
weight, so much weight on the academics. I don’t really feel like I did my reports. I think my
reports used a lot of the academic stuff to motivate. Well, given how strong
Asians are doing on apalling on academics. What is that?
What do they have to look like on the rest of this to justify what they’re low admit rates?
But I came out of this process much less
convinced of the value of the holistic admissions. I think
actually, if you’re going to have racial preferences, I think something formulaic could actually be better
than than what we’ve got here. I think it’s too prone to corruption.
I mean, you see it with the varsity blues scandal, of course. So those types of things,
you know, I find that this disconcerting. You know, there’s so much noise,
noise in the process as a result of different people having different perspectives on these
files. And, you know, I read the files, I read a few thousand testified
about one of them and. That’s what makes it harder to detach
from the case, not the affirmative action side, which people can have different perspectives on. But on
the Asian-American discrimination side, you know, the finally testified on
I think showed clear. Clear discrimination. We are U.T. Here have
a 10 percent rule where if you finish in top 10 percent of your any high school public high school in the state
of Texas, you’re guaranteed a spot at a public university in Texas. I think any
of your choice and then there’s a matching process to try to. Because too many might want to come to teach a certain
major and so on. That, of course, has a ratio in better racial component to it
because because the high schools are not deterred. There are different racial compositions,
but that I like the idea of a clear rule. Nobody’s messing with it. It just like,
you know, you might disagree with it. The level, whether there’s some reforms to that rule. But it’s a very clear rule. Everybody
knows what the game is for that. Now, that’s for 75 percent of our students, 25 percent
of our students come from something that is holistic and has been litigator, although the Supreme Court and that’s what the holistic
word comes from his foes. Now, I think they come from Michigan. Right. That was the first decision was against Michigan.
Yeah, Michigan. The formulaic approach was ruled unconstitutional.
Do you have let’s say that the other part to it, which I think would in my mind solve some
of these issues, too, is transparency. You know, so thinking about this legacy admissions
and such. You know, people always knew that we have legacy preferences, but we don’t
know the extent. Well, now we kind of do how we do. And now
I think that’s a more honest approach. Is that okay? Lay the cards out. I
understand how things work. I haven’t seen much of a of a reaction in terms of of you to
from Harvard or even from critics saying that this is. I mean, some but very small, I think on
you have focused on the racial component of this, not so much on the like. I can believe hard, but does that
so openly and so strongly. And, you know, any calls for reforming that inside Harvard
from there, even from their boards and so on? I haven’t I haven’t I haven’t seen any any of that yet.
There’s a question, too, too, to sort of wrap up a little bit and and go back to the paper.
You do some exercises here of thinking, okay, what would happen to the Harvard class if they were to drop?
Using basically their formula of admissions made their model of ratings and so on. If they were to
drop the use of race, legacy and athletes on it and I’m assuming all you
in here, are you dropping also Dean’s preference and. No, no, I didn’t do that.
You have completed the stuff that was that was in the public. OK. OK. So. So if they
were to do that. The total number of whites in the Harvard class
would be about the same. I think you’re saying here would go from forty eight hundred to forty nine hundred in a given
year. That’s right. But it would be different, like different whites. Exactly. But it just thinks it is racial prospectuses
and athletes. Right. Right. Much, much fewer like a seasoned athlete.
The number of Asian Americans that would get in, we’ll go from 23:00 to thirty five
hundred. That’s right. That’s a huge, enormous bump. Meanwhile, the other two components will be enormous
decrease for African-Americans for 200 to 400, 13:00 to 400 and Hispanics 13:00
for 2 4 4 4, 4, 4. Not for whites. A different mix of whites is, as you say, but fit for
the other groups. And I think that the delta that you see here in Asians, it’s that’s the part. That’s
that’s that’s it. That’s exactly the the focus of the case. Right. And how justifiable
is that? That’s right. Those are I mean, I encourage all listeners due to this
paper. This paper is public. You can find it through and I will post a link to it. And
in our Web site as well. But the analysis and all the information is papers, I think gives
us an incredible insight on on what goes on in the topics at the
top institution of of of of higher education, our country and perhaps in the world.
One more question that I had something I haven’t thought about. I didn’t realize that a foreign applicants were not here. Foreign
applicants are, my guess, heavily Asian to Harvard.
You know, we didn’t really do much with that. It wouldn’t surprise me because, you know, it might this
penalty might not be a Asian-American penalty after all. Maybe actually it just straight up Asian penalty
once you put the 14 Asian applicants to it. Yeah. And, you know, I don’t know whether
that’s part of the reason that we see an Asian-American penalty.
You know, whether foreign applications have something to do with it, I don’t know. But we focused
on the domestic. And so to that end as well, when I look at the number of percentage of of
Asians, I think in Harvard over the past many, many years, that
has been that’s domestic. No, that’s interesting. That’s not actually overall. All right. All right. And
that number has hovered around 20 percent pretty steadily for a long time until
very recently. After the luck to it, one might say, came to play that they kicked up a little bit. We went up to 23
percent, I think had this last minute classes is definitely
more Asian-American. And one of the things that Harvard
did the summer before the trial was change a reader reader guidelines. Interesting.
And one of the biggest changes is in the past there was no
guidance as to how to use race in these ratings, no written guidance.
And now they make it clear that race is not to influence the personal rating. And in
their description of the personal rating, they talk about things that would be associated Asian-American stereotypes
as things that we. That shouldn’t penalize
people. Well, that’s I guess that’s a good thing that there’s some reforms or
not reforms as surreally belike they are trying to be more transparent about it. And even if
it’s a result of the existence of this case, if it even though the plaintiffs here did not then not win the first round
and might provide more, more transparency for things moving forward for sure. Peter, thanks
for joining us. Apolicy McCombs. Thanks for having me. Before we wrap up, you
can get more information in our medium page. Thanks for listening to Policy McCombs.
See you next time.