COMBINING PREDICTION MARKETS AND SURVEYS: AN EXPERIMENTAL STUDY

The track record of prediction markets suggests that markets may help to better foresee future developments and trends. However, looking at the range of applications, it becomes clear that there are certain limits. Complex forecasts, such as conditional or qualitative judgments are better gathered with traditional forecast methods such as survey-systems. In this study we integrate a prediction market with a survey in two distinct ways. First, trading in prediction markets indicates that participants believe to have additional information. Consequntly, we randomly trigger a one- qustion survey after a trade. Second, participants might find it convenient to answer survey qustions right on the same platform. Hence, we integrated a survey feature in a prediction market for 2013 German election. In a large scale field experiment with over 3,500 participants, we test consistency and the response rate of the integrated survey compared to a standalone version. We find that the integrated survey delivers robust responses and a 3.16 % (533.33 % relative) higher response rate. Moreover, the response rate in the trigger-based survey is above 90 %. These results highlight the great potential of integrating surveys with prediction markets