Racial Preferences By The Numbers: Two researchers lay out the data on affirmative action in college admissions

January 2010

It’s hard to get a straight answer as to how pervasive racial preferences are. On the one hand, many academics say preferences hardly even exist — they’re just a tie-breaker that admissions officers use on rare occasions. On the other hand, the same academics often say preferences are crucial to diversity, and their elimination would wreak havoc on campuses nationwide. Perhaps nowhere has this bizarre contradiction been on starker display than in No Longer Separate, Not Yet Equal — a book that manages, despite this contradiction, to shed light on various controversies in higher ed.


Using the National Study of College Experience (NSCE) - a collection of information from eight anonymous elite colleges — authors Thomas J. Espenshade and Alexandria Walton Radford are able to calculate various applicants’ odds of getting into a school. They discover some mildly interesting trends regarding social class (more on that later), but their results for race are truly stunning. After academic performance and demographic factors have been taken into account, black applicants are more than five times as likely as whites to be accepted at NSCE private schools, and 220 times as likely to be accepted at NSCE public schools. Asian applicants, meanwhile, are only about a third as likely as whites to get big envelopes from private institutions, and one-fifth as likely to gain admission to public ones.

Putting preferences in terms of test scores, at private schools, blacks get an advantage, compared to whites, worth 310 SAT points (out of 1600), Hispanics an advantage of 130, and Asians a disadvantage of 140. At public schools, the authors present the difference in ACT points: blacks 3.8 (out of 36), Hispanics 0.3, Asians –3.4.

If we look at students who actually matriculate, blacks are far more likely than whites to come from the bottom 80 percent of their high-school classes (27 percent versus 12 percent), have high-school GPAs of B+ or below (32 versus 18 percent), and have SAT scores below 1000 (21 versus 2 percent).

The logical conclusion from this mountain of evidence is obvious: Top-of-the-line schools use severe racial preferences. This shouldn’t be all that shocking; although colleges usually keep quiet about the degree to which they prefer blacks and Hispanics over Asians and whites, anecdotes and numbers have been trickling out for years. Even when California banned racial preferences, its state universities didn’t stop using them. Last year, a UCLA professor resigned from the school’s admissions committee in protest of its flouting the law and issued an 89-page report explaining his reasons. Few schools outright deny using preferences, and the Supreme Court allows the practice. The Center for Equal Opportunity has calculated the extent of countless schools’ preference policies, usually concluding that black and Hispanic candidates get a significant advantage.

But the authors resist this conclusion. Espenshade told an interviewer for the Inside Higher Ed website that he doesn’t have “smoking gun” evidence that Asians are discriminated against, claiming that factors he wasn’t able to include in his analysis — letters of recommendation, etc. — might have been so much worse for Asians that they explained the gap. The book makes a similar argument about blacks and Hispanics, going so far as to bust out the old tie-breaker meme in this jawdroppingly absurd passage:

It would be a mistake to interpret the data . . . as meaning that elite college admissions officers are necessarily giving extra weight to black and Hispanic candidates just because they belong to underrepresented minority groups. This may occur from time to time, especially in situations where two applicants are otherwise equally well qualified. But in our judgment, it is more likely that a proper assessment of these data is that the labels “black” and “Hispanic” are proxies for a constellation of other factors in a candidate’s application folder that we do not observe. These unobserved qualities — perhaps having overcome disadvantage and limited opportunities or experiencing challenging family or schooling circumstances — may be positively correlated with the chances of being admitted when a holistic review of an applicant’s total materials is conducted.

In the very same chapter, however, the authors mention “the black advantage” and refer to the disparities as “weight” and “preference.” They also note that at NSCE private institutions, students who are minority and poor get a sizable boost, whereas students who are white and poor actually get penalized — that’s how much admissions officers care about helping those who have “overcome disadvantage,” as opposed to engineering their schools’ racial balance.


Eventually, the mask comes off. The authors ask: What happens when you remove racial preferences, and what alternative policies are available to those who advocate diversity and/or the redressing of racial inequality?

It turns out the authors don’t really believe their statements that race isn’t a factor in admissions; if officials stopped considering race, they predict, minority enrollment would decline precipitously. If NSCE private schools eliminated both the black/Hispanic advantage and the Asian disadvantage, blacks would go from 8 to 3 percent of these colleges’ admittances, Hispanics from 8 to 5 percent, whites from 60 to 53 percent, and Asians from 24 to 39 percent.

Is there any way to have it both ways — to find a policy that uses legitimate, nonracial criteria, but that achieves results that liberal, race-obsessed admissions officers can live with? The authors consider an idea that has been around for a while and has gained some ground on the left recently: replacing race-based with class-based affirmative action.

