Valence Politics with Activists: Empirical and Theoretical Models of Multicandidate Competition

Washington University, Saint Louis MO

Investigators

Abstract

The spatial voting model is a sophisticated methodological device to study the interaction of political agents (whether candidates, or political leaders) and voters. However, the theory has predicted that all agents will be drawn to the electoral center in order to maximize votes. The conclusion, then, is that political competition is void of any relevance. Past research by the principal investigator and his colleagues strongly suggests that this conclusion is empirically invalid, both in polities based on proportional representation and in those based on plurality rule. Recent work by the researcher has shown that the mean voter theorem requires a domain constraint, on the support of the distribution of preferred policies of the electorate. Empirical analysis shows this constraint to be violated in polities as varied as Israel, Britain, the US, the Netherlands, etc. Instead, if electoral uncertainty is low, then parties should optimally position themselves at a distance from each other in the policy space determined by the differences in popularity (or valence) of the agents. "Popularity" is measured in the probit and logit models as constants in the voter utility functions. Under proportional representation, agents with low popularity are shown to leave the electoral center. Under plurality rule (as in Britain), agents with high popularity will contest the center. This research models the way activist coalitions make contributions of time and money to their preferred candidates. In order to maximize vote shares, agents will optimize between activist coalitions and the electoral pull of the center. The theory presented here develops the nature of response by agents to these electoral incentives. However, the rationality of activist contributions has not, as yet, been incorporated into the model. The investigation involves collaboration with Paul Whiteley of Essex University. Whiteley's work has focused on British leader popularity and activist support, and he has developed regression models of the relationships between these variables. The collaborative research draws on this research, together with completed probit models of voter and elite preferences, to construct an interactive model of activist behavior and political response.