Affirmative Action Quantified

The Study

A study using data gathered in undergraduate program affirmative action litigation at Harvard and the University of North Carolina, slices and dices the data many ways, but probably the most useful and wholistic summary of its results can be found in its Table 11:

The columns in Table 11 for "Share of Admits" and "Percentage Change" are flipped in Panel B for UNC Out-of-State students in the pre-print version linked above, but the numbers are otherwise not obviously in error.

The study is quite careful and lengthy (at 52 pages), and while there are judgment calls that go into their analysis, it is fair to accept that Table 11 is not far from the mark of being an accurate quantification of the situation.

It breaks out "out-of-state" admissions to UNC which are more competitive (and involve students paying higher before scholarship tuition bills) separately from "in-state" admissions to UNC which are more lenient and come with a lower tuition bill.

How Strong Is Undergraduate Affirmative Action

In undergraduate programs where admissions standards are very selective, no one will be surprised to learn that black students have a significant edge in admissions, that Hispanic students have a significant but smaller edge in admissions, that white students and Asian students have a harder time being admitted, and that at very elite schools like Harvard, that the admissions disadvantage of being Asian can be very significant.

At Harvard, about five out of six students admitted would be the same without affirmative action, but one out of six admitted students would be different each year.  At UNC for out-of-state students, about one in five admissions would be different. At UNC for in-state students, about one in twenty-three admissions would be different.

This isn't so different from the overall magnitude of the impact of other non-academic ability based preferences at colleges and universities with highly selective admissions like legacy and athletic considerations, although for quite a few individual applicants the amount that the "bar" for admission is impacted by race may be greater.

It is also worth noting that even in the arena of UNC in-state admissions which are sometimes described as "very selective" or "highly competitive", the affirmative action "penalty" for white and Asian students is already quite modest (it impacts the marginal five percent or so of white and Asian applicants with a more or less even impact), even though the benefit for black (about 36% of admissions) and Hispanic applicants (about 18% of admissions) is still quite significant.

In all of these cases, the actual difference in the study body would be somewhat less, because even with affirmative action, the admissions are strongly correlated and some students will be admitted to multiple elite schools, while others will be rejected by all of them.

The amount of affirmative action in UNC in-state admissions is pretty close to public opinion is comfortable with. Prior court opinions have also suggested that affirmative action of this intensity is legally acceptable to further educational interests in diversity. And the fact that UNC's affirmative action impacts white and Asian students similarly is also something that public opinion is more comfortable with. 

The amount of affirmative action in UNC out-of-state admissions and at Harvard, and the fact that affirmative action at Harvard impacts Asian students significantly more than white students, are facts that if known, would probably be outside the range of what public opinion is comfortable with and are closer calls in terms of what one might expect would be found to be acceptable based upon court opinions before these cases were litigated.

If you were to go down a notch in selectivity, you would find that the proportionate impact of affirmative action are more modest, until you reached a point of near open admissions four year institutions, and the bulk of community college programs, where affirmative action have a negligible impact.

Impact On Students Penalized By Affirmative Action

Affirmative action is often portrayed as a program that is penalizing working class whites at the expensive of middle class black and Hispanic students. But that is more wrong than right.

First, in the Ivy League at least, the typical student impacted is Asian American. 

An Asian American student among the 699 students denied admission to Harvard, for example, as a result of affirmative action, admitted with a change in the admission rate for Asian applicants from 5.13% to 7.17% as predicted, would still be mostly upper middle class, or at the high end of middle class socio-economically with immigrant parents who would be in a higher socio-economic class but for their immigrant status. Both academically and socio-economically, they are very similar to admitted students but without quite the stellar "plus factor" like inventing something or making a scientific discovery or having a record recorded, as the students who didn't make the cut.

Likewise, the 491 white students who didn't make the cut as a result of affirmative action at Harvard are also overwhelmingly upper middle class or middle class with academically inclined parents, and extremely academically talented, but lacking some "plus factor" if the admission rate increased from 4.89% to 5.78% as predicted, not poor white kids from Appalachia or low income families in rural America. 

Academic performance is more strongly correlated with socio-economic class in white Americans than that, and in the rare cases where there is a stark disparity, this is a big edge in college admissions, although not quite as great as being black or Hispanic. Even among black or Hispanic applicants, the typically successful applicant at the most elite colleges and universities is middle class or upper middle class, not working class or poor.

For the most part, the college admissions prospects of white working class and even middle class students would be almost unchanged if affirmative action were disbanded. White working class students may even get a weak plus factor themselves relative to academically similarly performing white students who grow up in middle class households in suburbia, in the status quo. They are being denied a benefit that working class and middle class black and Hispanic students are receiving, but if the system as a whole were scrapped, they would be no better off.

Equally important, no white or Asian student who would have been admitted to Harvard or UNC-Chapel Hill but for these institutions affirmative action policies will not be able to get into a very solid and reputable four year college or university as a result, although many will end up attending a less prestigious or less selective institution than they otherwise would have attended but for the consideration of race in college and university admissions.

