Party Discipline in Elections and Latent Policy Ideals

Individual members of Congress can have different policy preferences from the party leadership or special interest groups (SIGs), and the latter two may pressure members to shift their positions. To study this tension, we develop and estimate a multi-stage election model that incorporates party discipline with election spending. First, we uncover the unobserved ``ideal'' policies of these different political agents. Second, we estimate disciplining constraints and the importance of ``policy gaps'' with candidates to parties and SIGs. We then study various dimensions of discipline: the conditions under which representatives become more responsive to SIGs, the effects of safe seats on disciplining ability, and the limits of disciplining in the U.S. We find that safe seats make discipline less effective and that voter preferences are the main predictors of incumbent positions