Social Networks: Birds of a Feather Do Flock Together

A study by Harvard and UCLA researchers on Facebook is finding that social networks tend to lead to bonding social capital (people associating with others like them). These are preliminary findings and the study is continuing through 2009.

They have found that race and gender have the largest influence on who one befriends in social networks online, and white students (especially men) have the least diverse social networks. The study also found that the size of the social network was largest for black students, followed in turn by mixed race students and white students.

While this finding is consistent with findings across lots of sociological settings that show that we tend to form friendships with others who are like us (what researchers call ‘homophily’), it is a blow to Internet utopianists who hoped that the Internet would somehow make it far easier for us to form friendships with those who are different than us. [As the dog using the Internet in the famous New Yorker cartoon articulated, “no one knows you’re a dog on the Internet.”]

The interesting research project is being conducted by Jason Kauffman and Nicholas Christakis at Harvard University and by Andreas Wimmer (a sociology professor at UCLA). [We wrote earlier about Nick Christakis’ research on how obesity spreads through social networks.]

Putting a positive spin on the fact that facebook tends to lead toward like befriending like, Kevin Lewis (a third year PhD working on the project) asserted that this finding may buttress the case that the friendships formed online are real, if they exhibit traits (like homophily) that we see in real-world friendships.  Harvard-UCLA researchers are also examining”triadic closure” with these data: the tendency found by socialists for people who have friends in common to themselves become friends over time.

This study is part of an emerging field of computational social science (analyzing the vast data trails that Americans leave with their e-mail, their online friendships, their call-logs, etc.). My colleague David Lazer recently convened a meeting at Harvard of scholars doing computational social science or interested in doing more.  (For a brief post, see here.)  Some of the projects are quite fascinating including one by a scholar who captured all of his child’s communication and utterances from infancy through toddlerhood through an always-on digital camera, and then transcribed all the conversation to observe patterns of speech development.

And the New York Times yesterday in their article, “On Facebook, Scholars Link Up With Data” (NYT, 12/18/07) mentioned not only the Kaufman et al. study but other interesting recent studies. “Scholars at Carnegie Mellon used the site to look at privacy issues. Researchers at the University of Colorado analyzed how Facebook instantly disseminated details about the Virginia Tech shootings in April….Social scientists at Indiana [Eliot Smith], Northwestern [Eszter Hargittai], Pennsylvania State [S. Shyam Sundar], Tufts, the University of Texas and other institutions are mining Facebook to test traditional theories in their fields about relationships, identity, self-esteem, popularity, collective action, race and political engagement.”

This is all a wonderful development as we hope it will help to sort out some of the ethereal claims on social networks from the actual practices observed.  And given that these networks are longitudinal, one can actually watch friendships being made and see what factors at time 1 predicted friendships at time 2 which is quite exciting from a social science perspective.