Is the glass half empty or half full?
Thomas J. Espenshade, a professor of sociology at Princeton University, used that question to answer a question about his new book, No Longer Separate, Not Yet Equal: Race and Class in Elite College Admission and Campus Life  (Princeton University Press), co-written with Alexandria Walton Radford, a research associate at MPR Associates. In fact, he could probably use the glass image to answer questions about numerous parts of the book.
While Espenshade and Radford -- in the book and in interviews -- avoid broad conclusions over whether affirmative action is working or should continue, their findings almost certainly will be used both by supporters and critics of affirmative action to advance their arguments. (In fact, a talk Espenshade gave at a meeting earlier this year about some of the findings is already being cited by affirmative action critics, although in ways that he says don't exactly reflect his thinking.)
Unlike much writing about affirmative action, this book is based not on philosophy, but actual data -- both on academic credentials and student experiences -- from 9,000 students who attended one of 10 highly selective colleges and universities. (They are not named, but include public and private institutions, research universities and liberal arts colleges.)
Among the findings:
- Significant advantages and disadvantages exist for members of some racial and ethnic groups with regard to the SAT or ACT scores they need to have the same odds of admission as members of other groups. While advantages and disadvantages were also found based on economic class, these were far less significant than those based on race and ethnicity.
- Just about every existing idea for reforming college admissions would not, by itself, preserve current levels of racial and ethnic diversity -- if current affirmative action policies were eliminated or scaled back.
- Most undergraduates at the institutions studied do have significant interactions with members of different races and ethnicities, and these interactions result in learning about the experiences of different groups. At the same time, the data suggest significant gaps in the kinds of meaningful cross-race interactions that take place with some groups much more likely than others to have such interactions. (By far, the most common interactions are white-Latino, while the least common are black-white).
- On measures of academic performance, graduation rates across racial and ethnic groups show only modest gaps at the institutions studied. But analysis of class rank suggests major gaps in academic performance. More than half of black students and nearly one-third of Latino students who graduated from the colleges studied, for example, finished in the bottom quintile of their classes.
Based on these findings, and the reality that some states have barred affirmative action and that the U.S. Supreme Court's blessing for consideration of race in admissions came with a 25-year time limit, the authors suggest that it's time for a massive federally supported effort, equivalent in intensity to the Manhattan Project, to determine the source of academic achievement gaps and to develop plans to shrink them.
The Test Score Advantage
Among the potential bombshells in the book are data on the advantages or disadvantages of SAT or ACT scores by race, ethnicity and economic class. Many studies -- including those released annually by the College Board and the ACT -- show gaps in the average tests scores by members of different racial or ethnic groups. This research takes that further, however, by controlling for numerous factors, including gender, status as an athlete or alumni child, high school grades and test scores, type of high school attended and so forth.
The "advantage" referred to, to take an example from the book, is what it would take to have equivalent odds of admission, after controlling for other factors. So the table's figure of a 3.8 black ACT "advantage" means that a black student with an ACT score of 27 would have the same chances of admission at the institutions in the study as a white student with a score of 30.8.
As the following table shows, there are large black advantages in the way colleges consider SAT and ACT scores, and notable disadvantages for Asian applicants. On issues of wealth, the SAT shows an expected affirmative action tilt, with the most disadvantaged students gaining and the wealthiest losing. But there is also a gain for upper middle class students. On the ACT, analysis found the advantages go to wealthier students.
The table uses ACT scores for public institutions and SAT scores for privates. The "norm" score was considered white for the race section, and middle class for the class section.
Advantages by Race and Class on the SAT and ACT at Selective Colleges, Fall 1997
|Group||Public Institutions (on ACT scale of 36)||Private Institutions (on SAT scale of 1,600)|
Much of the debate about affirmative action historically has focused on the advantages given to those from some minority groups. But the research in No Longer Separate, Not Yet Equal may also be of particular interest to advocates for Asian students. Many such advocates and guidance counselors who serve those students have charged in recent years that elite colleges have de facto higher standards for Asian applicants. Is the Asian disadvantage of 3.4 points on the ACT and 140 points on the SAT evidence to bolster that claim?
Espenshade said in an interview that he does not think his data establish this bias. He noted that while his formulas are notably more complete than typical test score comparisons by race and ethnicity, he doesn't have the "softer variables," such as teacher and high school counselor recommendations, essays and lists of extracurricular activities. It is possible, he said, that such factors explain some of the apparent SAT and ACT disadvantage facing Asian applicants.
At the same time, he said he understood that these numbers would certainly not reassure Asian applicants or those who believe they are suffering discrimination.
"I understand the worry of Asian students, but do I have a smoking gun? No," he said.
As to the large racial gaps on SAT scores, he said it was "distressing" in that it showed the difficulties colleges face in using their traditional criteria for admissions and still producing diverse student bodies.
The book notes that dropping the SAT or ACT as requirements would result in gains for black and Latino students. Espenshade has given papers previously showing that the biggest gains in such models are for colleges that drop consideration of testing entirely,  as opposed to just making it optional. (To date, only one institution -- Sarah Lawrence College -- has taken that step.)
