A Dynamic Forecasting Model for the 2017 German Federal Election

We present efforts to develop a dynamic forecast of the German Federal Election 2017. In contrast to the predominant academic approach to forecast incumbent vote shares from measures of government popularity, economic conditions and other fundamental variables, we combine data from published trial heat polls with structural information. Opposite to common practice in the news media, we do not take isolated polls as election forecasts in their own right, but exploit historical data to assess the relationship between polls and election outcomes. Furthermore, the model takes care of the correlated evolution of party support over time and the multi-party nature of the setting. The value of our approach goes beyond the interest of the public and the media: Being able to dynamically track public opinion over the course of an election campaign can help understand how political events affect citizens' voting preferences.