Spatial models of candidate competition and the 1988 French presidential election: Are presidential candidates vote-maximizers?

Existing applications of spatial modeling to historical elections suggest that parties rarely maximize votes. This complicates the analyst's task because explanations of party behavior must incorporate, in addition to vote-seeking motivations, such factors as the parties' calculations concerning postelection coalition negotiations. We argue that candidates in national presidential elections have greater incentives to single-mindedly seek votes than do political parties, and we refine this argument by developing two additional hypotheses: competitive candidates are more likely to be vote-maximizers than are noncompetitive candidates, and candidates are most likely to be vote-maximizing on issues highly salient to voters. Based on data From Pierce's (1996) French Presidential Election Survey, 1988, we fit a conditional logit model that incorporates voters' party identifications and sociodemographic characteristics and use its parameters to determine the vote-maximizing locations for the major candidates. Our empirical results suggest that a spatial model that incorporates voters' nonpolicy-related motivations accounts well for the candidates' behavior, especially for the more competitive candidates.