Wednesday, November 23, 2011

Criminologists: IRB Demands Threatened Confidentiality

Two professors of criminal justice--Mitch Librett of Bridgewater State College and Dina Perrone of California State University at Long Beach--describe IRB demands that endangered research participants. They suggest that "the oversight element and informed consent requirement of IRBs may in fact represent the greatest threat to the well-being of participants, which is contrary to the mission, purpose and objective of IRBs."

[Mitch Librett and Dina Perrone, "Apples and Oranges: Ethnography and the IRB," Qualitative Research 10 (2010): 729-747, DOI: 10.1177/1468794110380548.]

IRB Demanded that Boss Be Informed

At the heart of the article of the article are the authors' accounts of IRB interference in their own criminal justice research.

The first concerns Librett's dissertation research, which he began while still a serving police officer. Librett wanted to interview five serving officers about the effects of undercover work, including "work-related psychological disorders [especially post-traumatic stress disorder (PTSD)], or any other difficulties (i.e., alcohol or other substance abuse)." (Brackets in original.) Since the disclosure of some of these effects could get a police officer fired or even prosecuted, Librett wanted to offer his narrators "an absolutely ironclad guarantee that any revelations would never be traceable to an identifiable person." Thus, he planned to interview them outside of their work settings and to avoid publishing any information that could identify them or even their agencies.

Librett describes the resulting exchange with the IRB:

IRB: You've presented a very persuasive argument. However we are unable to approve your request. We absolutely must have letters of cooperation from the agencies involved, there's too much liability involved here.

M.L: But sir, if I ask for official participation from the agencies, then the participants' identities might be compromised. We are talking about very small agencies here, and it wouldn't take an intimate knowledge of quantum physics for a police chief to figure out which undercover officer I was interviewing – when there is only one in the department.

IRB: Well, we just can't approve it without official permission.

Librett eventually settled for interviewing two retired police officers at significant cost to his research. He preferred to sacrifice his findings rather than accede to an IRB demand that would have put his narrators at risk.

IRB Demanded that Researcher Disclose Names and Tapes

Perrone had less trouble getting initial approval for her study of "club drug" users, but she was frustrated by her IRB's demand that she get signed consent forms. As she and her narrators understood, such consent forms posed more threat than protection, since they would create a document stating, in effect, that the signer used illegal drugs.

[Yes, I know that 45 CFR 46.117 allows IRBs to "waive the requirement for the investigator to obtain a signed consent form for some or all subjects if it finds . . . that the only record linking the subject and the research would be the consent document and the principal risk would be potential harm resulting from a breach of confidentiality." The problem here is that Perrone's IRB didn't understand that, and the rule itself requires the investigator to offer subjects forms that will endanger them.]

Perrone got around this by giving each informant the "option to sign any name she/he chose." How this process boosted trust or protected subjects is anyone's guess.

Perrone's IRB more seriously threatened her subjects when it investigated her research as part of an audit of randomly selected projects.

As part of the audit, the IRB demanded "the keys linking the pseudonyms the participants had chosen to their biographical information" and "an actual audiotaped recording of an interview session held with informants." Both demands raised the chances that the narrators' illegal drug use would become known in ways that hurt them, and forced Perrone to violate the promises she had made to them. Still, she gave in.

Perrone was perplexed and stressed that she had assured her participants that no one would listen to the tapes but her. Therefore, the IRB's 'audit' violated the trust participants had in Perrone, and breached confidentiality. The auditors stressed that as members of the IRB, they have the legal authority to listen to the tape. Perrone conceded, however there was not much choice here. As in all cases, refusal to comply with requests from IRB results in cancellation of projects, failure to complete dissertations, and the resultant damage to careers. Perhaps more importantly, disinformation and grapevine anecdotal evidence may prevent some projects from ever being attempted in the first place.

Questioning the Model

Reflecting on such experiences, Librett and Perrone assert "a fundamental disconnect between what the typical Institutional (or Ethical) Review Board will perceive as essential to safeguard the rights of human subjects and critical ethnographers' interest in maintaining a high degree of trust and partnership with their research participants."

They note that ethnographic projects "are by definition open-ended and unpredictable as far as the sort of questions that will be asked, the activities that will take place in the field, and the direction of interview protocols" and that "Ethnographers are hardly psychic. They are unable to accurately predict possible events with any degree of certainty in the field. While psychologists or hard scientists are not psychic either, their research is often conducted in closed and controlled laboratory environments."

Because the course of research is so hard to predict, Librett and Perrone note that "the greatest potential for harm to participants in an ethnography is the release of the participants' true names and/or identifying information. Thus, the IRBs' focus on research plans is quite beside the point."

