Party Voting In The American States: How National Factors And Institutional Variation Affect State Elections

Abstract

This dissertation seeks to answer two questions. First, how do national-level conditions associated with the president's party, the presidential election cycle, and presidential approval influence state legislative and gubernatorial elections? Second, how does variation in state government power mediate how state elections are influenced by those national factors? I approach the first question by testing existing theories that help to explain how party voting is affected by the presidential election cycle. Specifically, I examine the effect of presidential coattails, surge and decline theory, national referendum voting theory and vertical policy balancing theory. Surge and decline and referendum theories have been applied to state elections; however, vertical policy balancing has not. In order to assess the independent effects of each of these theories, I include variables for all theories in comprehensive statistical models. These models cover gubernatorial elections from 1948-2010 and state legislative elections from 1968-2010. Using these models, I deconstruct how each of these three theories contributes to the phenomenon of midterm loss in state elections. I find evidence for surge and decline, referendum voting, and electoral balancing in both gubernatorial and state legislative elections. Having established the effects of theories of midterm loss in state elections, I then turn to the question of whether variation in state government power mediates midterm loss. I find that formal institutional power and the size of state government do not systematically affect midterm loss. However, I show that there are important differences between states with and without the direct initiative. I show that coattail effects and referendum voting are lessened in states with the direct initiative and that presidential punishment is increased. These results, along with the findings associated with theories of midterm decline, add to our understanding of elections in the American states as well as the American electoral process in general.