Experimentation as part of a social science research strategy

In this paper, the author argues in favor of increased federal funding of large-scale long-term social science research projects, citing some benefits gained from the Rand Health Insurance Study and the Panel Study of Income Dynamics. The former illustrates some advantages of a well done experiment: (1) one can be relatively confident about the findings; (2) one can study certain issues that cannot be studied by analyzing existing data; and (3) all parties debating an issue will have a better understanding of it. The author suggests that the social sciences should have the same tradition of trials as the biomedical sciences do.