Beyond shifting test policies, may other ideas have been proposed over the years to achieve a racially diverse student body without affirmative action as currently practiced. Here the book is quite discouraging. It reviews simulations based on class-based affirmative action (extra points for low-income applicants), reducing the emphasis given to academic credentials and priority admissions for those in the top 10 percent of their high school classes. And the book considers various combinations of these policies, looking for a formula that would yield diversity similar to what colleges have obtained to date.
"In this exhaustive examination of a wide variety of potential admissions policies, we have looked for but have not found any feasible policy alternative to the current practice of race-sensitive admission that has the capacity to generate the same minority student representation on campus," the book says. "The closest we have come among private institutions is a 15 percent minority student share among all students, achieved by lifting affirmative action, adding more weight for low-income students, and paying no attention whatsoever to students' academic qualifications. This policy stands no chance of being implemented at any academically selective institution."
Do Students Mix?
The new book doesn't just explore how students get into college, but what happens with them once there -- especially in terms of interactions with people from different backgrounds. The book notes that this is a question with important legal ramifications because colleges have justified affirmative action by pointing to the educational value of educating students in heterogeneous groups.
Here, the book finds evidence of significant interactions outside students' own racial and ethnic groups.
- 62.8 percent said that they often or very often socialize with someone of a different race.
- 51.2 percent of students reported having lived with at least one student of a different race.
- 50.9 percent of students reported having a "close friendship" with at least one student of a different race.
- 35.8 percent of students reported having dated at least one person of a different race.
The figures reflect all students, so the numbers are boosted in part by minority students on largely white campuses who may have relatively few fellow minority students with whom to interact.
Espenshade said that there is "no gold standard" for how much social interaction one would like to see among members of different groups, so it's hard to judge whether these numbers over all reflect positive or negative news. But he was heartened, he said, that survey questions showing that students who developed friendships across racial lines reported learning from those perspectives and gaining from the experience. Generally, he said, students reported the most gains in understanding coming from informal activities, such as socializing, and not from formal activities.
So if a college wants to encourage this sort of relationship, Espenshade said he would favor random freshman roommate selection, so more students end up living with people different from themselves, and policies that encourage groups that are based on race or ethnicity to co-sponsor events with other groups. But Espenshade said that the data suggest students are not moved by formal requirements. "I wouldn't advise diversity training," he said. "Students react negatively if they think they are being forced to take a diversity orientation session."
Across the various types of cross-racial interactions, not all groups interact evenly. Looking at who interacts, the data give the following order of likelihood: white-Hispanic, white-Asian, Hispanic-Asian, black-Hispanic, black-Asian, black-white.
The data in the book also suggest that ethnic studies courses are reaching a significant minority of all college students, but that the percentages of students at the colleges studied who majored or minored in them is extremely small, even with regard to their own groups. Nearly 40 percent of students at the colleges studied -- including nearly one third of white students -- took at least one ethnic studies course. But only 2.2 percent of students are majoring.
Ethnic Studies Coursework, by Race
|One or more of the above|
Measures of Academic Success
One of the most sensitive issues in discussions of affirmative action concerns academic success. Critics of affirmative action have long argued that the intended beneficiaries are in fact victims, because they might have more success in college -- and gain more confidence in themselves -- at less selective colleges. This "mismatch theory"  was recently repudiated in a landmark study of public flagships, the book Crossing the Finish Line.  That book found that minority students have the greatest level of success (measured by graduating) at the most competitive institution that admits them.
With regard to academic performance at the colleges studied in No Longer Separate, Not Yet Equal, the data on graduation rates largely back the conclusions of Crossing the Finish Line. The average six-year graduation rates for these institutions is 89 percent, with Asian students most likely to graduate (92 percent) and black students the least likely (78 percent). Similarly, those from the upper classes are more likely (90 percent) to graduate than those from working class families (79 percent). But here, even the numbers for black students and working class students far exceed national averages, and many institutions report much larger gaps by ethnic and racial groups.
It is among graduates that the new data raise questions about academic performance, because there are large differences in academic achievement (as judged by class rank) found both by race and economic class.
Class Rank by Race and Economic Class
|Group||Highest Quintile||Second Highest Quintile||Middle Quintile||Second Lowest Quintile||Lowest Quintile|
|--Lower and working||13.0%||10.9%||19.9%||20.1%||36.1%|
|--Upper and upper middle||25.7%||21.6%||20.8%||16.9%||15.0%|
Asked about the class rank data, Radford said that she doesn't think it's very significant, compared to the graduation rate data, which show that minority students are finishing their degrees.
"How much does a G.P.A. difference affect your life?" she asked. "It's not preventing these students from attending prestigious graduate schools or going on to have successful careers."
Espenshade said that he realized that there are data in the book that will be embraced by people on all sides of the debates over affirmative action. Describing himself as a "staunch moderate" on such issues, he said he will be pleased if advocates with differing views find evidence they like in the study.
"My main objective here is to be a mouthpiece for the data," he said. "My job is to let the data talk. What I may or may not feel about affirmative action doesn't matter. What matters is how the Supreme Court feels about it and how the voters feel about it."