How Voters Use Issues

• In Event: Poster Session: Political Psychology

Abstract

While issue voting is a well studied concept in political behavior, we actually know quite little about how voters engage in issue voting. What issues are voters considering and how are these considerations combined to shape vote choice? Relying on research under the bounded rationality paradigm, I propose that the ways voters use issues are fundamentally different than the normative expectations researchers often have when studying issue voting. Voters do not completely ignore, nor completely process, all available policy information about candidates. Instead, voters employ heuristic decision strategies that allow them to engage with issue information in meaningful, but also efficient, ways. This relationship between issue information and vote choice has been difficult to uncover in past work due to observational equivalence – where a voter could reach the same decision regardless of if they were primarily using issue information to inform their decision or non-issue criteria like partisanship or a candidate’s appearance. I designed two experimental surveys to test how voters engage with issue information. The surveys were designed to identify how well different known heuristic decision rules correspond to actual voter choices when the decision rules take respondent preferences and candidate policy positions as inputs. This experimental setup also avoids the problem of observational equivalence that is found from observational studies.

The first study, conducted in 2017, had respondents participate in six different candidate choice tasks. Each choice task had a variety of candidates take binary issue positions on a random set of fifteen political issues; with the respondents’ goal being to select their preferred candidate in each task. The choice tasks varied in the number of candidates and issues present, allowing for analysis of how task complexity could impact respondent use of issue information. I also collected respondent preferences on each of the fifteen issues that could be present in the choice task. This allowed me to simulate which candidate in each choice task a respondent would have chosen if they were to utilize issue information according to a certain decision rule. For the analysis, the simulated rules were compared to what the respondents actually did in each decision task, allowing for classifiers of each decision rule to be formed. A second survey, conducted in 2019, expanded on this analysis. Instead of randomly generating choice tasks for respondents and then comparing their actual choices to choices made by different decision rules, the second survey used respondent information on the first part of the survey to create decision environments that would differentiate certain decision rules from all the others. This alternative method cut down on the high correlation between simulated rules that was experienced in the first study and allowed for error of an attempted rule usage by a respondent to be estimated.

Results across the two studies largely confirm my expectations. Voters appear to have a variety of heuristic strategies at their disposal that allow them to meaningfully engage with issue information, while considering only a subset of available information. This conclusion differs from alternative perspectives that describe voters as using non-issue information to vote as-if they were issue voters, as well as perspectives that state voters disregard issue information almost entirely in favor of decisions based on political identities. By allowing for voters to engage with issue information while still accounting for cognitive and motivational constraints, and creating an experimental setup that avoids the pitfalls of observational equivalence, we can see that political issue information is impactful on the voter calculus.

Author

• Gabriel Madson, Duke University