'Who Gets In?'

New book shows the impact of grades, test scores, race and gender on admission to competitive colleges.

May 15, 2017

Rebecca Zwick's views on admissions and diversity can't be easily categorized.

She has divided her career between academe (where she is professor emerita at the University of California, Santa Barbara) and the Educational Testing Service, one of the major players in standardized testing in the United States. But her new book, on which ETS holds the copyright, features data and analyses that may boost the arguments of those who dislike testing. Zwick is a major proponent of affirmative action, but some of the data in the book may well be useful to those trying to eliminate the consideration of race in admissions.

But that doesn't bother her. In an interview, she said she believes the data point to real challenges in education and should be used to bolster colleges' commitment to educating students from all groups. She also says colleges need to be more open about why they value certain characteristics in applicants.

She explains her views and shares data in Who Gets In? Strategies for Fair and Effective College Admissions (Harvard University Press).

Zwick's tool for analysis is the Education Longitudinal Study, which has data on 13,000 high school students who graduated in 2004. The study has their grades, test scores, and the colleges at which they enrolled. (We are of course more than a decade after the set of graduates Zwick studied, which she acknowledges, but arguably many of the patterns in postsecondary enrollment don't appear to have changed radically.)

Zwick focuses on the minority that applied to and enrolled in competitive colleges (those that admit fewer than half of applicants) and she also looks at models she creates for admission to the most competitive colleges as well. Those models may startle in some cases.

Consider gender patterns and enrollment in competitive colleges. Women make up a majority (53.4 percent) of actual students who enrolled in these colleges that admit less than half of applicants. That might suggest a healthy advantage for women, but Zwick's research finds that test scores (the SAT) are making men much more competitive than they would have been otherwise. If colleges that admit less than half of applicants made their decisions based only on high school grades, the proportion of female students would rise to 62 percent. And if this took place at the more competitive colleges (those admitting only 10 percent of applicants), the female share of students would rise to 64.8 percent.

Asked if these figures gave her pause about the use of tests in admissions, Zwick said she still believes that tests are relevant, given grade inflation and grade inconsistency in high schools. But she said these findings would make her worry about any colleges that rely overly or exclusively on test scores.

Zwick also said that she thinks grades are a proxy for other highly desirable qualities that may indicate why a grades-only approach would benefit women. Grade-point averages, she said, "reflect important qualities like effort and persistence and time management."

Of course much of the debate about grades and test scores ends up focused on race, ethnicity and affirmative action.

On this issue, Zwick discusses alternative models for colleges that admit no more than 10 percent of applicants. She starts by showing what the student body looks like now (with colleges of course weighing factors in different ways).

Then she presents three models for how these most competitive colleges would look if they admitted based simply on a composite of grades and test scores; if they did that but also used race-based affirmative action for underrepresented minorities; and then if they used the composite but favored those from low-income families. To measure the impact of these two forms of affirmative action, she adds 200 points on the SAT to the groups being targeted for help. The rules about data usage prevented Zwick from sharing separate findings on black, Latino and Native American students, but she was permitted to combine them for the following analysis.

What she found is that an admissions system based solely on grades and test scores would result in significant increases in Asian enrollments and declines in enrollments of underrepresented minority students. To a slightly lesser extent this is true for a system based on socioeconomic status.

Model for Impact of Different Admissions Models at Colleges That Admit Less Than 10% of Applicants

Race/Ethnicity Current If Decisions Based Only on Grades/Test Scores If Race-Based Affirmative Action Added If Socioeconomic Affirmative Action Added
Asian 12.1% 15.6% 12.5% 14.9%
Underrepresented 12.4% 2.5% 15.6% 5.6%
More than one race 3.8% 1.5% 1.4% 2.3%
White 71.7% 80.4% 70.5% 77.2%

Zwick said that these figures reinforce her views about the importance of affirmative action. But she notes that she doesn't focus on the reason cited by several Supreme Court decisions: that a diverse student body produces a better education for everyone. She agrees, but argues that colleges should be able to argue that our country is full of inequities that mean students from some groups are less likely than others to get a good elementary and secondary education.

"I think affirmative action should be viewed as a means to deal with a problem in society," she said.

But Zwick says her models encourage her to seek a major reform for college admissions. Her research shows that if a college values various qualities (grades over test scores, diversity in various forms), the odds of various types of students being admitted or rejected change. Much of the public frustration with elite college admissions, Zwick argues, is based on the feeling that colleges don't reveal completely what they are looking for. They say they want diversity, but do they say how much that matters, compared to other factors?

Zwick acknowledges that, for legal reasons, most elite colleges probably wouldn't want to say exactly how they value various factors (and of course, most of these colleges use holistic admissions and say that there is no formula). But Zwick said that "more transparency" would benefit everyone. And she hopes her various models encourage that.


Back to Top