Unexplored Tributaries

America in Black and White, based on an article published in the Journal of Blacks in Higher Education. These data, for freshmen enrolling in 1992, paint a different picture than the data offered in The Shape of the River. Among these universities, the difference between the average white SAT score and the average black SAT score is smallest at Harvard, which engages only in relatively modest affirmative action and is able to take the best of the applicant pool because of its reputation. But as one moves down through the Ivies, the gap increases. It becomes huge at Virginia, whose undergraduate student body is twice the size of Harvard's, and at the University of California, Berkeley, whose undergraduate student body is over three times the size of Harvard's. Even if one concedes the importance of the elite schools, their small overall enrollment means that Bowen and Bok have directed their attention only to the tip of the iceberg. Most black students, like most white students, attend the large state universities that are underrepresented in this study. It is there that both the student populations and the SAT gaps are largest.

The underrepresentation of large state universities seriously weakens the study for a second reason. Virtually every flagship state university practices affirmative action, and there is some reason to suspect that they will be less skilled at the art than private colleges and universities. It is easy to define what a color-blind program entails: ignore race. But affirmative action programs come in infinite varieties and intensities. In all likelihood the success of an affirmative action program depends on how far it goes and how quickly it can be revised in light of new information. Private colleges and universities have much better feedback mechanisms than large state schools that are beholden to legislatures and electoral politics. In those settings the heavily subsidized nature of public education makes the claim of black (and Hispanic) students for their share of public largess hard to resist. Schools will tend to lurch too far in the direction of affirmative action, then too far the other way.

Some evidence for this hypothesis comes from Texas' reaction to the judicial invalidation of affirmative action under the Equal Protection Clause of the 14th Amendment. Surely (as Bowen and Bok are inclined to agree) the new system, which guarantees admission to students who finish in the top 10 percent of their high school class, is far worse than the one it replaced, which pooled applicants from around the state, so as to allow the selection of the ablest students in each racial group, regardless of which schools they graduated from. The old policy flirted with quotas by explicitly incorporating lower standards for minority students. But it avoided the major vice of the new policy, which admits black students with poor records who finish at the top of weak schools while excluding stronger black students who did less well at more competitive schools. It will similarly distort the pool of white applicants, so that overall student quality will decline.

In light of this problem, racial quotas may offer the best solution to the affirmative action question. A quota has the advantage of picking the best students from each group. By making transparent the extent of any racial preferences, it allows them to be debated publicly in a responsible manner. Those who are opposed to affirmative action could resist any quota; other institutions might impose certain limitations below which admission is not offered, and the process could be continuously reviewed in light of the progress made by the enrolled students. Courts should think long and hard before invalidating affirmative action programs, unless they are prepared to take the next step and invalidate the Texas 10 percent program as a pretext or sham (which under traditional civil rights law it surely is). Otherwise, legislatures will seek to avoid the stain of racial preferences by adopting dubious selection procedures based on the premise that it's OK to make able white students worse off so long as able black students are made worse off as well.

In dealing with the elusive ideal of "merit," Bowen and Bok are surely correct to note that SAT scores and grades do not tell a student's entire story. A student's grades might suffer, for example, because he treats theater as his academic major and physics as an optional extracurricular activity. But in and of itself, this valid point does nothing to help the supporters of affirmative action. By any account, scores and grades remain an essential part of the mix. So long as that is the case, we should expect that college-bound black students as a group will start off with a substantial disadvantage. Their combined verbal and math SAT scores still hover around 200 points below those of white candidates, both for the students who take the test and the smaller fraction who apply to college.

Within the white pool, the stronger the scores and grades, the more likely the student to have those extracurricular activities that prove so attractive to college administrators. The same is true within the black pool. Even though the rank order of individual black and white students might change when the full file is considered, it is highly unlikely that blacks as a group will do better on intangibles than whites as a group; indeed, there is some reason to think, given their lower SAT medians, that they will do worse. An affirmative action program is not a system that uncovers hidden veins of merit that lie beneath the surface. When all is said and done, it is a program designed to boost black candidates relative to whites.

The boost may even be bigger than itappears. Bowen and Bok repeatedly express puzzlement that, with SAT scores held constant, black students seem to do worse than white students in college. Yet that result can be explained by the fact that the black and white applicants come from pools with different medians and (roughly speaking) the same variance. In each case, of course, the score that the applicant received could have been in error, either higher or lower than some hypothesized ideal. Given a bell-shaped distribution, both white and black applicants will tend to regress toward the means of their respective distributions, which in turn suggests that the corrected scores of black applicants (measured by repeated testing) would be lower than those of white applicants, given the difference in their respective means. The thoroughgoing defender of individual merit would therefore adjust downward the SAT scores of black students and upward those of white students to take this difference into account, now that the statistical theory has been confirmed by data. Here is one technical reform with no political future.

