Extremism, Intensity, and Perception in Congressional Voting

Abstract

Social psychology, since the days of Thurstone, has emphasized the study of attitude distributions on quantitative dimensions. The discipline has, moreover, emphasized the study of individual differences. In particular, social psychologists – and sociologists and political scientists as well – have emphasized that the behavior of extremists differs sharply from that of moderates. Economists, in contrast, have been concerned with the development of mathematical models of choice along a dimension based on the concept, in one dimension, of a single-peaked indirect utility function. At the same time, empirical work by economists has neglected individual differences at the psychological level. In economic models, individuals typically differ only in their ideal points (which may differ because of budget constraints or socioeconomic characteristics) but not in the form of the utility function. In this paper, we use the findings on extremism developed in social psychology to inform a choice model commonly used in econometrices. This is the logit or double exponential model. In fact, as shown by Yellott (1977), the double exponential distribution is the only distribution that satisfies Thurstone case V discriminal processes, equivalent to Luce's choice axiom, for all possible choice experiments. Consequently, the research reported here accomplishes a form of Thurstone scaling that allows for the behavior of extremists to differ from that of moderates.