The Multidimensionality of Political Polarization in the U.S. Congress: Topic Networks
Chen Liang
Computational social science research has been focusing on depicting the salient inter-party heterogeneity between Democrats and Republicans, but there is a lack of quantitative research in assessing the level of ideological heterogeneity within each party. We propose a new approach to study the inner-party ideological heterogeneity in Congress from a social network analysis perspective. We web-scraped the information and content of 316,921 bills from 1973 (93rd Congress) to the present (116th Congress). Based on this dataset, we constructed a network with each node representing a bill topic and the weight of each edge representing the number of legislators who support bills addressing connected topics. For example, if there are three legislators who co-sponsor certain bills about health and certain bills about national security, then there is an edge (weight = 3) between the two topic nodes. The level of inner-party heterogeneity in the Congress co-sponsorship network can, therefore, be identified by measuring how closely topic nodes cluster in this network throughout different congress periods; that is, by measuring modularity of each network graph.
We show that while Republican House Representatives tend to cooperate in several core policy areas, they have divisive interests in economic and social affairs. This observation corresponds to existing qualitative studies that indicate a historical tension between social and economic conservatives within the Republican party. On the contrary, with Democrats, there is not a clear distinction in either house between those who focus on economic affairs and those who focus on social affairs. Furthermore, from the 93rd Congress to the present, senators, in particular, have statistically significantly increased their cooperation beyond individual policy preferences and have been more frequently setting policy agenda based on partisanship. 
← Schedule