Public event

Leveraging data science to forecast the 2021 German federal election – more than just polling

In recent years, elections everywhere have become notoriously hard to predict, as traditional polling has become less reliable. Data science offers new tools that go beyond polling for forecasting various election outcomes, such as modelling coalition possibilities.

Join us as Mark Kayser, Dean of Faculty and Research and Simon Munzert, Assistant Professor of Data Science and Public Policy, present their forecast model and predictions ahead of Germany’s federal vote on 26 September.

The event will also feature Marcel Neunhoeffer from Ludwig Maximilian University of Munich and Klara Müller from University of Mannheim, part of the team behind, a model that aims to go beyond the snapshots of political sentiment provided by opinion surveys (e.g., Germany’s well-known “Sunday poll“). Christian Endt, Senior Data Journalist at Zeit Online, will add his take on this super election year in Germany, with it s one federal and numerous regional elections. And he will discuss whether modelling can help reduce uncertainty about the election outcome.


Christian Endt works as a data journalist for ZEIT ONLINE in Berlin and reports from a quantitative perspective mainly on topics covering the Corona pandemic, elections and climate and energy issues. Prior to this, he worked for the Süddeutsche Zeitung in Munich. Endt studied mathematics and physics in Augsburg.

Klara Müller is an associate member at the PhD Centre for Social and Behavioural Sciences at the Graduate School of Economic and Social Sciences (GESS), University of Mannheim. Her research interests lie in the area of comparative political behaviour and electoral forecasting.

Mark Kayser is Dean of Faculty and Research and Professor of Applied Methods and Comparative Politics at the Hertie School, Berlin. His core research interests address the economic underpinnings of democracy, including political accountability, the effects of electoral competition on policy-making, and economic influences on electoral preferences. He was previously an assistant professor of political science at the University of Rochester, a Postdoctoral Prize Research Fellow at Nuffield College, Oxford and a fellow at the Center for Advanced Studies in the Behavioral Sciences at Stanford University.

Marcel Neunhoeffer is research associate at the Chair of Statistics and Data Science for the Social Sciences and Humanities at the Ludwig-Maximilians-Universität Munich. His research focuses on the application of deep learning algorithms to social science questions with an emphasis on data protection.


Simon Munzert is Assistant Professor of Data Science and Public Policy at the Hertie School and part of the Hertie School Data Science Lab. His research interests include opinion formation in the digital age, public opinion, and the use of online data in social research. He is the principal investigator of an international cooperation project funded by the VolkswagenStiftung entitled "Paying Attention to Attention: Media Exposure and Opinion Formation in an Age of Information Overload", and the recipient of a postdoctoral scholarship awarded by the Daimler and Benz Foundation. He received his Doctoral Degree in Political Science from the University of Konstanz.