The linkage between predictive dimensions and candidate issue positions in American presidential campaigns: An examination of group differences

Abstract

The purpose of this paper is to investigate the empirical utility of a new theoretical model of elections developed by Enelow and Hinich (1982, 1984a, 1984b) and Hinich and Pollard (1981). At the heart of this model is the assumption that candidate positions on campaign issues can be reduced to a set of positions on a smaller set of predictive dimensions. Factor analysis can be used to test the empirical adequacy of this assumption. Further, factor scores can be used as estimates of each voter's linear translation coefficients, measuring the linkage between candidate positions on the underlying dimensions of the campaign and candidate positions on real campaign issues. To obtain estimates of these translation coefficients, we factor analyze six candidate variables, using voter perceptions of each candidate's issue positions in the 1980 Pre-Election NES Survey. We then assign scores to each voter-issue pair on each of two underlying factors and examine group differences in these scores. We find systematic differences in mean factor scores among racial, ideological, partisan, educational, and income groups. These differences are consistent with the results obtained from a direct examination of perceived issue differences between Jimmy Carter and Ronald Reagan.