Whither the Institutional Review Board?

Historian Zachary Schrag's Institutional Review Blog (and his recent book, Ethical Imperialism) has been a valuable resource in tracking how federal regulations are shaping research in the humanities and social and behavioral sciences.  Today he posted a summary of a recent workshop conducted by the U.S. National Academies to look at potential upcoming changes in the federal regulations governing scientific research with human subjects, which may include substantial revisions to the Common Rule and other federal guidance.  Schrag's thoughts include the following concern about the inclusiveness of the process:

One participant—I think it was Plott—noted that the committee and the panelists almost all come from major research universities, not smaller colleges and universities who may have trouble affording competent IRB staffs. I would add that most of the projects discussed were major, grant-funded quantitative studies. Who will speak for the undergraduate?

Schrag's point about undergraduates is well-taken.  In a paper for a semester-long course or for a student working on an honors thesis, that might be presented at a statewide conference, a requirement for IRB approval for common, non-sensitive research projects is overkill, and may be used as an excuse to avoid allowing students to work on controversial topics to protect the interests of the institution rather than simply protecting the subjects of the research from potential harm.

An additional concern of mine is that the needs of those of us working with already-collected or secondary data are also underappreciated.  Even if the American National Election Study is defunded as a result of the Coburn Amendment, the same issues arise with the General Social Survey, Current Population Survey, World Values Survey, and hundreds of other data sets, and federal guidance is at best unclear as to whether or not working with this data is “human subjects research” (and whether the researcher can decide him/herself if this is actually the case)—it clearly fits the definition of “research,” but given the lack of any interaction between the researcher and the human subjects in question it's difficult to see how the researcher would be able to harm the latter using anonymized, secondary data—and I have had conflicting guidance from IRBs on this point over the years.  Clarification of this point would be extremely welcome.