Voter Preference and Emotion: Using Emotional Response to Classify Decided and Undecided Voters1

Abstract

This study tested the utility of using the emotioneliciting qualities of candidates to classify both decided and undecided voters. Based on Mehrabian and Russell's (1974b) theoretical perspectives, a hypothesis was developed that emotional response to candidates could be used to discriminate between voters.Three hundred eighty‐six registered voters in Bexar County, Texas, were randomly polled. These voters were asked a series of attitude and behavior questions. Of special interest for this study were questions about the Texas Governor's race between Bill Clements and Mark White. Voters were asked to identify their preference for either man and to respond to questions that tapped the emotioneliciting qualities of each candidate. The results indicated that the emotioneliciting qualities of the political candidates could be used to discriminate both decided and undecided voter preference. Specifically, over 90% of the decided voters and 80% of the undecided voters were accurately classified. This was seen as important for two reasons. First, it supported the claim that emotions may play an important role in voter preference. Second, it suggested a method for discriminating between voters even when voters were unable to immediately identify their own candidate preferences. Also of importance was the finding that the emotioneliciting quality scales became a “clarification” tool for some voters who were “on the fence” about the two candidates. After these voters responded to questions about thek feelings, candidate preference could be more clearly verbalized.A refinement of the “emotion‐eliciting quality” instrument was suggested for future research.