(cross posted at KungFuQuip.com)

Gaines & Sendak: The Friend of My Friend is My Friend? Ideological Clustering on Facebook

Session 1 of the Politics: Web 2.0 Conference brings us to an examination of Facebook and clustering of ideological types and research done by Brian Gaines and Jeff Mondak at the University of Illinois. One of the fears about the growing rise of the Internet is it fosters a cognitive dissonance selective exposure (sorry, brain synapse misfire) and allows individuals to ignore information with which they disagree. It also allows them to congregate together with only those who align with their views and self-reinforce.

Gaines and Mondak looked at whether friend networks had commonality of belief systems or whether there were significant variations. One interesting aspect they explored was whether the urge to add more friends - ths raising their social capital - would lead people to connecting with more divergent networks and therefore exposing themselves to more diverse opinions.

One interesting slide in the presentation covered the ideology by Big Ten Schools. It's on page 26 of the paper linked above (I'll try to go back and add a grab later. What's interesting in that slide is the fact that the number of people who chose "very liberal" and "very conservative" remain largely unchanged across universities (as do the number of moderates, though with greater swing). Micah's take is that is probably consistent with the percentage of American's who are firmly ensconced in the wings of the two parties.

Only the number of "liberal" and "conservative" identifiers change dramatically with Northwestern proving to be the most liberal and Purdue the most conservative.

Interestingly, the study found that conservatives are slightly more likely than liberals to surround themselves with other conservatives. The change was fairly minor and when only "Top Friends" were explored, there was no indication that Top Friends were more alike politically.

One interesting notion they introduced is the idea of "the stealth conservative". Essentially the theory is (especially among college students) that conservatives would not want to out themselves and be somehow denigrated as conservatives by their more liberal friends. Since liberalism is more prevalent at college campuses, these conservatives may otherwise go uncounted. It's an interesting concept, but one that may become less and less meaningful as Facebook's population grows older.

I asked during Q&A whether they had considered looking only at those who had added a political application or otherwise indicated a level of political activism to see if they were more likely to select only friends with whom they share political beliefs. They indicated they had begun with a purely random sample, but have considered going back to do some subsampling based on political involvement (both online and to the extent possible off) and other factors.

The paper is an interesting read if you're interested in the study of selective exposure and the web.