Research question/goal:

In this research project, we devise a forecasting endeavour for the 2025 German Federal Election, which provides three main contributions. First, it advances the dynamic scientific forecasting of elections in multiparty systems by developing and applying new methods that account for uncertainty over time and efficiently pool information from various sources and levels. Second, it collects primary and secondary polling data, which will inform local and national level forecasts as well as research on the consequences of election forecasting exposure. These data will be made available to the scientific community for further analyses. Third, it will provide observational and experimental evidence on the consequences of election forecasts on the attitudes and behaviours of citizens and elites.

External Directors: Lukas F. Stoetzer (University of Witten/Herdecke) and Simon Munzert (Hertie School Berlin)

Current stage:

Election Forecasts for the German Federal Election 2025
In the 2025 federal election year, we conducted online pre-election surveys with over 50,000 eligible voters as well as surveys among journalists and candidates. Based on these data, our model generated rolling forecasts of party vote shares, probabilities of parliamentary entry and majority, district winners, and—for the first time—vacant districts. These forecasts were published on zweitstimme.org and attracted more than 1,500,000 visitors. The results were published in a Special Issue of PS: Political Science & Politics and presented at conferences of EPSA, APSA, and the AK Wahlen, and in the MZES Social Science Data Lab. The year concluded with a workshop on Elections, Public Opinion & Elite Behaviour at the Hertie School, bringing together researchers from across Europe.