The American Association of University Professors (AAUP) recently released a report titled “Regulation of Research on Human Subjects: Academic Freedom and the Institutional Review Board”. The AAUP’s report is based on an analysis of responses from various professional societies (including LSA) to a July 2011 document released by the Department of Health and Human Services, the ANPRM (advance notice of proposed rule-making), which set out some proposals for IRB reform and asked for feedback.
Here, I discuss one of the report’s recommendations, which is relevant for linguists, but wasn’t the focus of the LSA’s official response to the ANPRM. I hope this can start a conversation following up on the earlier (August 2011) discussion on this blog.
The AAUP report recommends that all minimal-risk research on autonomous adults be included in the “exempt” category, not just the short list of procedures (such as survey forms, interviews, and observational research) that is currently included. This would include much of what currently qualifies for expedited review. Furthermore, the report proposes that the determination of “exempt” status no longer be subject to IRB review, but rather should be determined by the researchers themselves.
This is similar to one of the proposals in the ANPRM — a new category of “excused” research to replace the “exempt” category. This category would similarly expand the scope of the exempt category, but the researcher would still have to “register” their project with the IRB. The registration forms would then be subject to possible random audit later, but the researcher would be allowed to begin research right away without waiting for approval.
I think that both proposals would enhance academic freedom and relieve administrative burdens from researchers and IRBs. Of particular value for linguists is the expansion of the exempt category to include additional types of minimal-risk research. For example, my own experimental protocols, which consist in adult participants speaking sentences and/or pressing buttons in response to linguistic stimuli, are non-exempt under the current guidelines. Under either set of revised guidelines, most types of psycholinguistic experiments and other low-risk methods would become exempt. This would be a welcome change for me, since I would no longer have to wait for approval (which sometimes takes several weeks) every time I need to make a minor change in the protocol.
Would this revised system for exempt research provide adequate protection for participants?
In a recent email to the Ethics Committee, James Crippen brought up two potential problems regarding the proposed expansion of the exempt category: (1) how to define low-risk methods for each discipline; and (2) how to take into account subject matter in addition to the actual methods.
The AAUP solution to both problems is to allow researchers themselves to make those determinations, with the help of a standardized list of exempt methods and the advice of the IRB or a departmental representative as to how discipline-specific methods should fit into the standardized categories. But does this give too much discretion to researchers?
I think the idea of having researchers on exempt projects register their projects with the IRB is a good idea, and helps ensure appropriate protection of participants in minimal-risk studies. Registration of the project forces researchers to explain their procedures and how the rights of participants will be protected (just as in the current system). This would encourage researchers to make a careful, reasoned decision about whether a particular project qualifies as exempt and to seek advice from the IRB or other experts when needed. I believe that the threat of a random audit would discourage people from inappropriately stretching the definition of exempt status.
What do you all think? Do you agree with the proposal to expand the list of exempt categories of research to a larger number of minimal-risk procedures? Would simple registration of exempt projects be enough to ensure appropriate protection of participants? If the proposed changes go through, should the LSA provide their own guidelines for how to interpret the new exempt categories with respect to common research protocols in linguistics?
Please post your comments here!
Elaine Francis
Purdue University
Junior Co-Chair, LSA Ethics Committee
anggarrgoon
cahillm
drword563
ejfranci
jacrippen
Jack Martin
Jeff Good
jpboyle
posted by LSA Ethics Committee
natashawarner
speer2014
Sylvia Reed
tanyamatthews
LSA Ethics Discussion Blog 
It’s interesting that your work using recording of oral response or button-press response to linguistic stimuli is not currently exempt at your university. As long as the content of the stimuli is non-sensitive, the oral responses are non-sensitive, and the participants are adults and not members of a vulnerable group, it would be exempt at mine. Until a year or two ago, no audio recording could ever be exempt at my university, but they’ve now changed their interpretation of the exempt categories so that data that is potentially identifiable (because someone could recognize a voice) but would pose no risk even if it became public can still be exempt. (This was my interpretation of the exact wording of the exempt guideline in the first place. I think it hinges on a scope issue.)
My current protocol was approved about three years ago, and was deemed non-exempt because of the audio recordings. I will check whether Purdue IRB might have changed their interpretation– thanks for the suggestion!
