The Shape of the River: Long-Term Consequences of Considering Race in College and University Admissions, by William G. Bowen and Derek Bok, Princeton: Princeton University Press, 472 pages, $24.95
The debate over affirmative action in college admissions has often been couched in terms of moral imperatives. On one side stand the advocates of a color-blind world, who take their cue from Justice John Marshall Harlan's powerful 1896 dissent in Plessy v. Ferguson, which protested judicial acquiescence to racial segregation in the South. Today the opponents of affirmative action hark back to Harlan's theme: Race is irrelevant; only individual merit should determine the distribution of awards and honors in society.
Pitted against the opponents of affirmative action are the equally vocal and passionate educators and social leaders who view with abiding skepticism any abstract moral commitment to color blindness. They believe that affirmative action is necessary in many walks of life. Some of them insist that race-conscious policies must be used to rectify the many forms of discrimination--some invidious, some unconscious--that have scarred the American past and continue to shape its future. Others, including William G. Bowen and Derek Bok, do not emphasize this backward-looking justification. Rather, they advocate diversity in admissions to ensure minorities a meaningful place in the university community, not only for their benefit but for the benefit of everyone else as well.
Much is left unsettled in a debate that takes place at so high a level. The opponents of affirmative action too often sound dogmatic and absolutist in their attacks on what has become a widespread social practice. For their part, the defenders of affirmative action come off as special pleaders who always manage to advance some convenient social justification for policies from which they reap enormous benefits. The constant struggle between this white applicant who has been passed over (or claims such) and that black applicant with inferior academic credentials adds a sharp, human edge to the controversy, which is only partially blunted by the reassurance that affirmative action policies (sometimes gussied up as "race-sensitive" admissions) make everything work out in the long run.
Into this maelstrom have waded two formidable defenders of affirmative action, both with impeccable academic credentials. Bowen, a former president of Princeton University, now heads the Andrew W. Mellon Foundation. Bok, a former president of Harvard University and dean of the Harvard Law School, now teaches in Harvard's John F. Kennedy School of Government. Both men presided over great American universities noted for their unwavering commitment to affirmative action. In The Shape of the River, they defend in principle the policies that they implemented in practice. They do so in a low-key, measured fashion, with all hyperbolic observations carefully excised. Like Jack Webb, they ask for "just the facts." They believe that once these have been assembled, their detailed account of affirmative action programs will speak for itself.
Bowen and Bok's odd title is consciously lifted from Mark Twain's Life on the Mississippi, where that great author notes that any riverboat pilot has to "know the shape of the river perfectly" for navigating both up and down the Mississippi, day or night. That image does not quite work, however. The riverboat pilot has to proceed from detailed firsthand knowledge precisely because maps and charts offer fallible guides. But Bowen and Bok do not rely primarily on experience, anecdote, hunch, or intuition to make their case, even if they intersperse their book with some well-chosen remarks of black and white devotees of affirmative action (without once offering any opposing voices). Recognizing that pleasing testimonials will not win over the determined doubters, Bowen and Bok appeal to data.
The data come from surveys of black and white students who enrolled in the freshman classes of 28 selective private and public universities in 1976 and in 1989. The schools covered by the study are mostly private, with only three public institutions among them, all of which rank in the third (lowest) tier of selectivity. Bowen and Bok measured selectivity based on the combined math and verbal SAT scores of the students who enrolled. The first tier, with average scores over 1300, consists of Bryn Mawr, Duke, Princeton, Rice, Stanford, Swarthmore, Williams, and Yale. The second tier, with average scores between 1151 and 1300, consists of Barnard, Columbia, Emory, Hamilton, Kenyon, Northwestern, Oberlin, Pennsylvania, Smith, Tufts, Vanderbilt, Washington University, Wellesley, and Wesleyan. The third tier, with average scores of 1150 or below, consists of Denison, Miami (Ohio), University of Michigan (Ann Arbor), University of North Carolina (Chapel Hill), Penn State, and Tulane.
Bowen and Bok chose two widely separated dates to represent different eras. They excluded more-recent years because there was not enough opportunity to follow these students after college graduation. They also omitted any assessment of affirmative action for Hispanic or Native American populations. In addition, the study leaves out some highly selective universities (Harvard and the University of Chicago among them) and, more important, most major state universities, including those in the California and Texas systems, whose affirmative action programs have been curbed by external pressures. (California's programs were brought to a halt by its Board of Regents, whose decision was ratified in a 1996 state referendum. The Texas system reached the same result through a 1996 ruling by the U.S. Court of Appeals for the 5th Circuit.) Nor does the book tackle the thorny problem of affirmative action in faculty hiring--a critical omission.
