Doctoral Dissertation Research in Political Science: Too Close for Comfort: Elite Cues and Strategic Voting in Multicandidate Elections

Duke University, Durham NC

Investigators

Abstract

The act of voting is one of the most sacred acts in a democracy. It is in the voting booth where citizens register their preferences over who should lead their nation. In some ways it seems a clear and simple act.: one should vote for the candidate they most prefer. However, in the presence of third parties, the evidence shows that many third party supporters abandon their fIrst preference to vote for their second preference. The mechanism behind this choice usually lies in the situation in which a voter's fIrst preference has no chance of winning so that voters do not want to cast a "wasted vote." Political scientists have formalized this logic in mathematical terms, taking into account the probability of a voter being decisive between pairs of candidates and their difference in utility between these pairs. While many studies have found that this model performs well in accounting for levels of strategic voting in different elections, one problem plagues these studies, namely most do not use a measure of the probability term that accounts for whether the voter will be decisive between pairs of candidates. While the logic behind strategic voting makes sense intuitively, the probability that one will cast a decisive vote will always be close to zero in a large electorate, such as the United States. If voters know that the probability of their vote being decisive is essentially zero, then why would they not just vote for their first preference? Riker and Ordeshook themselves acknowledged this problem and argued that people will systematically inflate the probability of being decisive, thus the p term will not be zero (1968). They argue that people might inflate the probability on their own accord or because of messages sent by elites indicating that p is much higher than it objectively is. Though this justification seems credible, few studies have examined whether people do inflate probabilities in the context of multicandidate elections (Blais, 2000) and whether elites play a role in influencing the variables in the voter's calculus (Hsieh et al, 1997). This research proposal address this question in the literature by examining elite cues in multicandidate elections. The primary purpose of this research is to show that one solution to the puzzle of why we even see strategic voting lies in the efforts by elites to influence the probability and utility terms in the individual's calculus. There are three key questions that the research seeks to answer. First of all, do elites try to send signals to convince voters to abandon their first preference and vote for their second preference? Second, are these signals effective? Finally, are there individual and contextual factors that account for who will accept such messages and who will not? The first hypothesis is that third party voters respond to persuasive cues by elites to abandon their first preference and vote for their second preference. Thus, the individual's perception of the variables in the individual calculus will be a function of the messages sent at the elite level. The results should show that individuals who receive messages encouraging them to vote strategically exhibit higher levels of such voting than individuals who do not receive these messages. Second, the effectiveness of such messages will be contingent on contextual factors, such as the source and type of message, as well as individual factors, such as one's level of political sophistication, external and internal political efficacy, and past voting behavior. An experimental design can clearly test these hypotheses by randomly assigning subjects to control and treated groups. This project makes several important contributions to political science. First of all, the project builds on the literature on strategic voting by filling in a missing part of the puzzle, the efforts by elites to influence the variables in the voter's calculus. The project also speaks to the literature on political persuasion. Most studies of persuasion look at the effect of messages on respondent's opinions on various issues. This study examines whether such messages can actually change the behavior of voters. Finally, this project contributes to the campaign and election literature, where the main fmding has been one of minimal effects for campaign messages. There might in fact be a stronger effect of campaign messages once we examine voters who have an incentive to switch their vote.