Policy Perspective

College Roommates: "Odd Couples" Can Expand Horizons
by Greg J. Duncan

Fall 2004, Volume 26, Number 2

This fall, college freshmen who have been assigned roommates by university housing officers will meet one another for the first time. Many of them are probably wondering how they will get along: What will their roommates be like? Will they share similar tastes in music, movies, and politics? Or will they wind up with someone so com-pletely unlike them that they feel like they are living out an “Odd Couple” rerun?

Our research suggests that pairing up roommates from different backgrounds with diverse opinions on such issues can mold attitudes and influence behavior. Even binge drinking and grades might be affected, but more on that later.

We have focused on how the attitudes and behaviors of students in their second through fourth college years are affected by their first-year roommates. Since everyone in our studies participated in a random roommate lottery, roommate characteristics were beyond their control.

Affirmative action naysayers argue that mixing whites with less qualified blacks breeds antipathy, while proponents argue that social interactions promote tolerance and understanding. We found that whites reported more frequent and comfortable interactions with minorities two to four years later if assigned black rather than white roommates, and they held much more favorable attitudes toward affirmative action policies as well.

Mixing across class lines also mattered. Middle-class students assigned lower-income roommates engaged in more volunteer activities two to four years later. But empathy cuts both ways: Students become less supportive of higher taxes for the wealthy when assigned roommates from high-income families.

Consistent with evidence from social psychology, our general findings confirm that close personal inter-actions with people from different groups can lead to greater understanding and empathy. But at the same time roommate effects were not all encompassing. We found no evidence that roommates affected life goals, for example.

That brings us to the binge drinkers. The good news is that, despite parents’ fears, an abstinent child is unlikely to emulate the poor behavior of a binge-drinking or potsmoking roommate. Though binge drinking is more common in college than high school, children who abstained in high school drank no more in college whether they were assigned a roommate who drank or a teetotaler. The situation was similar for marijuana use.

But we did find one very worrisome result that suggests colleges should consider taking steps to avoid bringing binge drinkers together as roommates. We found that male—but not female—roommates who were both high school binge drinkers led to much more frequent binge drinking and lower grades than when a binge drinker was paired with a nondrinker.

The college years can be an exciting opportunity for expanding horizons and learning, both inside and outside the classroom. Classmates can be an important part of that process, but only if students take advantage of opportunities to interact with peers very much unlike themselves—like Felix and Oscar, the original odd couple.

Greg J. Duncan is Edwina S. Tarry Professor of Education and Social Policy and an IPR faculty fellow. Project collaborators include Johanne Boisjoly, Université du Québec à Rimouski; Michael Kremer, Harvard University; and Dan Levy, Mathematica Policy Research.

For more information on this research, please see "Peer Effects" to download a copy of the paper.