Using Roll Call Estimates to Test Models of Politics

Measuring the preferences of political elites is critically important for analyzing the determinants and consequences of elite behavior. The decisions that elites make when casting roll call votes seem to provide an ideal opportunity for measuring elite preferences and testing theories of the political process. The fact that the resulting ideal points are a consequence of applying a statistical model to a model of individual choice, however, may affect their usefulness for measuring elite preferences and testing predictions regarding individual and collective decision making. When analyzing roll call votes, scholars should be mindful of how their decisions may affect the estimates from their analyses. I use simulations to illustrate how the nonrandom selection of roll calls may affect the ability to estimate ideal points that accurately reflect the preferences responsible for generating the observed votes, and I discuss work integrating the many models involved in the production and consumption of roll call estimates.