Jeremy Hunsinger said
I can't agree with the definition of a subject as simply *someone* you intervene with or *someone* whose private information you have.
that is the federal definition though as cited on the umich irb and many others.
I don't know where you got just that. The original definition is
A human subject is* a living individual* about whom an investigator (whether professional or student) conducting research obtains (a) data through *intervention or interaction* with the *individual*, or (b) *identifiable* private information the devil is in the details. "A living individual" IMHO disqualifies imaginary personas and as long as you can't rule them out, you write off your entire work (even if we talk of just one in a hundred.). On the first note you certainly need a specific someone to intervene on or interact with , since you need to be able to distinguish between pre- and post- interaction / intervention states. On the second note you can't identify private information which is unconfirmed.
A subject should be a *unique someone* in one's research, a single person and not a single persona (unless of course your research design specifically addresses this issue of multiple personas). While the opinions expressed in a blog like Jeremy's can be attributed relatively safely to him, opinions in a forum or behaviors in an online world cannot be attributed to any single person. In my book, failing to convincingly address the issue of web anonymity is sloppy research. A signed consent form goes some way (but not all the way) in addressing this very real problem.
which problem? that your research needs identifiable individual subjects? in that case, you are very likely going to have the ethics issues noted earlier, which will then be resolved in some way like a consent form.
Name one field of research which does not "need" identifiable individual subjects. I'm not aware of any. George Floros, MD MSc Medical research methodology, Thessaloniki , Greece.