Measuring Legislators' Policy Positions

Abstract

Numerous scholars have used roll-call votes to analyze legislative behavior and the impact of party and constituency on legislative behavior. It is argued in this article that roll-call votes provide an insufficient data base on which to determine the overall impact of party and constituency because they do not indicate a legislator's degree of support for legis lation. A number of findings presented support this argument. First, although legislative analyses often require the assumption that a legislator is indifferent between two pieces of legislation voted on in a similar manner, such an assumption appears theoretically and empirically unsound. Second, even though the scaling of roll-call votes implies legislators' preferences among the bills making up the scale, the preferences inferred from this scaling are not equivalent to legislators' stated preferences. Finally, policy positions estimated on the basis of roll-call votes are not equivalent to those estimated on the basis of legislators' stated preferences. The article concludes by analyzing how conclusions about party and constituency influence are modified by utilizing legislators' stated preferences among bills in lieu of roll-call votes to estimate policy positions.