Utilizing Political Campaign Contribution Networks to Identify Political Affiliations and Important Actors in US Elections

Abstract

We seek to identify important political committees and states in recent election cycles by applying centrality measures to networks formed from political campaign data. One network we focus on is the senatorial bipartite committee-state network, which denotes relationships between committees and senatorial candidates from a specific state. We find that a state’s weighted PageRank score correlates to the contentiousness of that state’s senatorial election, and that the important committees of a given senatorial election cycle tend to reflect contemporary trends in national political discourse. The political action committee associated with the National Rifle Association is found to be the single-most important committee by weighted PageRank. By analyzing the community that contains this committee as detected by the Louvain algorithm, we find that network community structure corresponds to real-life political relationships. Furthermore, we utilize another community detection algorithm to split this network into two groups, and find that a state’s community tends to correlate to the party that won that state’s senatorial election.