White students who are assigned to classes with black peers in college are more likely to spend time with other blacks in the future, according to a new working paper that examined the experience of cadets at the Air Force Academy. The results suggest that universities that recruit diverse, talented classes can overcome their students' subtle biases and prejudices against members of other races.

The freshmen at the academy were assigned more or less at random to squadrons of 35, with whom they took classes and played sports. They were reassigned when they became sophomores, finding new roommates within their new squadrons. The authors of the study, Scott E. Carrell of the University of California, Davis, Texas A&M University's Mark Hoekstra and Baylor University's James E. West, found that white freshmen at the academy who were assigned to squadrons with more black students as freshmen were more likely to choose a black classmate as a roommate in their sophomore year.

Using the Air Force's composite measurements of aptitude among high school students, the three economists also found that blacks of greater academic ability had a larger effect on their white peers. Another indication that the white freshmen's racial biases were a factor was that the results were driven by white male students from the South. If a white male cadet from that region happened to be assigned to a squadron with a group of black students who were smarter than average, he was 35 percent more likely to choose a black roommate the next year than if his black peers had been below average in academic aptitude. (There were no differences in aptitude between white and black cadets as groups.)

Other research has shown that most white Americans across the country display a similar degree of racial bias - even if they don't know it - with slightly greater intensity in the South. This map is from Project Implicit, with the darker red sections indicating higher degrees of implicit bias:

The idea that whites' views of blacks change when they work with them as equals toward a common goal is decades old, dating at least to psychologists' observations of U.S. soldiers of both races who fought together in World War II. The new study is interesting because it suggests not only that peoples' attitudes about others can change, but also that people act on their changed attitudes and spend time with members of groups against which they might have once held a prejudice.

Other researchers have documented the benefits of diverse classrooms for students in academic terms as evidence for the value of affirmative action.

Yet Hoekstra, one of the authors of the paper, noted in an interview that the results are not necessarily in favor of affirmative action. Assigning whites to classes with black peers who are less academically prepared could have little effect on their negative perceptions, although at least among the cadets, larger numbers of blacks had a greater effect on their white peers, even if the blacks as a group had relatively less academic aptitude.

"In our data, the positive impact of increased contact appears to outweigh the negative impact of somewhat lower-ability peers," Hoekstra said.

The results, all the same, suggest another reason that university officials might feel a need to maintain diversity in admissions. The Supreme Court has recognized the importance of this goal, even while issuing several rulings in recent years limiting the ability of schools to use race as a factor in evaluating applicants.

Of course, the purpose of affirmative action isn't mainly to benefit white students, but to redress the disadvantages confronting black youth as a result of decades of segregation, according to its proponents.