With this said, the inescapable question is how large a premium should be attached to each black applicant. Here Bowen and Bok would have done better to avoid the appeal to hidden merit and concentrate on the shape of the applicant pools. They are clearly right that no amount of massaging the traditional admissions standards could boost the percentage of black matriculants. Yet the explanation for that result is somewhat more benign than is sometimes supposed. All that matters under the traditional merit-based system is the rank order of the candidates in the applicant pool. If A is better than B, he gets selected for the spot, whether the gap between their records is large or small. It may be possible, therefore, to substantially increase minority admissions without substantially reducing standards.

Whatever the perceived quality of an institution, a large number of high school students will choose it as their "reach" alternative. A very large number of unadmitted white applicants could thus stand between the weakest cohort of the admitted white applicants and the strongest of the excluded black candidates. Those bulges hurt. It is quite possible, therefore, that we will leap over a very large number of white applicants to include more black applicants, yet the actual difference in performance ratings (and the predictions of success associated with them) would be relatively small.

The pain of the excluded white students still counts in any ultimate calculus. But to the extent that the above proposition is true, the defenders of affirmative action now have a pretty strong answer to the complaint that the admission of black students will lower the level of education by reducing the average quality of the students. On the other hand, the large number of affirmative action programs that accept large racial gaps in the scores of admitted students cannot take refuge in this argument. It is a pity that Bowen and Bok did not supply the evidence that would allow us to explore this issue in depth.

More concretely, it is not enough for Bowen and Bok to show that graduation rates at the most selective private schools are about the same for white and black students. The real question in many cases is whether the education that students receive is equal to what they would receive under a color-blind system. On this view, the increase in rates of graduation between the 1976 and 1989 freshman cohorts could be regarded as bad news about the decline in the overall rigor of undergraduate education in the United States rather than a statement about the improved performance of black students.

Likewise, it is not enough to show that the same percentage of black and white students major, for example, in the sciences (often with an eye to enrollment in medical school) or gain admittance to graduate and professional schools. Perhaps the practice of affirmative action that begins with college is simply replicated by the graduate and professional schools, where it is likely to have more dramatic consequences. Now the white students all come from the top of a nationwide pool that includes the best not only of the elite private institutions but also of the second-tier private schools and the large state schools that were not part of Bowen and Bok's study.

The same concerns apply to the authors' analysis of earnings data for graduates of the colleges in their study. Within the 1976 cohort from the most selective schools, for example, they find that the mean earnings of black women in 1995 ($71,800) were about the same as that for white women ($74,400). By contrast, the mean earnings of black men ($88,400) trailed those of white men ($110,500) considerably. Once again, these data cry out for an explanation. The usual story holds that black women are doubly disadvantaged, by sex and by race. Yet that theory is confounded when black women do as well as white women while black men fare considerably worse than white men. More important, these numbers do not provide an independent, color-blind confirmation of affirmative action's success. Affirmative action programs are as common in businesses as they are in colleges and graduate and professional schools. The earnings data that Bowen and Bok present therefore could be taken not as validation of the value added by affirmative action but as a sign that affirmative action is pervasive at all levels of American life, for good or for ill.

Bowen and Bok do not wish to scratch below the surface of this delicate question. In particular, they spend no time at all in looking at what happens to black and white college graduates in Ph.D. and professional school programs, including bar passage rates and medical board certifications. Data on these issues are sometimes hard to collect, but there is some evidence that the affirmative action problem persists beyond college. In 1997 bar passage rates for graduates of California law schools were 82 percent for white applicants and 57 percent for black ones. And those figures do not reflect the lower graduation rates of black law school students. Ph.D. trends are also disturbing. The number of black Ph.D.s dropped from 1,047 in 1982 to 951 in 1992, from 3.4 percent to 2.5 percent of the total. In 1992 no black candidates received a Ph.D. in many branches of mathematics and the physical sciences. One possible conclusion is that affirmative action raises the performance of the black students clustered at the middle of the pool, while lowering the performance of those at the top.

Unfortunately, Bowen and Bok explore none of these system-wide implications. Therefore they do not address the magnification effect introduced when affirmative action programs are resorted to not once but at each stage of career advancement. Within the white pool, the students become ever stronger as selection by traditional tests weeds out the lesser candidates. As one progresses within any professional hierarchy, the definition of merit becomes somewhat narrower and more performance-bound, but scores and grades are rightly regarded as strong predictors of professional excellence.