Research & Reports
Opponents of Affirmative Action Use the Same Misleading Statistics Rejected by the U.S. Supreme Court
October 2006
By Stephen W. Raudenbush
Lewis-Sebring Professor of Sociology and Chair, Committee on Education,
University of Chicago
Stephen W. Raudenbush was the expert witness on statistics for the University of Michigan
in Grutter vs Bollinger and Gratz vs Bollinger.
On November 7, Michigan voters will decide whether to keep or discard affirmative action at public universities and other public bodies. As the election approaches, opponents of affirmative action have distributed inaccurate information about the role of race in university admissions.
Writing in the Free Press about a study by the Center for Equal Opportunity, Roger Clegg said, “Admissions data recently obtained from the University of Michigan show that race and ethnicity play a huge role in determining who gets in.” He picked out limited pieces of data to assert that African American students have a sizeable advantage over white students in being admitted to U-M programs in the law school, medical school and undergraduate studies.
Such claims arouse strong emotions.
These claims, however, are false.
And the statistical methods they are based on are exactly the same flawed methods rejected by the U.S. Supreme Court in 2003.
The truth is that U-M’s use of affirmative action has a negligible effect on admissions decisions for white candidates. That was my testimony during the university’s successful defense of affirmative action, testimony affirmed in the U.S. Supreme Court.
Sandra Day O’Connor, who wrote the Court’s finding rejecting the same claims made by Clegg, put it this way: “Student body diversity is a compelling state interest in the context of university admissions” and “the Law School’s race-conscious admissions program does not unduly harm nonminority candidates.”
A puzzle emerges here: How can a policy that effectively creates a diverse student body not “unduly harm” white students? If affirmative action increases the odds of admission for minorities, don’t the white students lose out?
The key to this puzzle is a basic fact that lies at the heart of the controversy but one that Clegg and his allies never mention to their readers: comparatively few minority students apply to the university. Among those few who do apply, the subset who are highly qualified are, it is true, admitted with high probability. If they were not, U-M would not be racially or ethnically diverse. However, U-M’s use of affirmative action has very little effect on nonminority applicants.
Here are the simple facts. In 2000 only 262 African Americans applied to the U-M Law School and 95 were offered admission. In contrast, 3,042 non-minorities (mostly whites) applied, of whom 1,171 were admitted. Suppose now that every African American who had been admitted had instead been rejected. This policy of total exclusion of African Americans would have “opened up” only 95 places. If all of these slots had been given to nonminorities, the number of nonminorities admitted would have increased to1,266. As a result the fraction of non-minorities admitted would have gone from 38.5 percent to 41.6 percent, a shift of 3.1 percent. In fact, the effect would be even smaller because many of the 262 African American students would have been admitted even if test scores and grades were the sole criteria for admission.
The same is true at the U-M Medical School and in undergraduate admissions, as well as in analyses of more recent data. For U-M’s undergraduate programs in 2005, only 1,579 applications were received from African Americans, and 806 were admitted. Even if every African American student were excluded, the chances of admission for nonminority students would rise at most by 4.2 percent.
It is therefore indisputable that eliminating affirmative action would have at most a tiny positive effect on the chance of admission of nonminority students. But it would have extremely negative effects on the ethnic diversity of the learning environment experienced by all students. The Supreme Court ruled that re-segregating universities in this way would undermine educational quality for everyone regardless of race.
Imagine you are a white student walking into your first class, a class of 25 students. Under the current system, it is extremely unlikely that the class would be segregated (that is, having no African Americans, Hispanics, or Native Americans). In fact, using the 2000 data, I calculated the probability to be .02. But if affirmative action were abolished, that probability would jump to .36. Over one-third of such classes would be racially segregated in this way.
Now imagine you are an African American student walking into the same class. Under the current system there is some risk that you will be the sole African American. If affirmative action were abolished, this probability jumps from .18 to .69, meaning that, in two-thirds of your classes, you would find yourself a minority of one. The U.S. Supreme Court was convinced that diversity is needed to “encourage under-represented minority students to participate in the classroom and not feel isolated,” to contribute to “a robust exchange of ideas,” and to have an education that “better prepares students for an increasingly diverse workforce, for society, and for the legal profession” and to ensure that military, corporate, and professional leaders reflect the nation they serve.
Clegg and his allies misled their readers, claiming, falsely, that affirmative action has a big effect on the chances of admission of white applicants. They neglected to report that abolishing affirmative action would segregate many classrooms, undermining the quality of learning for everyone.