What many advocates present as a panacea turns out to be of little help. While class-based affirmative action brings in more minorities than a race-neutral policy would, the numbers aren’t impressive. If schools eliminated racial preferences, instead giving “lower-class” students the weight blacks currently get and “working-class” students the weight Hispanics currently get, black admittance would fall from 8 to 4 percent and Hispanic admittance from 8 to 6 percent. The only way to achieve current levels of diversity with this system is to completely eliminate test scores, GPA, and high-school class rank as considerations.

There’s still a good case to be made for the class-based approach: If we’re going to use our college-admissions practices to try to combat economic inequality, we should base our preferences on actual economic disadvantage rather than on skin color. But those who advocate ethnic diversity, and those concerned with racial economic inequality as opposed to economic inequality in general, will not be happy with a class-based approach.

The authors also consider “10 percent plans,” in which schools would automatically admit students who ranked in the top 10 percent of their high-school classes. However, that doesn’t work either: Under these plans Hispanic enrollment would stay constant, but black enrollment would fall from 8 to 4 percent.

The only realistic way to keep diversity without preferences, the authors say, is to eliminate the achievement gap at lower levels of education. In an article for Inside Higher Ed, they suggest a “New Manhattan Project” toward this end. Good luck to them, but the notion that we can dramatically increase kids’ test scores by improving their schools warrants skepticism.

One important thing to bear in mind is that the authors’ sample — the elite schools in the NSCE — is not representative. Without affirmative action, the minority students who failed to get into NSCE schools would likely go to lower-tier schools rather than skipping college entirely. It’s hard to tell what would happen at those lower-tier schools. After California banned preferences, black enrollment at its elite schools dropped significantly, but black enrollment at other schools didn’t change much. (Of course, the caveat here is that California administrators didn’t fully comply with the law.)

Still, the conclusion is inescapable: We cannot reconcile high-end colleges’ desire to enroll substantial numbers of blacks and Hispanics with the public’s opinion that racial preferences should be illegal. One side must win, and the other must lose.


The question of which side should win is, of course, a highly contentious one. Perhaps the most powerful argument against affirmative action was put forth by UCLA School of Law professor Richard Sander in his study of law schools. Sander found that affirmative action brings students into schools that are too demanding for them. As a result, they’re more likely to achieve poorly and eventually drop out. In the end, affirmative action actually decreases the number of black lawyers that law schools produce. If this is true, and if a similar process unfolds at the undergraduate level, it essentially ends the debate. Even if admissions departments really believe what they say about the benefits of diversity, they’ll have trouble convincing anyone that achieving it is worth hurting minorities.

Unfortunately, the NSCE data do not provide a good chance to test this theory, because the study includes only elite schools. To perform the same kind of analysis that Sander did with law schools, one would need to compare similar students who went to very different institutions.

Nonetheless, the authors are able to divide their universities into three tiers: those whose students have an average total SAT score of above 1400, those with an average score between 1300 and 1400, and those with an average score below 1300. The nationwide average for a college-bound high-school senior is just above 1000, so these categories don’t reflect the full range of colleges, but they do allow the researchers to figure out whether comparable students fare worse at the more demanding schools. Looking at graduation rates, they find the opposite: Students who go to more selective NSCE schools are actually more likely to get diplomas.

This is certainly notable, but the authors also find good evidence that affirmative-action students perform differently from their peers. Compared with whites, blacks and Hispanics are more likely to choose social-science majors and less likely to choose natural-science ones, while Asian students are overrepresented in natural sciences and engineering. As for time needed to graduate, 57 percent of blacks and 71 percent of Hispanics finished college in four years, compared to 80 percent of Asians and 75 percent of whites. Even after six years, 22 percent of blacks have not graduated, as compared to about 10 percent each of Hispanics, Asians, and whites. When they do graduate, half of blacks and a third of Hispanics rank in the bottom 20 percent of their classes; the authors estimate that they’d have ranked higher had they gone to less selective schools.

So, while affirmative-action students may not be hurt by going to NSCE schools — their likelihood of graduating seems to get higher as they attend more demanding schools, at least within the NSCE, and their classy degrees presumably help them in the job market more than their lower GPAs hurt them — they don’t perform as well as non-affirmative-action students. This must be counted amongst the costs of affirmative action at elite schools: The students admitted out of preference will need more time to earn their degrees, and will achieve less in doing so, than students admitted on merit alone.

This is a big book, exhaustively researched and packed full of facts, numbers, and prose. The authors weigh in on a number of additional topics, giving statistical snapshots of NSCE schools’ applicants, accepted students, and matriculants and discussing how students pay for school and how often they interact with peers of different races. Whatever its problems, No Longer Separate, Not Yet Equal is a must-have reference for everyone who pays attention to race and class controversies in higher education.