In this article, I suggest a different and more liberalized path. In Part I,I describe the regulatory metastasis of IRBs and some problems it is causing for social science research. In Part II, I offer some thoughts on the ways in which these problems might arise from the pro-regulatory incentives to which IRBs are exposed.
Finally, in Part III, I outline some modest liberalizing reforms to counter the effects of these pro-regulatory incentives. The reforms I propose broadly fall into three categories: IRB membership and structure, substantive IRB jurisdiction, and institutional liability. In the first category,IRB membership and structure, I propose that we should require basic First Amendment training for IRB members and include a First Amendment expert as a member of the IRB; that we should require that more than one, perhaps even a majority, of the members of the IRB have the expertise and competence to evaluate the risks and benefits of the particular research being reviewed; and that every research institution using IRBs should establish separate boards for biomedical and social science research. In the second category, substantive IRB jurisdiction, I propose that oral history and other interview-based research should be exempt from IRB approval; that IRBs should be permitted to prohibit or alter research in the social sciences only where the risks of the research substantially outweigh the anticipated benefits; that rather than have IRBs screen social science research before it is performed, they should review it (and enforce internal discipline on researchers, if necessary) only after ethical breaches cause some harm; and that social science researchers themselves, rather than IRBs, should determine at the threshold whether their research is exempt from prior IRB approval. In the third category, institutional liability, I propose that evidentiary rules in civil trials should exclude evidence of a university’s failure to adopt the Common Rule for non-federally funded research. Many details of these proposals will need to be worked out, but I offer them here as a starting point for reform efforts.