Hierarchical Ideal Point Estimation with Influential Peers

• In Event: Scaling and Ideal Point Estimation

Abstract

Strategic voting is one of the biggest barriers to accurate ideal point estimation. One such source of strategic voting lies in the desire of rank-and-file legislators to appease the agenda setter or other party leaders on floor votes as a means of currying favor in future settings, whether in the legislative chamber or on the campaign trail. This behavior, combined with the agenda-setter's control over the agenda makes it difficult to obtain reliable estimates of legislators' ideologies. This project attempts to address this concern using a hierarchical model that extends the analysis of Clinton, Jackman, and Rivers (2004) to account for legislators' beliefs about the voting decisions of their peers and their desire to `follow the leader.' In particular, I develop an extension in which legislators' condition their votes on the votes of influential peers rather than solely (or even primarily) on the content of the bill. I am able to use this model twofold: On one hand, it allows me to empirically identify influential members of the legislature who are able to effectively control large voting blocs, and on the other hand, it allows me to identify those legislators who are truly independent of the party apparatus and able to vote their conscious without concern for the party's support for reelection. I support this model with evidence from the United States Congress in which I estimate the influence of party leadership over time, especially via the control of campaign spending, as well as individual members' susceptibility to such influence.

Author

• Garrett Darl Lewis, Washington University in St. Louis