Experiments in economics and psychology have critically contributed to the development of new theoretical (behavioral) models of individual and social behavior. Experiments may not only be used to falsify existing models but can also suggest the right way of modeling. This will be exemplified by a novel study on the incompleteness of preferences in decisions under uncertainty where the major (behavioral) models fail to account for observed behavior. Moreover, in a strategic setting it will be shown that economics experiments that ignore the power of partner choice in social interaction are likely doomed to produce misleading predictions for field behavior and to give wrong guidance for theory development.