Party Polarization in Congress: A Social Networking Approach.

In this paper, I construct networks between Members of Congress using roll-call data and identify the community structure of these networks using the concept of modularity. Modularity maximization provides two key measures. First, the maximum modularity value provides a conceptually clear measure of Congressional polarization, derived from the data without making assumptions about legislator utilities, as NOMINATE based measures do. Second, the community detection algorithm used to maximize modularity yields a unique community structure for each Congress that may be quantitatively compared to alternate community structures using the modularity concept. Analysis of maximum modularities over the history of the United States Congress reveals high levels of partisan polarization to be the historical norm, with the period from the 75th to 95th Congresses being the notable exception. Comparison of the modularity measure to NOMINATE based measures of polarization reveal interesting substantive differences, especially during periods of declining party importance and partisan fractionalization. ..PAT.-Unpublished Manuscript [ABSTRACT FROM AUTHOR]