Sunday, July 21, 2013

Comments on "Racial Segregation Patterns in Selective Universities"

Peter Arcidiacono of Duke University has been publishing a steady stream of papers examining the role of race in college admissions, with a particular focus on the effects of affirmative action.  I've discussed his work on this blog before, and given the substantial attention that generated, I'm sharing thoughts on another relatively new piece.

In the new paper, Peter and his colleagues suggest that friendships among students attending selective universities are no more likely to be interracial in composition than friendships in high school.  Of persistent racial segregation, they write:

"This is particularly true for blacks where on average their share of friends who are of another race is no higher in college than in high school despite their colleges having a much smaller share of black students than their high schools. However, the extent of interracial friendships, both before and during college, vary significantly depending on academic preparedness. The percentage of black friendships that are same-race is lower for those with SAT that are relatively low given the college they attend. Ordered probit estimates of the number of friends of different races show that, within a college, increasing one's own academic preparation makes inter-racial friendships with blacks less likely while increasing friendships with whites and Asians.

The multiple waves of friendship reporting ... tell a story of substantial racial isolation among blacks that slightly increases over time. Despite only comprising eight percent of the Duke student body, Black students report on average that 68% of their friends are black during their freshman year, a number which increases to 72% in their senior year. Ordered probit results again suggest that friendships with other races are more likely to occur the more similar one's academic preparation is to those of other races.

Taken as a whole, our results suggest that similarity in academic background is an important
determinant of interracial friendship formation. That black friendships are no more diverse in college
than in high school, despite blacks being substantially less-represented in their colleges, points
to a potential cost of a ffirmative action. Namely, by introducing a mismatch between academic
backgrounds of di fferent groups, interaction between these groups is discouraged."

I have several concerns with how the authors reach their conclusions and the policy implications they draw.

1. Their assumption seems to be that where educational settings have fewer black students, those students will be more likely to have interracial friendships.  Why expect this and not the opposite?Couldn't scarcity lead to a survivalist instinct to focus on key friendships among people who might seem to come from similar places?

2. The (very modestly) increasing isolation of the black students at Duke could be due to any number of factors, including campus racial climate. In this paper there are no tests for competing explanations other than academic mismatch. And the authors make nothing of the increasing segregation of white students from black friends as well- the % of white students with black friends as freshmen falls from freshman to senior year.  Moreover, the biggest increases in isolation are among Asian students-- as freshmen, 41% of their friends are Asian, but by senior year that is up to 48%.

3. There is no evidence that the observed relationship between academic "mismatch" and friendship composition is causal. It is quite a leap to suggest that racial segregation of friendships directly results from affirmative action.  In other words, there is little overlap between the empirical tests in this paper and the policy conclusions.  The authors conclude the paper with some words that suggest they know this-- and yet they make the policy statement in their opening abstract with direct and inflammatory language, stating that affirmative action plans "drive a wedge between the academic characteristics of different racial groups" creating problems for friendships.

4. The authors do not explain what their intended alternative policy might be.  Without affirmative action there would be fewer minority students on campus at all-- given their concern with friendship integration, what would the authors suggest happen then?