Policy, valence, and identity: Voter decisionmaking in primaries and non -partisan elections

This dissertation develops a new theory of voter decisionmaking in primaries and nonpartisan elections by integrating a rational choice framework with concepts derived from behavioral research.;In comparing candidates, voters make a counterfactual assessment: how would each candidate, if elected, react to the office's challenges, including new issues that arise? My theoretical framework predicts that voters rely on three measures in their decisionmaking. In the first approach, voters focus on the candidates' announced policy positions. Voters assess how close the candidates' positions on issues are to their own positions, while also weighing the issues' relative priority for them.;Voters also must evaluate the candidates as people. The second way to evaluate candidates is by comparison to the ideal candidate. Does a candidate have the virtues of a leader and statesman? The last way is to ask whether the candidate is 'like me;' i.e., someone who, because of a similar background, responds to issues as the voter would. Voters infer whether a candidate has similar preferences to them if they are proximate in "identity space" -- the web of reference group affiliations.;I test this theory through a web-based survey experiment using a random selection of university students. Each student participated in mock elections to see how they alter their vote as the candidate combinations are systematically changed (varying the candidates' platforms, race and ethnicity, and reported competence). In addition, to expand external validity, I analyzed a national survey of registered voters conducted during the 2008 presidential primary elections.;The analysis of the experimental and survey data showed that voters used three measures to compare candidates, preferring the candidates most proximate to them in policy and identity, as well as highly regarded on valence. In the process, I established a new way to measure how voter identity affects vote choice. As a result, I could clarify that Hispanic primary voters did not prefer Clinton over Obama because of identity concerns. Additionally, I examined primary campaigns, showing how they shape the voters' assessments of candidates' policy and nonpolicy qualities. Finally, I analyzed political accountability, identifying when voters will reject an ineffective incumbent.