Partisan structure in online social networks
As part of a continuing project which makes use of data from the social microblogging service Twitter, I presented a paper at the 2010 MPSA in which I derived inferences about elite partisanship and ideology from only the patterns of connections between Twitter users. That is, given only knowledge of which of Twitter’s millions of users were following a set of Congressmen and other political elites, I am able to accurately predict both the partisanship and ideology (as measured by NOMINATE) of those Congressional elites.
This is surprising, because it implies that Twitter users’ preferences, individually somewhat uninformative, actually contain interesting and reliable information when aggregated. The paper for the talk may be seen here [PDF], and the slides are embedded below.
There is massive potential for the use of Twitter (and other online sources) to aid in our understanding of mass political behavior, largely by virtue of the volume of voluntarily expressed sentiment and expression that can be found. I am currently pursuing this line of inquiry in collaboration with Aaron King and Frank Orlando.