| This dissertation analyzes how preferences, parties, and constituencies jointly impact legislative policy making. It consists of three essays, each addressing this issue in different ways. In the first, I develop a new statistical model to formalize Barbara Sinclair's (2002) observation that legislators' decisions are a weighted average of multiple sources of influence. Applying this approach to the U.S. Senate since 1995 shows both its general usefulness and generates a number of important substantive results. For example, one key finding is that Republican moderates are much more sensitive to electoral and partisan pressures, reducing the weight they put on their own personal ideologies, than Republican extremists or Democrats of all ideological types. My second essay analyzes how external conditions affecting constituencies impact legislative behavior in a non-partisan environment. Specifically, I present a theory of how legislative district occupation led to observed preference change in the non-partisan Confederate Congress. I find that the crisis imparted by the occupation of legislators' districts led them to shift their behavior in favor of strengthening the Confederate government. My final essay changes gears and examines legislative behavior from the perspective of voters. Since voters are often unable to locate their legislators on an ideological scale, I present a statistical method that allows scholars to better understand the mechanisms behind voters' decisions whether to place legislators on a seven-point scale. The results suggest that informational, racial, and ethnic factors are influential in terms of saliency, but that education is a powerful predictor for decisiveness. |