Within their carefully constrained universe, Bowen and Bok offer a moderately persuasive account of affirmative action's successes. They note that race-blind admissions policies at these selective schools, no matter how extensive their outreach programs, would sharply reduce the black presence on campus, from more than 7 percent of students to about 2 percent. They also make a strong case that any sensible admissions program takes into account more than grades and test scores in determining the relative "merit" of different applicants. They show that, within the pool of black applicants, admissions practices proceed along rational lines: Black students with stronger records generally do better than black students with weaker records. They demonstrate that the black students admitted to the first tier of selective schools have better records, higher graduation rates, better success in graduate and professional school, higher earnings profiles, more community involvement, and higher levels of personal satisfaction than black students who enroll in second-and third-tier institutions. The demoralizing effects of affirmative action are more hypothetical than real to the black students who benefit from their participation in these programs. It is no mystery why most black students strongly support diversity on campus and beyond.
Bowen and Bok believe this procession of happy findings renders affirmative action bulletproof. They display a certain impatience with critics of racial preferences, such as Abigail and Stephan Thernstrom, for harping on the conspicuous failures of affirmative action while ignoring its quieter, daily victories. My own views on this subject are more mixed, but it seems to me that Bowen and Bok's celebration is a bit premature. Here are some reasons.
The authors regrettably understate the connection between what happens at elite institutions and what happens at other schools. The ability of selective colleges and universities to gobble up all the credible black candidates influences the success or failure of affirmative action programs at second-tier institutions. Even if those students are "qualified" for admission, they are likely to be bunched at the bottom of the class, creating internal tensions that could more easily be obviated if black and white candidates within institutions competed on a more equal footing. It may be unrealistic to expect strong institutions to abandon affirmative action programs so their less distinguished rivals might succeed. But even if it is unwise to ask the highly selective colleges and universities to change their policies, it would be nice to see some acknowledgment that their gain produces some external pain.
Bowen and Bok's narrowness of vision leads to graver omissions. They argue that the issue of affirmative action arises only in selective universities with the luxury of rejecting some portion of their applicant pool. They tell us: "Many people are unaware of how few colleges and universities have enough applicants to be able to pick and choose among them. There is no single, unambiguous way of identifying the number of such schools, but we estimate that only about 20 to 30 percent of all four-year colleges and universities are in this category." But that reassurance rings hollow unless we are confident that all the large, flagship state universities are not on the list of selective schools. If some of them are, then 30 percent of the colleges and universities contain a lot more than 30 percent of the students. The affirmative action problem looks to be a lot more ubiquitous than Bowen and Bok make it appear.
To see why, it is important to count the number of students in each tier. I do not claim to have done anything like Bowen and Bok's exhaustive research, but I did whip out my son's copy of the Kaplan/Newsweek guide How to Get Into College for 1999, and here is what I discovered. For the 1997-98 academic year, the eight schools in Bowen and Bok's first tier had a total reported enrollment of 30,825 students, of whom 2,289, or about 7.4 percent, were black. The three state schools in the lowest tier had a total reported enrollment of 73,565 students, of whom 4,872, or 6.6 percent, were black. That number may be deceptively low because Penn State, with over 34,000 students, had only a 3 percent minority student enrollment.
If we add to the third tier just two schools, the University of California at Berkeley and the University of Texas at Austin, the numbers jump to about 7,500 black students out of 129,000, or 5.9 percent. But even those numbers are misleading, because by far the larger minority student population at both places is Hispanic, constituting 14 percent and 15 percent, respectively, of their student bodies (as opposed to 6 percent and 4 percent, respectively, for blacks). Bowen and Bok thus engage in a peculiar form of lumping that overstates the importance of the elite schools and underestimates the differences between blacks and whites that can be seen when the large state universities are taken into account.
Consider a table (page 62) from the Thernstroms' 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 it appears. 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.
Which leads to the other portion of the affirmative action debate that Bowen and Bok ignore in their study: What is the proper role for affirmative action in the hiring of faculty? Here the stakes of any single decision are surely higher than with the admission of individual students. Faculty members often engage in research--cutting-edge research in the leading institutions of higher learning. What chances are there that some applicant who has been boosted by one or two affirmative action programs can compete with other faculty members who have survived a rigorous winnowing process at every stage of their professional development?
It is not only research that is at stake. It is also teaching. Weaker students can receive remedial education or outside tutoring. They can take easier courses or a lighter schedule. They can rely on their stronger classmates for assistance. But faculty have no place to hide. They stand in the front lines the first day they walk into class. It is not surprising that many of the complaints about affirmative action are directed at weaker faculty rather than weaker students. Strong black faculty members have to fight student preconceptions that they are there because of their race and not their ability. The heavy concentration of black teachers in black or race studies only increases the perception of ghettoization.