While it at least ostensibly reviews the comments of all social science organizations it could dredge up on the regulations.gov site, the AAUP report puts a lot of effort into arguing against the proposal that Rena Lederman and I made in the AAA’s response last fall. There we take the position that the concept of ‘minimal risk’, while surely the right guiding idea at some level, is not actually workable in practice and so is not a good regulatory criterion. The effort to list methods or interventions that constitute minimal risk is a hopeless task — they are simply too many and too context dependent. (I reviewed a protocol a couple years ago where the researcher was interviewing farmers about their beliefs and practices surrounding cacao trees. Is this risky? Doesn’t seem so, till you know that the farmers’ lands were under threat of alienation by the government when they lay fallow for a certain period of time. Documenting that they were laying fallow could lead to terrible outcomes for the participants. Of course, cacao trees would be a ZERO risk topic to ask about in many other situations.) So instead, we suggest that what is needed is a more clearly defined object of regulation than ‘research’, which is way too broad and varies in meaning from discipline to discipline. We encourage the regulations be limited to (1) physical interventions and (2) interventions that involve deception or withholding information from participants (this would include any kind of experimental research where the hypotheses or purpose of research are withheld — sorry Elaine! — but that necessarily messes with your ability to get truly informed consent). That would leave a lot of research uncovered, not because it’s not risky but because more sensitive tools are needed to evaluate it. We proposed forming a commission of humanists and social scientists to develop evaluation guidelines for these other kinds of research. If this came to pass there would almost certainly be a big role for organizations like the LSA to get involved. So, may it come to pass!!
Lise, that makes good sense that certain kinds of humanities and social science research are just too context-sensitive to fit into pre-existing categories, and such research would be better served by a completely different kind of review process. The AAUP report really did not take this issue seriously. That also makes sense that most experimental research should continue to be regulated by IRBs under the existing federal guidelines (or revised version thereof). It’s interesting that you bring up the issue of deception. It’s true that my research and similar research in experimental syntax/semantics withholds information about the exact structures being tested until after the expt is over. But this has never been an issue for my IRB, as long as we state the general goals of the research in advance (e.g. “to investigate which factors lead people to favor one type of sentence structure over another structure, in cases where there are multiple ways of saying the same thing”). Acceptability judgments tasks have routinely qualfied as exempt, despite this kind of withholding of information. The sticking point here seems to be audio recordings and the potential for someone’s voice to be recognized (even though the recordings are simply of people speaking pre-determined sentences and contain no sensitive material).
Our IRB makes a distinction between deception, which involves a manipulation that makes the participant believe something that is not true, from withholding information. So, e.g. if I explicitly tell a participant that the purpose of an experiment on language production is to create tapes for use on the radio, but really there will be no such use, that’s deception. But if I don’t tell participants the nature of my experimental hypotheses before conducting research with them, that’s withholding information. As long as the information withheld is not likely to change the participant’s mind about participation, it is not a problem.
About registering research with the IRB, and having researchers themselves decide whether or not they are exempt: Since there is no review of the registered research, both of these proposals leave the consideration of the welfare of human subjects up to one person (or team) – the researcher(s). If you go back to the reason that IRBs were initially established, a primary purpose was so that someone not involved in the research could examine the content and methods and consider the effect on participants’ welfare – the idea was that researchers are sometimes not objective judges of the risk/benefit ratio for the use of human subjects. So researchers, who are very familiar with their research and often don’t consider it unusual or taxing when others would beg to differ, could use feedback and potentially re-design experiments to make them more acceptable for subjects. This can be useful – E.g. a couple of years ago a version of the famous Milgram studies was run w/ IRB approval after the debriefing process was substantially changed.
I agree that there are circumstances (like the one Lise mentions) where the most innocuous method can turn up results that would do harm were they not kept confidential. Those circumstances are rare, but that has two consequences: 1) most of the time if there wasn’t review, it wouldn’t result in any harm 2) people who assume that the possibility of harm is very small can end up blind to the harm when it is there. As a former social sciences IRB chair, it gives me pause to think that there would be no review at all (except audit after the research is already conducted) – studies that could cause harm don’t come up very often, but they do come up, and sometimes the researchers (who are perfectly good, thoughtful people operating in good faith) have not noticed.
Thanks Shari, I think your arguments in favor of maintaining oversight for determining low-risk status make good sense. After considering these and other comments, I have changed my mind about endorsing the AAUP’s proposal.
You are welcome – thank you for considering my arguments.