Accounting for Voter Heterogeneity within and across Districts with a Factor-Analytic Voter-Choice Model

José Afonso MazzonFaculdade de Administração e Economia, University of São Paulo, Ave. Prof. Luciano Gualberto, 908, CEP 0558-900 São Paulo, Brazil. e-mail: jamazzon{at}usp.br e-mail: kamakura{at}duke.edu (corresponding author) In this study, we propose a model of individual voter behaviorthat can be applied to aggregate data at the district (or precinct)levels while accounting for differences in political preferencesacross districts and across voters within each district. Ourmodel produces a mapping of the competing candidates and electoraldistricts on a latent "issues" space that describes how politicalpreferences in each district deviate from the average voterand how each candidate caters to average voter preferences withineach district. We formulate our model as a random-coefficientsnested logit model in which the voter first evaluates the candidatesto decide whether or not to cast his or her vote, and then choosesthe candidate who provides him or her with the highest value.Because we allow the random coefficient to vary not only acrossdistricts but also across unobservable voters within each district,the model avoids the Independence of Irrelevant AlternativesAssumption both across districts and within each district, therebyaccounting for the cannibalization of votes among similar candidateswithin and across voting districts. We illustrate our proposedmodel by calibrating it to the actual voting data from the firststage of a two-stage state governor election in the Brazilianstate of Santa Catarina, and then using the estimates to predictthe final outcome of the second stage.