A Heteroscedastic Spatial Model of the Vote: A Model with Application to the United States

Abstract

How do candidate policy positions affect the citizen’s vote choice? From the Downsian tradition, a common response to this question is that voters identify where contending candidates are located on policy space and then select the candidate closest to them. A well-known finding in current models of political psychology, however, is that voters have biased perceptions of the ideological location of competing candidates in elections. In this chapter we offer a general approach to incorporate information effects into current spatial models of voting. The proposed heteroscedastic proximity model (HPM) of voting incorporates information effects in equilibrium models of voting to provide a solution to common attenuation biases observed in most equilibrium models of vote choice. We test the heteroscedastic proximity model of voting on three U.S. presidential elections in 1980, 1996, and 2008.