Vote choice, economic interests and uncertainty

This dissertation consists of three articles on vote choice, economic interests and uncertainty. In the first article I use the actual income voters report to have earned under the government of two different parties to measure their economic benefits from the two parties. Using this measure I find that in the 1997 and 2001 British General Elections voters do tend to vote for the party that leaves them better off. This relationship is stronger among voters with a high level of information. I also find that the relationship is the same whether voters have past data of their party differential or need to guess it prospectively, supporting the prospective economic hypothesis. Finally, I find that even though sociodemographic characteristics explain to some extend these party benefits the relationship remains robust for highly informed voters to the introduction of ociodemographic controls. The second paper develops a spatial voting model that includes uncertainty of voters about their ideal points. An implementation of the model using beta distributions to represent uncertainties is introduced. This is defended against current models which I argue mistake uncertainty for lack of importance of the electoral dimensions and lead to downward biased estimates of the impact of these dimensions on the electoral choice. Several tests against data from the 1996 U.S. Presidential Election support the claims of the paper. The last paper introduces a unified model of vote choice and abstention consistent with the classic rational choice model of the electoral decision, an algorithm to estimate its parameters and a statistical package that implements such an algorithm. This is necessary because as pointed by the literature non-unified models will produce biased estimates of the true parameters. Unlike previous unified models of electoral choice this model directly introduces party differential as a variable that affects the utility function of abstention, as implied by the theory behind the electoral choice. Simulated data shows that the estimating algorithm is not very efficient but it properly samples the true value of the parameters, unlike a non-unified model that, as expected, underreports the true value of the parameters of interest.