Introduction

While the theoretical results presented in Chapter 11 are suggestive, they rest upon simplified voting models that omit several factors that influence realworld voters' decisions. These include influences that vary across voters, such as party identification and sociodemographic characteristics, as well as voter-specific random disturbance terms that render voters' choices probabilistic from the candidates' perspectives. In this regard, simulations based upon election survey data offer a possible bridge between our theoretical results and empirical applications. Here we explore policy-seeking candidates' strategies, using survey data from the two-candidate 1988 American presidential election and from the second round of the 1988 French presidential election, which was also a two-candidate contest.We also report applications to the 1997 British general election, which featured a three-party contest.

The 1988 U.S. and French presidential elections provide an appropriate testing ground for spatial models of policy-seeking candidates, since the presidents of both countries have broad constitutional powers to influence government policy (see Pierce 1995; Safran 1998). As in our theoretical analyses, we first explore these issues using the assumption that the candidates had full information. Then, in the next section, we extend our analysis to incomplete information scenarios, and we explore candidate strategies when uncertainty centers on the electoral effects of valence issues and, alternatively, when uncertainty revolves around positional issues.

The Contexts of the 1988 American and French Presidential Elections

Both the 1988 U.S. presidential election and the second round of the 1988 French presidential election pitted a candidate from a traditionally rightwing party against an opponent from a traditionally left-wing party.