Genauigkeit von Wahlerwartungsbefragungen zur Vorhersage des Ausgangs der Bundestagswahl 2013
Prior research found simple surveys that ask voters to predict the election winner to be among the most accurate methods for forecasting plurality elections (Graefe, 2013). However, little is known about the accuracy of such vote expectation surveys for forecasting proportional representation elections. The present study discusses theoretical and methodological barriers to the use of vote expectation surveys for forecasting proportional representation elections and provides empirical evidence for their accuracy in predicting the results of the 2013 German federal election. The accuracy of forecasts from two convenience samples of citizens (SoSci panelists and self-selected webusers who accessed the survey through a forecasting website) is compared to two expert groups (scholars and political journalists). The results suggest that the sampling of citizens is critical to forecast accuracy. While differences in accuracy were small for dichotomous forecasts (of, for example, who will win), panelists were less accurate than experts and webusers when predicting vote shares. However, citizens appear to form accurate expectations of election outcomes if they are provided with useful information.
|Titel (deutsch):||Genauigkeit von Wahlerwartungsbefragungen zur Vorhersage des Ausgangs der Bundestagswahl 2013|
|Titel (englisch):||Accuracy of citizens and experts in forecasting proportional representation elections: A study of the 2013 German federal election|
|Stand der Informationen:||02.12.2013|