They do not say so explicitly, but a logical next step would be to replace prospective review of field research with "ethical proofreading" of manuscripts just prior to publication to make sure that they did not violate any promises of confidentiality. [See Carole Gaar Johnson, "Risks in the Publication of Fieldwork," in Joan E. Sieber, ed., The Ethics of Social Research: Fieldwork, Regulation, and Publication (New York: Springer-Verlag, 1982).]

Additional Points

Librett and Perrone also pose some minor points, which perhaps could use elaboration. They write that "ethnographies may not fall within the definition of research, as it is currently defined in federal regulations. Ethnographies are by nature decidedly unsystematic. There is no testing and whatever knowledge is generated is heuristic. While ethnography most certainly is science, in that it informs our understanding of significant phenomena, it is not systematic or generalizable; it lies beyond the realm of positivism." Given the history of the National Commission, which specifically mentioned anthropological research in its IRB report, it's bold to claim that ethnography does not fit the regulatory definition of research. But perhaps there is a case to be made.

The authors also claim that "if a legal authority – police or court – had asked for [tape-recordings of interviews] Perrone would have certainly been free to state that the records had been 'lost' (without consequence) . . ." A researcher who receives a subpoena for her recordings may indeed seek to quash the subpoena, but falsely claiming that the records had been lost, or destroying the records, could in fact have very unpleasant legal consequences.


Anonymous said...

Interesting. I do not see any evidence that the disconnect is "typical". You'd think researchers would know better than to generalize from their own personal experience and anecdote.

The requirements in the Librett case are clearly ridiculous and contrary to the requirement to minimize risk. The IRB seems to be more concerned about institutional risk.

However, Librett is wrong if he thought he could offer "an absolutely ironclad guarantee that any revelations would never be traceable to an identifiable person." Researchers should never make promises that might not hold up. You'd think a detective, whose job is to identify people who beleive they won't be identified, would no better. Better to describe for the subject the steps taken to greatly reduce the risk. Telling subjects that there is "no risk" is unethical.

In the Perrone case there should be no requirement for signed informed consent because this is covered, as noted, by 117(c)(1). I agree that requiring documentation be made available as an option is stupid but following the minimizing risk requirement elsewhere in the regs an IRB is more than justified in ignoring it. If there are conflicting requirements I think it is fair game for the reviewers have to make a sensible choice.

They allowed her to keep audio recordings and keep keys to the participants identities despite the legal risk? If they were doing their job they'd want to severely limit the collection of identifiable data and the time it could exist in an identifiable state. And if you collect this identifiable data that makes you ineligible for 117(c)(1). It states "the only record" linking is the consent form. If you want that documentation exemption you can't record other identifying data. That doesn't make any sense. I wonder if they required her to get a certificate of confidentiality? I'm skeptical about how effectiveness of CoCs but at least it's something.

Perrone seems to be almost as clueless as she makes her IRB out to be. The part about the option of lying to the court shows really bad judgement.

If a breach of confidentiality creates significant risk, you want to severely limit any data that could be identifying including documented consent. You don't want audio recordings or keys files or anything like that. You want to create de-identified data that can't be re-linked without great difficulty or where there is significant doubt about the validity of the recreated links. If the only links are in the researcher's memory, then there is plenty of room for "I'm not sure who said what". Do enough interviews, and believe me, that's true.

So, where does this leave us? Sure there are IRBs who do a bad job, but there are researchers who need adult supervision.

Zachary M. Schrag said...

Thanks for these thoughtful comments. I should have mentioned that the authors do not merely "generalize from their own personal experience and anecdote"; they also cite a number of other publications noting similar problems with IRBs.

I agree that these two researchers could have used some guidance on the limits of confidentiality. Unfortunately, their IRBs seem to have lacked the expertise necessary to provide such guidance, and instead made matters worse. Nor is this surprising. As Joan Sieber noted back in 2001, "There is now a literature of virtually hundreds of approaches to protecting privacy or assuring confidentiality. This literature is rarely sought out by IRBs, researchers, or teachers of research methods. Most are not even aware that it exists. . . ."

The coercive power of IRBs relieves them of the need to achieve and demonstrate true expertise while discrediting the idea of any kind of ethics review. Deregulating social science research might encourage individual researchers and departments to seek out true expertise. See "Michael Rowe on Situational Ethics."

Mitch Librett said...

My 'absolutely ironclad guarantee' represented my personal word, as a cop-to other cops-while I was a serving police officer. This section of the article referred to a grad school project that never was published- for that and other reasons (I promised the participants that I wouldn't ever publish what they gave me for one). Lessons learned! But in regarding the IRB, their demands were impossible to satisfy; there are many, many similar cases cited in this extensive literature.

Zachary M. Schrag said...

Thanks for this comment. As Boston College prepares to supply prosecutors with interview materials for which the narrators were promised confidentiality, I think all researchers must ask themselves if their "personal word" includes a promise to go to jail before complying with a subpoena.