Files
Abstract
Economic experiments are prized for their high internal validity due to the randomization tech- niques they employ. However, especially in recent years, they have been criticized for their lack of external validity—the ability of their results to apply to different populations, settings, and con- texts. Despite these critiques and concerns, an economist seeking to optimally design an experiment for external validity lacks quantitative guidance, even though there is a wealth of literature on op- timal design for maximizing internal validity. This paper bridges that gap by outlining and solving a series of novel optimal design problems, providing experimental economists with the quantitative tools necessary to maximize the generalizability of their results.