BCAM-UPV/EHU Seminars on Applied Statistics: Who will form the government? KOALA (coalition analysis): A new strategy for analysing opinion polls

Abstract:
Common election poll reporting is often misleading as sample uncertainty is addressed insufficiently or not covered at all. Furthermore, main interest usually lies beyond the simple party shares. In multi-party electoral systems like in Spain or in Germany, the main question is, who will be able to form the government in coalition with other parties. In the talk, I will present and discuss new strategies for communicating results of election polls. We propose shifting the focus towards the reporting of survey-based probabilities for specific events of interest. An approach for multi-party electoral systems, focusing on probabilities of coalition majorities will be presented. Then, the probabilities, e.g. for a majority of a coalition between PSOE and U. Podemos or PP and C´s and others, is of interest. A Monte Carlo approach based on a Bayesian Multinomial-Dirichlet model is used for estimation. We also introduce visualization techniques that facilitate a more adequate depiction of relevant quantities as well as respective uncertainties.

For German election polls, we present our results on our web page based on polls from 8 providers with automatic update after new poll releases:
Web page (in german) : https://koala.stat.uni-muenchen.de/ 
Twitter (in german): https://twitter.com/KOALA_LMU 
Paper (in English) : https://arxiv.org/pdf/1807.09665.pdf 

Click to see the poster