Essays in Electoral Behavior and Bayesian Data Analysis
<p>The resurgence of Bayesian statistics in political research and, in particular, the rising popularity of Markov chain Monte Carlo (MCMC) methods, has unlocked estimation problems long thought to be considered impossible or intractable. Besides opening new terrain to political methodologists...
| Summary: | <p>The resurgence of Bayesian statistics in political research and, in particular, the rising popularity of Markov chain Monte Carlo (MCMC) methods, has unlocked estimation problems long thought to be considered impossible or intractable. Besides opening new terrain to political methodologists, these developments have allowed scholars to explore new problems or to revisit longstanding puzzles. This dissertation takes advantage of the generality and power of the techniques comprising MCMC methods to address novel substantive and methodological questions about abstention, voter choice and turnout misreporting, areas where substantive controversies remain despite the rich story of academic studies on electoral behavior and the considerable attention that has been paid to them.</p>
<p>The second chapter of the dissertation develops a statistical model to jointly analyze invalid voting and electoral absenteeism, two important sources of abstention in compulsory voting systems that had so far not been simultaneously examined. I illustrate the application of the model using data from Brazilian legislative elections between 1945 and 2006, underscoring relevant differences in the determinants of both forms of non-voting. The third chapter presents a study of voter choice in Chile’s 2005 presidential elections, examining substitution patterns in voters’ preferences over the competing candidates and highlighting the influence of candidates' entry and exit strategies on the election results, an aspect that has received virtually no attention in previous analyses of Chilean electoral politics. Finally, the fourth chapter develops a model to correct for misclassified binary responses using information from auxiliary data sources, and applies it to the analysis of voter turnout in the U.S. While the main contribution of the chapter is methodological, the empirical application has clear implications for researchers interested in the influence of race on voting behavior in America.</p>
|
|---|