There have been all sorts of defenses of affirmative action over the years, but just over a decade ago William G. Bowen and Derek Bok’s The Shape of the River appeared to change the debate. The authors use the carefully honed tools of social science to seemingly show both that affirmative action was necessary both to get minority students into elite colleges and ensure that these students could do well in their careers.

If there was a sequel to The Shape of the River, it might look something like Thomas J. Espenshade and Alexandria Walton Radford’s No Longer Separate, Not Yet Equal. Espenshade, a Princeton sociologist, and Radford, a higher education researcher with MPR Associates, use the same format as The Shape of the River, in that most of their work consists of analyzing a database of information compiled by colleges and universities. In fact, the colleges and universities that supplied information to Espenshade and Radford are some, but not all, of the schools that gave information to Bowen and Bok. Moreover, the Mellon Foundation—whose president for much of the study was William G, Bowen--funded Espenshade and Radford.

The flaws of The Shape of the River and No Longer Separate, Not Yet Equal are similar, too. Both books use proprietary databases that are unavailable to other scholars, a violation of the fundamental scientific rule that research results can—and should—be checked by other experts for accuracy. And neither book satisfactorily answers why helping a tiny number of minority students get into good schools does anything at all for the vast majority of African-American and Latino students.

Espenshade and Radford use as the core of their book a survey called the National Study of College Experience, which includes data from eight colleges and universities. We do not know which these eight schools are, except they are among the 34 Bowen and Bok studied for The Shape of the River. We do know that the authors originally were going to include ten schools, but decided not to use two historically black colleges because they decided that those schools were not representative of the typical American university.

The authors say the identities of the schools—and the information in their dataset--cannot be divulged because the schools supplied the data on the condition that it not be made public because it contained confidential information about students. But good social science is based on research that can be fact-checked and replicated by other scholars. Surely there must be a way to supply the database to reputable researchers while stripping out any references that would identify particular individuals.

Finally, it should be noted that the survey Espenshade and Radford conducted was for 1997, and readers are supposed to assume that data collected 13 years ago describes today’s colleges and students. It is far from certain that this assumption is valid.

Most readers will have three questions about affirmative action in colleges today. Who gets admitted under affirmative action? How well do students who are admitted under affirmative action do well in school? And why is affirmative action still necessary?

Let’s throw in a fourth question. Justice O’Connor, in her decisive fifth vote in the 2003 Grutter decision that declared affirmative action in universities constitutional for at least another generation, declared the policy necessary so that students could become friends with students from different races. What evidence is there that this inter-racial friendship has been enhanced by affirmative action policies?

Now to the answers. First, regarding admission. There’s a rising suspicion that an increasing number of the black students admitted to elite universities are not African-American, but are either of African or West Indian. Some blacks talk of “descendants,” or African-Americans who are descended from ancestors who were slaves. Affirmative action is, presumably, designed to help these students, not alumni of Caribbean prep schools with rigorous standards.

Espenshade and Radford find some evidence of this trend at the private colleges included in the survey. They calculated the percentage of African-Americans who applied to universities in their database who were first-through-fourth-generation immigrants. They found that it is a black fourth-generation American is far more likely to apply to a public school than to a private one. At the public universities the authors surveyed, 80.2 percent of the black students admitted were fourth-generation Americans, compared to only 51.2 percent of the black students admitted to private colleges. And 14.7 percent of the blacks admitted to private colleges were first-generation immigrants, compared to just 1.9 percent of those admitted to public universities. The largest shares of these black immigrants come from Jamaica, Nigeria, and Trinidad and Tobago. (Kenya comes in eighth.)

Whether they are immigrants or descendants, the black students in the Espenshade/Radford survey are likely to be less prepared than their white or Hispanic counterparts. The average SAT score among black freshmen was 1141, compared to 1288 for whites and 1266 for Hispanics. Thirty-nine percent of the black freshmen had SAT scores of 1100 or less, compared to ten percent of the whites and 9.4 percent of the Hispanics.

How well do black students do in college? Blacks—and, to a lesser extent, Hispanics—tend to get lower grades. Just over 50 percent of the blacks surveyed had grades that were in the lowest 20 percent, compared to 32.8 percent of the Hispanics and 15.8 percent of the whites.

Before addressing whether we still need affirmative action, let us consider Justice O’Connor’s criterion that affirmative action was necessary so that students from different races could become friends. Here the authors say the results could either be seen as a “half-empty” or “half-full” glass. Two-thirds of the students said they socialized with other races—but one-third didn’t. Half of the students had roommates who were of a different race—but half didn’t. Half of the students said one of their five closest friends was from a different race or ethnic group—but half said that their friends were all of the same race.

Now, the big question: Given the continuing poor performance of blacks in elite colleges, why is it necessary for these colleges to continue to have massive affirmative action schemes? Wouldn’t a lot of these students be better off succeeding at a state university than struggling at Harvard?

Espenshade and Radford recognize that this is an honorable position, one that dates back to James Davis’s classic 1966 paper in the American Journal of Sociology in which he argued that students who would do badly at an elite institution might prosper at a less competitive institution. “It is better to be a big frog in a small pond than a small frog in a big pond.” Davis wrote.

Two recent studies support Davis’s position. In a 2004 Stanford Law Review article, Richard Sander looked at the 27,000 students admitted to law schools in 1991 and tracked their careers. He found that black students admitted under race-based preferences were less likely to graduate, less likely to pass the bar, and less likely to be hired by law firms than those students who didn’t receive any preferences. Sander concluded that affirmative action was less likely to produce black lawyers than a race-neutral admissions system.

Martin Morse Wooster is a senior fellow at the Capital Research Center in Washington, D.C., and a contributing editor of Philanthropy magazine.