State and National Factors in Gubernatorial and Senatorial Elections

Abstract

Theory: Voters link candidates for govemor and senator to the president through partisanship. However, voters also distinguish between the functional responsibilities that govemors and U.S. senators have regarding the health of state and national economies, respectively. Hypotheses: Voting in elections for both governor and senator should respond to presidential approval. Voting in senate elections should respond to evaluations of the national economy while voting in gubernatorial elections should respond to evaluations of the state economy. Method: We replicate Atkeson and Partin's (1995) analysis of American National Election Studies (ANES) date controlling for the pooled and clustered nature of the data. We then crossvalidate their study using media exit polls. Results: Our analysis confirms the above hypotheses, which contrasts with the findings of Atkeson and Partin (1995). These differences result from controlling for the pooled and clustered nature of the data and from the well-documented misreport problem in the ANES data.