Ecological Inference, Hybrid Models, and Voter Transitions
Project by Dr. André Klima, Prof. Dr. Helmut Küchenhoff, Prof. Dr. Paul W. Thurner
We develop and implement statistical solutions for the assessment of voter behavior and voter transitions in German and other contexts (France, UK etc. ) based on aggregate administrative data and on the enhancement of aggregate data with individual data (so-called hybrid models).
Based on a local project in the City of Munich at the occasion of the German Federal Election 2013 and the Bavarian Parliament Election 2013, we meanwhile support, e.g. the following cities for the upcoming 2017 German Federal Elections:
- Berlin
- Munich
- Frankfurt / Main
- Köln
- Düsseldorf
- Freiburg
Our Publications:
Klima, André, Helmut Küchenhoff, Mirjam Selzer, and Paul W. Thurner, 2017: Exit Polls und Hybrid-Modelle. Ein neuer Ansatz zur Modellierung von Wählerwanderungen. Springer: Wiesbaden. http://www.springer.com/de/book/9783658156732
Klima, André, Thomas Schlesinger, Paul W. Thurner, and Helmut Küchenhoff, 2017: Combining Aggregate Data and Individual Exit Polls for the Estimation of Voter Transitions. Sociological Methods and Research (2016 Impact Factor: 3.604, 2016 Ranking: 3/143 in Sociology | 1/49 in Social Sciences, Mathematical Methods) http://journals.sagepub.com/doi/pdf/10.1177/0049124117701477
Klima, André, Paul W. Thurner, Christoph Molnar, Thomas Schlesinger, and Helmut Küchenhoff, 2016: Estimation of voter transitions based on ecological inference: An empirical assessment of different approaches. AStA Advances in Statistical Analysis, 2(100), 133-159.
Our Software
We have developed the easy-to- use package “eiwild” for the widely used statistical software “R”. In this package, the ecological inference model proposed by Rosen et al. in 2001 was extended to a hybrid model, allowing the inclusion of individual level data. See:
Thomas Schlesinger, 2014: eiwild: Ecological Inference with individual and aggregate data. R package version 0.6.7. http://CRAN.R-project.org/package=eiwild
