Presidential Evaluations and Voting for Congress

Abstract

elections constitute a referendum on administration's performance. Some studies in psychology and in political science have suggested that negative cues are likely to be more potent than positive ones as influences on individual opinions and behavior. This study focuses on question of whether negative evaluations of president are more powerful than positive evaluations in affecting turnout and vote choice in congressional elections. The study does find a with respect to vote choice, but it finds no affecting turnout. It concludes by presenting evidence on role of party competence evaluations as a mediating link between presidential evaluations and congressional voting. Although parents of young children have long suspected that there is an inherent negativity bias in human nature, social scientists have only recently begun exploring impact of on behavior. Lau (1985) broadly defines effect as the greater weight given to negative information relative to equally extreme and equally likely positive information in a variety of information-processing tasks. In other words, under many circumstances negative information is more powerful in affecting attitudes and behavior than positive information that seems comparably credible and extreme. Early research in this area was conducted by psychologists, and only recently have political scientists begun assessing its utility in political realm. This paper will examine impact of in explaining a well-known political phenomenon, loss of support by president's party in midterm congressional elections. In particular, it will focus on whether there is a differential impact of positive and negative presidential evaluations in voting for Congress. After showing that negative voting does not provide a fully adequate explanation, paper will lay out an alternative model based on party competence evaluations. This model works well at aggregate level, it can be applied to congressional voting in both midterm and presidential election years, and there is evidence that it can be applied at individual level as well.