Abstract:
Typically, laboratory experiments suffer from homogeneous subject pools and selfselection biases. The usefulness of survey data is limited by measurement error and by the questionability of their behavioral relevance. Here we present a method integrating interactive experiments and representative surveys thereby overcoming crucial weaknesses of both approaches. One of the major advantages of our approach is that it allows for the integration of experiments, which require interaction among the participants, with a survey of non-interacting respondents in a smooth and inexpensive way. We illustrate the power of our approach with the analysis of trust and trustworthiness in Germany bycombining representative survey data with representative behavioral data from a social dilemma experiment. We identify which survey questions intended to elicit people's trust correlate well with behaviorally exhibited trust in the experiment. People above the age of 65, highly skilled workers and people living in bigger households exhibit less trusting behavior. Foreign citizens, Catholics and people favoring the Social Democratic Party or the Christian Democratic Party exhibit more trust. People above the age of 65 and those in good health behave more trustworthy or more altruistically,respectively. People below the age of 35, the unemployed and people who say they are in favor of none of the political parties behave less trustworthy or less altruistically, respectively.