The case for affirmative action at the faculty level has to be weaker than it is for admission to college. Yet even here, no generalizations come easily because a lot depends on the mission of the college or university and the substantive demands of the discipline. The point is even more true once we abandon the high moral ground to examine, as Bowen and Bok urge us to do, the long-term impact of affirmative action on the vitality of universities.
The Shape of the River does deliver on its promise to supply facts that justify affirmative action in college and university admissions. The evidence that Bowen and Bok offer should be sufficient to answer across-the-board objections to affirmative action based solely on the self-evident truth of the color-blind position. But they still leave key questions about the appropriate use of affirmative action unanswered.
Even if there is a single, distinct color-blind approach to college and university admissions, there is no similar unity of approach for affirmative action. It could be practiced on a small scale or on a grand scale. It could be used to make sure that a critical mass of black students is enrolled in a given college. Or it could be used to make sure that the percentage of blacks in any institution is identical to that of blacks within a local community, a state, or the nation as a whole. There is no single answer to the question of how much affirmative action is ideal for any given institution of higher education. The right way to think about this question therefore starts with a strong preference for decentralized decision making. Let each college and university decide, free from government coercion (and that requires the repeal of the present version of 1964 Civil Rights Act as it applies to private colleges and universities), how much affirmative action it wants.
Many people on both sides of the affirmative action debate will recoil at the thought that private institutions should be allowed to chart their own course on this issue. The defenders of color-blind norms are often conservative defenders of free markets, but they are prepared to sacrifice freedom of contract and association to prevent what they regard as unacceptable discrimination. Meanwhile, the defenders of affirmative action want to use the stick of civil rights laws to maintain affirmative action, fearing that without coercion private colleges and universities will backslide into the traditional racial practices of Jim Crow.
This nightmare scenario is hard to take seriously. The key reason affirmative action survives is that it commands strong support within colleges and universities. The explanation for this support is both simpler and more powerful than is often acknowledged: It is simply not possible today for any major college or university to present itself to its alumni, its faculty, or its prospective students as a lily-white (or white and Asian) institution. Perhaps diversity doesn't enhance the educational experience. Perhaps it doesn't promote tolerance and understanding across groups. Perhaps affirmative action creates internal resentments, some serious instructional difficulties, and invidious individual comparisons. Perhaps it also creates an unfortunate set of desserts and rewards. No matter. So long as the demand for it remains, the practice will remain. Given this reality, the wise course is to decentralize the debate so that each college and university can make its own decisions, thereby shifting the focus from the lofty question of "whether" to the grubby question of "how much."
These observations are incomplete because they do not address how affirmative action should be implemented, if at all, within state-run institutions. I confess that I am uneasy about the predictable conservative response that the Equal Protection Clause requires that states hew to a color-blind approach. That principle works well when it requires that the same penalties be given to all individuals who commit the same offense, or at least to rule out race as a ground for more serious punishment of one individual than another. I have heard little sentiment anywhere for affirmative action in the criminal code, even though many voice concerns about an implicit racial bias in law enforcement.
Yet here we are not dealing with the forms of protection supplied by the night-watchman state. We are dealing with the distribution of public benefits to various citizens. I freely confess that there is no first-best solution to this problem, because no one can explain why higher education should be part of the public sphere in the first place. But there it sits, and it dispenses huge subsidies to a fortunate few who gain admission to its programs. These public institutions are in direct competition with private ones. Why, then, should they be precluded from adopting practices that private educators think indispensable for their colleges and universities in the current social and intellectual environment? It hardly matters whether I agree with their judgment or not: No one should be allowed to elevate his own views on affirmative action into an immutable first principle. But owing to incessant political pressures, it is not possible to decentralize decisions by public institutions about whether, and if so how, to implement an affirmative action program. And so we lurch from extreme to extreme.
The idea of a "national conversation" about race rests on the facile assumption that dialogue will yield a single solution backed by a broad social consensus. But this industrial-policy approach to affirmative action is subject to the same fatal objection that is raised against it in banking or computer research: too much central planning. Our core national commitment should be to a rule of law that creates zones of private authority where individuals may form voluntary associations that reflect the aspirations of their members and not those of the population at large. Bowen and Bok would have done us all a great service if they had stressed the sound institutional framework for making decisions about affirmative action rather than pushing their own preferred policy outcome. "Leave us alone" offers a much stronger foundation for affirmative action than "we are surely right."
Richard Epstein (firstname.lastname@example.org) is the James Parker Hall Distinguished Service Professor of Law at the University of Chicago. His latest book is Principles for a Free Society: Reconciling Individual Liberty with the Common Good (Perseus Press).