Greetings, To build on what Ben has said. I too am an IRB member. I also do online research with a protected population (U.S.A.) so my work comes under additional IRB scrutiny. When I apply for approval my requests include a variety of wavers and requests. First, I do not anonymize for my public research...when I am working with their webpages only. I write that fact into my IRB requests. I take the position that the adolescents I study have already chosen to be known by their online monikers and that their published work is available to the entire internet....while mine is only available to a subset thereof. This logic has been consistently approved. Likewise I do not request approval from the participants to use their online words because they have already made them available to the entire internet. This logic would not prevail for protected sites...where more than a a cursory registration is necessary, if you can register and immediately receive registration approval then the registration is considered to be perfunctory so basically public. This logic is analogous to studying letters to the editor in newspapers since the writer chose to submit the letter for publication, and even though access to the newspaper is limited to subscribers....anyone who pays their subscription fee including public libraries, etc. can see their letters. The above changes when I communicate with the adolescent website producers directly. Then I ask for waivers so I don't have to get parental consent...this is only appropriate when the questions are not too personal or controversial, or when the teens would be in more danger from getting parental approval then if there were no approval. My work falls into the first category, I have not asked particularly personal information beyond basic demographic information. I also ask for waivers of documented assent from the adolescents themselves. That way I have nothing that ties their real names to the monikers they choices to use online...which may or may not be related to their real names. Even with this waiver I have been allowed to ask survey participants for contact information if they would like to be interviewed at a later date....their choice. In the states, local IRB's can approve waivers for most any of the requirements if the waiver request meets criteria. However not every college and university will entertain waiver requests....so you need to know if waivers are possible before you design you research project. Hope that helps Lois Ann Scheidt Doctoral Candidate - School of Library and Information Science, Indiana University, Bloomington IN USA Webpage: http://www.loisscheidt.com CV: http://www.loisscheidt.com/cv.html Blog: http://www.professional-lurker.com On Sat, Oct 27, 2012 at 7:35 AM, Bates, Benjamin J <bjbates@utk.edu> wrote:
I've served on a couple of University IRBs and recently had a chapter published on getting online research approved by IRBs - so here's my take (and from a strictly US perspective)
Getting permission from the site owners is helpful but not determinative. The two key issues that I'd say would be the emphasis of IRB concerns are: 1) Is the site and content public? 2) How are individual responses recorded?
For 1)- it's a little tricky since one could argue that anything on the internet is inherently public, but the federal IRB guidelines define "Public" in an online sense in terms of whether the people providing the responses have any expectation of privacy for their responses and the site. If there's some expectation of privacy, the IRB's likely to ask you to get consent from participants before collecting responses. Otherwise the responses are public and fair game.
For 2) - to qualify as "exempt" research the key is that the data/responses must be recorded anonymously- Basically that means that you don't record any personally identifying information (like usernames). If the raw data comes with identifiers there are two options- first, strip the identifying information from the raw data and create a research data set with no identifiers - then destroy the raw dataset (so you can't go back and peek); if you want to somehow link an individual's responses, you can also replace the personal identifier (username) with a generic identifier (Subject1) to create a new research dataset before destroying the raw data. It's also helpful (but not necessary) to have someone other than the researchers do the anonymising. The key to qualifying as exempt is that the data you actually use for the research is recorded anonymously - and that you don't keep raw data with personally identifying information long (or better, as Pis you don't even see the raw data (the anonymising is done by someone else)).
I hope this helps,
Ben Bates College of Comm & Info U of Tennessee, Knoxville