unusual irb requests
I’m reposting this from Scatterplot:
And another question on behalf of someone else. My IRB thinks it is not possible for them to approve to network research using a methodology in which subjects are handed a list of names and asked which people on the list they know. The reason for this, per IRB, is that people have to sign a consent form before their names can be put on any such list. Thus the researchers are being told that everyone has to sign two consent forms, first for the compilation of the list, and second for doing the survey. This IRB regularly says that organizations cannot turn over lists of their employees or members to researchers for the purpose of initiating a request to be in a research project. Is this a common objection? Does anyone have examples of research with a similar methodology getting approval from other IRBs? Would it make a difference if the list in question is public or semi-public, i.e. a paper neighborhood or school directory that is delivered to everyone in a neighborhood or school, or a web site that lists all of a group’s members? Please cross-post elsewhere if you know of another pool of people who might know the answer. (I’m thinking of orgtheory here, but there may be other groups.)
This seems like an abnormally aggressive position for an IRB. Any suggestions for OW? Has anyone else had a similar experience working with their IRB?
It seems like the real privacy issue is protecting the people on the list from knowing if ego picked him or her as a friend. It’s not as if ego doesn’t already know who works in his or her company. As long as you were able to protect the anonymity of subjects once the data were compiled in a data set, I’m not sure why this is a concern at all.
I don’t have experience with this sort of network research so I can’t speak to this particular issue. I can say that I have colleagues who have had the experience that an IRB at one institution approves a protocol easily and uncontroversially while another one says it’s absolutely unapprovable.
Work like this has been published so presumably this stuff has been IRB-OKed in the past. Presumably, the OP can mine their own bibliography for a list of folks who’ve done similar things and ask those people for details.
Benjamin Mako Hill
February 10, 2011 at 6:33 pm
If the research involves existing data that are publicly available, then the research should be “exempt” under 45 C.F.R. s. 101(b). If the research is deemed “exempt,” then no consent would be required and only minimal ethical criteria have to be met. But it’s up to the IRB to interpret the regs and figure out whether the research is exempt. Right?
Brookes
February 10, 2011 at 10:57 pm
Some specificity is required here. The original posting asked ego if the names are known. Brayden says “friend”. Will alters be contacted as a result of being identified by egos?
In the case that subjects sign a consent form and are asked whom they know, there is no privacy issue and no one is placed at (social) risk. IRB should be satisfied as long as they agree to the original recruitment protocol.
In the case that new subjects are recruited from being identified on the list, then this secondary protocol must be approved and consent forms signed. One must argue that this secondary recruitment does not put either the first group of respondents or the second at risk. So it matters what you ask: friend, acquaintance, informant, peer, …
Randy
February 11, 2011 at 2:57 am
I think the IRB is concerned that studying the relationship between a person A and person B requires the consent of both A&B.
Steve Borgatti says something to the effect of:
A’s knowledge of a relationship with B is A’s to talk about and they are allowed to by most ethical standards. http://www.steveborgatti.com/papers/ethics.pdf
James
February 11, 2011 at 3:37 am
Sorry, clarification.
Borgatti might say that in most situations this kind of reporting on perceptions of relationships is ethical.
It might be unethical to ask about such relationships if – for example – the relationship were criminal in nature or socially stigmatized.
See page 339 of the paper I linked to in the previous comment and look for the sentence beginning “To be fair, what respondents are normally reporting on is…”
James
February 11, 2011 at 3:48 am
Thanks folks. I’m especially curious even though this isn’t my research because I’m on my IRB now. As I understand it, what the IRB here is objecting to is compiling a list of everyone who is a member of organization X and then doing a survey of people (who consent to be surveyed) which of the people on the list they know. The IRB says it is a violation of people’s privacy to have their names on a list without their consent.
It is also my impression that this IRB would object to the standard network-finding protocol in which you ask people who are all members of some organization the names of their friends. And (although it has not come up yet) I think from things they’ve said that they would also find objections to organizational network studies in which the representative of organization X is asked about which other organizations X has connections to, and asks whether those connections are formal or informal.
