Measuring ‘closeness’ in 3-candidate elections: Methodology and an application to strategic voting

Abstract

Past research suggests that voter behavior is influenced by perceptions of electoral competitiveness. For example, when an election is perceived to be close, voters will be more likely to turnout and/or cast strategic votes for their second-most preferred candidate. Operationalizing electoral competitiveness in three-candidate elections presents previously unrecognized methodological challenges. This paper first shows that many past strategies for measuring ‘closeness’ in three-candidate contests have violated at least one of three basic properties that any such measure should satisfy. We then propose a new measurement grounded in probability ratios, and prove formally that ratio-indices satisfy these axiomatic criteria. Empirical analyses using this new index provide novel and nuanced findings on the extent and causes of strategic voting in the 2010 British general election. The paper's operational strategy should be generally applicable to research on voting in elections, legislatures, and organizations.