Toward a Model of Congressional Elections

Previous research on congressional elections has focused on either national or local events as explanations of vote choice. National forces consist of partisan identification, issue proximity, and retrospective evaluations. Research on local forces has stressed the candidates themselves, particularly with regard to the advantages of incumbency, challenger quality, and the comparative evaluation of candidates. Research in each of these areas has been extensive and has paved the way for a more comprehensive model of congressional elections, a model that combines these elements into a general framework. We have constructed and estimated a comprehensive covariance structure model that brings together these theories on national and local forces to explain voting behavior in the 1978 House elections. The model provides an excellent fit for the data and explains a large percentage of the variance in the structural equations. In addition, we are able to confirm several long-standing findings, as well as shed some light on more recent debates on the forces affecting vote choice in House elections. We view this effort as more than just an exercise in fitting a model to data or explaining variance in a system of equations. Rather, we see this as a first step toward developing a general model with which to evaluate voting behavior in congressional contests.