Thomas Däubler (University College Dublin)

will speak on

Measuring party positions and modeling voter decision-making

Time: 12:00PM
Date: Mon 22nd November 2021
Location: Online [map]

Abstract: The talk focuses on recent empirical research in political science, pointing to possible extensions of the statistical modelling approaches used. The first part discusses methods for locating political parties in a one or two-dimensional policy space, starting from a document-category data matrix of statement counts (that originates from human coding of texts). We use a negative binomial latent variable model, implemented in a Bayesian framework in Stan. This approach could be modified, for example, to more flexibly handle overdispersion or to address the problem of cross-national measurement equivalence. The second part reports from a quasi-experimental study of ballot position effects on individual candidates' electoral performance. The unusual electoral system used in state-level elections in Bavaria allows us to observe the same candidates on different list positions and with varying sets of competitors. We find substantial ballot position effects over a large range of baseline ranks, and we conclude that the overall evidence is most compatible with the idea that ballot position affects both the order of consideration and the judgment of candidates. This line of work could be extended, for instance, by introducing preference heterogeneity to the voter utility function or by modelling the electorate as a mixture of different types of decision makers.

Join the Zoom call: https://ucd-ie.zoom.us/j/68316324831

(This talk is part of the Statistics and Actuarial Science series.)

PDF notice

Return to all seminars