The social structure of political behavior: Action, interaction and congressional cosponsorship

Abstract

The twin objectives of this dissertation, understanding political behavior as a social phenomenon and assessing the relative impacts of action and interaction on behavior, are realized through the empirical study of cosponsorship in the United States Congress. Cosponsorship, the formal support of a bill, is commonly said to be a rational action by a member of Congress designed to further electoral goals. However, it is also possible that cosponsorship is the arational result of social interaction. Processes based on the principles of action and interaction may occur within the Congress or with reference to entities outside the Congress. Combinations of principle and environment provide a simple theoretical framework from which a number of hypotheses are generated. To test these hypotheses, a random sample of 100 bills from the House of Representatives during the 105th Congress is generated. Information regarding leadership, reciprocity, congressional districts, campaign contributions, media coverage, election results, organizational memberships, member demography and bill cosponsorship is recorded for each combination of sampled bill, member of the House, and week the Congress was in session. Comparisons and relations between members are represented in matrix form. A combined network effects-discrete time approximation approach converts these matrices into individual-level predictions of a congressperson's likelihood of cosponsorship over time. Alternatively, QAP analysis regresses relations on relations to make cross-sectional predictions about any two members' cosponsorship overlap. Results illuminate the importance of interaction to political behavior.