Settings, institutions, campaigns, and the vote: Comparing House and Senate elections.

Abstract

This thesis proposes a simple three-stage model of the electoral process, comprising the pre-election period, the campaign period, and the vote. A small set of empirical indicators are employed at each stage--for the pre-election period, political diversity, media market efficiency, and partisan balance; for the campaign period, campaign spending and candidate quality; and for voting, voter information and learning, evaluative standards, and vote choice--and causal relationships between the stages are proposed. The model is tested in the context of United States House and Senate elections in the 1980s. The results indicate that Senate and House elections are quite similar. Many of the observed institutional differences are in fact a product mainly of different levels of candidate spending and quality. There are many House races that look like Senate races, and vice versa. While institutional differences do not disappear, they are substantially attenuated once differences in settings and campaigns are taken into account. Weak support is given to the hypothesis that campaigns react to political settings, but strong support is given to the hypothesis that information holding and voting is similar across the institutions. There are many reasons why electoral theories and models should not be confined to a single institutional setting. The results of this thesis indicate that pooling Senate and House election analyses challenges conventional wisdom, and allows one to test the impact of institutional arrangements on elections.