A New Approach to the Spatial Theory of Electoral Competition

In his unidimensional model of electoral competition, Downs argues that voters use party ideology as an informational short cut for forecasting the policies that a party will pursue if elected. Parties are perceived by voters as points on an ideological axis. In the Davis-Hinich multidimensional model, on the other hand, the axes are real issues, and the principal actors are politicians who are modeled as points in the multi-issue space. This paper reformulates spatial voting theory in terms of a model that connects what we call predictive dimensions with political issues that are salient during a given election campaign. This model is both a synthesis and an extension of the Downs and Davis-Hinich spatial models. We obtain a median voter result for one predictive dimension that is similar to the Downs result but with important differences. We also obtain results showing the electoral advantage of incumbency and the tendency for incremental change when there is a great deal of heterogeneity in voter perceptions about the candidates.