Behavioral experiments: how and what can we learn about human behavior

This paper addresses the experimental trade-off between the exercise of control over the actions of the experimental participants and the potential to provide understanding about human behavior. Control is a requirement of the experimental method to produce pertinent and intelligible results for scientific inquiry. But the more control is exercised the more the experimental results are the outcome of economists' actions. Economic experiments must therefore achieve a difficult balance. They must elicit intelligible behavior while ensuring that the actions of those taking part in the experiment are not determined by the design set-up and the rules of the experiment. The paper puts forward criteria for the analysis of the level of participation of experimental subjects in economics experiments and applies them to several experiments to illustrate the relevance of assessing the agency of experimental participants when deriving inferences about human behavior.