The literature on the wisdom of crowds theoretically points out the ability of large groups to make accurate forecasts. In the US and the UK, vote expectation studies have proven to be very successful, but to date there is no comparable study for Germany. Four weeks prior to the
2013 German federal election we asked 1,000 randomly selected eligible voters what they expected the vote shares (“Zweitstimmenanteil”) for each party to be. We find that the forecast based on expectations is relatively far away from the actual outcome and it is clearly inferior to projections based on vote intention polls, which is why we conclude that this approach does not work in Germany. Further, we find that political knowledge and education have a positive effect on the individual ability to predict the vote outcome, but the aggregate prediction of these individuals is no better than the overall forecast.
