"Community structure in Congressional cosponsorship networks"

Another of my papers on communities in Congressional networks just came out in Physica A. This one arose out of a SURF project I sponsored in 2006, and the student in question (Yan Zhang) is the first author of the paper. The next two authors, A. J. Friend and Mandi Traud, are also undergrads. Then we have all the non-students: Me, political scientist James Fowler (best known for a recent well-publicized paper on the spread of obesity through a social network), and applied mathematician Peter Mucha.

The abstract for the paper reads as follows:

We study the United States Congress by constructing networks between Members of Congress based on the legislation that
they cosponsor. Using the concept of modularity, we identify the community structure of Congressmen, who are connected via
sponsorship/cosponsorship of the same legislation. This analysis yields an explicit and conceptually clear measure of political
polarization, demonstrating a sharp increase in partisan polarization which preceded and then culminated in the 104th Congress
(1995–1996), when Republicans took control of both chambers of Congress. Although polarization has since waned in the U.S.
Senate, it remains at historically high levels in the House of Representatives.


Ye Pei's SURF project from 2007 followed up on this work and looked at political realignments using voting networks. Essentially, we have found that a particular graph-theoretic concept called "modularity" (which looks at the connections inside a group of nodes in a subset of a graph versus connections between nodes in different subsets) can be used to give a nice measure of political partisanship. Realignments occur when the best splitting gives a much higher modularity than the splits one obtains by dividing the graph purely by party identification. We are going to start writing up a research paper on this stuff pretty soon, but in the meantime, you can take a look at Ye's SURF report.