The white and Asian students who otherwise would have gotten into Harvard may the end up going to Tufts, or Middlebury, or UC-Berkley or UNC-Chapel Hill, or the University of Virginia, instead.

This isn't nothing. Affirmative action in elite college and university admissions is a highly sensitive issue because these elite institutions are gatekeepers to entry into the non-entrepreneurial sector of the American elite in big business, government, and academia. A more prestigious college pedigree translates into significant real world benefits. 

But the benefits aren't absolute either. While the most prestigious of these institutions recruit disproportionately from the most elite institutions, stand out candidates from other institutions are frequently chosen too.

Impact On Beneficiaries Of Affirmative Action

The article goes into great detail regarding what this translates to in terms of raising or lowering the bar for admissions for students in one of these four categories, and there is a significant difference regarding where that bar is by race that really can't be explained by any other factor. But one can get too focused on that, because it implicitly is drawing a comparison between the status quo and one in which everyone benefitted from a lower bar associated with affirmative action (which is impossible), not from an alternative where it was not available or weaker.

For example, at both Harvard and UNC-Chapel Hill, even after adjusting for the impacts of affirmative action, that admissions are still quite selective as illustrated by Table 2 in the paper:

The black and Hispanic students getting admitted to Harvard are generally among the best of the best in the nation of the black and Hispanic college applicants in the year in question. Likewise, black and Hispanic students admitted to UNC are much better qualified academically than those who are not admitted, on average.

Less affirmative action would be a slight career boost to graduates of highly prestigious institutions who are admitted and graduate anyway when they are black or Hispanic, and would probably reduce racial tensions within almost all institutions somewhat (although hardly to zero).

Graduate Programs

I know from other data that for law school admissions that the pattern is quite similar at highly selective law schools except those that have abolished affirmative action (often with a large quantitative impact on non-white enrollment), although I'm not sure about the relative white v. Asian impact.  Again, however, as in undergraduate affirmative action, this is largely a matter of the prestige of the institutions to which you are admitted, and not access to the profession or law school generally.

I also know from other data (previously blogged) that affirmative action based upon race is present but much less significant in medical school admissions, although this is the one area where affirmative action probably does impact who gains access to the  medical profession at all, rather than merely the prestige of the institution one attends. 

I strongly suspect that affirmative action is similarly modest in STEM graduate school admissions (basically, because graduating from an undergraduate STEM programs places a high minimum academic floor for admissions to a graduate STEM program, leaving a much narrower range of academic ability among eligible graduate STEM program applicants).

Graduate programs in education are frequently organized on an effectively continuing education basis for teachers and school administrators with effectively open admissions for licensed education professionals (at least at the sub-PhD level), so there is probably little affirmative action in these programs.

I don't have much information or insight on other kinds of graduate level higher educational programs. 

Apart from MBA programs (about which I've seen little data), most other non-STEM graduate school programs are fairly small and often fractured among professors in particular subfields, leading to less rigid and formulaic admissions policies and less statistically significant data in any given program.

Systemwide Impact

The number of higher educational programs where affirmative action has a strong impact is really a pretty modest slice of the total universe of higher education.

Selectivity in terms of grades and standardized test scores are often quoted in terms of the 25th and 75th percentiles. If affirmative action were ended, the 75th percentile would change very little, while the 25th percentile would move up, but only fairly modestly, because most of the marginal students admitted due to affirmative action are below the 25th percentile of grades and test scores, even in the status quo at the very most selective institutions. 

The proportion of non-white, non-Asian students who are academically struggling while in college, however, would decline dramatically if affirmative action were abolished, because it causes its beneficiaries to be disproportionately the least academically able students in an institution (along with other students like legacies and athletes who are given preferential admissions).

If you look at the overall situation, in the absence of affirmative action, a lot of academically strong black and Hispanic students would go to colleges and universities that were a tier or often two tiers less selective, and a fair number of white students and even more Asian students would go to colleges and universities that were a tier more selective. 

The very top colleges and universities would be a lot more white-Asian relative to being black-Hispanic than they are today, although we would not return to an era of all white Ivy League schools. 

In the middle range of selectivity, the proportions of students by race would be less starkly different, but the people in those slots would be different. 

Mildly selective colleges and universities would have significantly more black and Hispanic students, and somewhat fewer white and Asian students. 

Total rates of four year college and university attendance would differ only very slightly, if at all.

In my view, race probably is a stronger consideration in college and university admissions in 2022 than it should be at highly selective undergraduate and law school programs. But, it also has lower stakes than most people who are extremely outraged by it, liberal and conservative alike, realize. 

Many people who are outraged by affirmative action in college admissions, especially white conservatives, also have inaccurate perceptions of who is penalized by its most strongly. They probably wouldn't care as much about the issue if they knew that it most strongly impacted upper middle class whites and Asians, while having little impact on working class whites with whom they tend to identify culturally.