A Network Theory Study of Roll Call Votes in the United States Congress

I study the U.S. Congressional voting record using network theory and computations. I encode the roll call votes from both the House of Representatives and the Senate for 1789-2006 into adjacency matrices (graphs) that encode the extent of agreement between any two legislators in a given House or Senate. I apply concepts such as modularity and various notions of centrality to the adjacency matrices and determine rank orderings of the Congressmen, giving insight into voting behavior and trends for certain individuals as well as changes over time for the entire Congress. I apply the same procedures to legislation cosponsorship networks and compare the resulting rank orderings. I also compare my results to the DW-NOMINATE rankings and look for correlations and patterns by computing party modularity and maximum modularity. I show that party realignments in U.S. history and that party polarization has become very prominent in the U.S. over the past few decades.