What is odd about this is that people who say they are doing “snowball sampling” get a pass all the time even though the only way to do such sampling is to get the names of people who have not yet consented to have their names used. But if those names are to come to the researchers on a list from an organization, this IRB seems to believe that is impermissible.
That’s why I’m trying to get exemplars from other places, to push back. Or to learn what boundaries and distinctions are relevant elsewhere.
olderwoman
February 11, 2011 at 5:38 pm
Thanks for the Borgotti link; taken as a whole, I would say it tends to support the view that network research is risky for participants and is not likely to assuage IRB concerns.
olderwoman
February 11, 2011 at 6:57 pm
OW – I think there is plenty of research out there that uses similar kinds of methods (for example, see http://www.jstor.org/stable/2393451). I think comparing this sort of analysis to snowball sampling is a good move. As social scientists we’d hate to lose access to either one of these methods. I’d point out that if your university alone decides that network analysis is not ethical or violates privacy, then it would put it at a disadvantage in recruiting future network analysts to your school.
brayden king
February 11, 2011 at 8:47 pm
Brayden: the article has a 1992 pub date. There’s lots of old research methodologies that are no longer permitted under new IRB regs.
olderwoman
February 11, 2011 at 9:12 pm
I think this is really overreaching on the part of the IRB. They are imagining an expectation of privacy in which there is none, and conjuring up (I guess) a possibility of harm by the simple fact of being named from a public list or by someone who already knows you! Frankly IRBs should stick to research where real, non-negligible harm is a possibility and quit wasting our time and theirs.
As for the fact that different IRBs have different styles and that they don’t necessarily recognize one another’s decisions as precedent-setting, Laura Stark’s excellent dissertation explains the reasons for this.
andrewperrin
February 12, 2011 at 3:40 am
As usual, I agree with OW and AP. This is a scary case of overreach, and suggests a strange level of ignorance on the part of the IRB as to how organizations work. As AP says, there is no expectation of privacy in an organizational setting like this one. Let the organization’s managers worry about this; the IRB needn’t worry about it on their behalf.
ezrazuckerman
February 12, 2011 at 6:15 pm
OW – Here’s a link to a paper from 2010 that uses a very similar approach (http://orgsci.journal.informs.org/cgi/content/abstract/orsc.1090.0450v1). The analysts looked at the networks of all managers of a given rank in the organization. This is how they obtained the network measures:
“To collect the data for information, support, and friendship, we presented lists to our interview partners containing names of all managers of the respective company in a systematic order. To account for the fact that information and support ties are directed, we used two lists in each case by asking the following questions: (1) “Please mark all actors to whom you provide strategically relevant information” and (2) “Please mark all actors from whom
you receive strategically relevant information.”
I think this method for gathering network information is very common among network analysts. That’s why I’m surprised that the IRB of a major research university is coming down on this. To say that the researchers are violating the privacy of people on the list I think you’d have to make the case that people who receive the survey didn’t know or couldn’t easily find out who belonged to their organization and that, as Ezra mentioned, there was an expectation of privacy.
brayden king
February 12, 2011 at 7:32 pm
This is a scary case of overreach, and suggests a strange level of ignorance on the part of the IRB as to how organizations work.
Clearly what’s needed to deal with this ignorance amongst IRBs is for social scientists to do some basic organizational research and——oh, crap.
Kieran
February 12, 2011 at 9:45 pm
Two more thoughts.
1) I have a feeling that IRBs are reacting to the fact that names appear on a document and that the interview subject is known to the researcher. In network analysis you need this to establish relationships, but to IRBs it must look as though we are seeking to violate the anonymity of the research subject.
2) How do ethnographers and anthropologists deal with the fact that their research subjects are often groups with a number of people that they don’t know about at the outset (people who may not consent to being researched even if their culture is).
James
February 12, 2011 at 11:00 pm
It might be interesting to think of experiments that your IRB – or any IRB in its right mind – would reject. You could incrementally push to the reductio from a single questionable suggestion.
Michael E. Marotta
February 14, 2011 at 1:38 am