In the case of my project, discussion of the topic matter has the potential to incriminate the interviewee, if they were to talk openly about their own illegal drug-related behaviours without concealing their identity or without encrypting the content. Ethics approval was granted for me to set up a process whereby communication was encrypted or the participant's identity was adequately concealed. Obviously both anonymity and encryption would be the best option, legally and ethically.
You correctly point out that encryption does not ensure anonymity in the strongest sense; I think this is important to keep in the forefront.
I have put together an Encryption Guide to assist participants in setting up secure IM to use both while completing my interview and within their normal IM use. I have consulted with some of the academics who created the open source IM encryption software Off-the-record, and have (hopefully) skilled myself up enough to assist interviewees in setting things up.
In terms of secure IM, using OTR seems like a good choice, in that it moves the encryption/decryption step away from some intermediary service provider and onto the computers of the participants/researchers. It probably would be equally reasonable to implement any secure IM strategy that relied on public key cryptography - whether PGP/GnuPG/OpenPGP or server- and client-side SSL certificates.
I will begin interviewing in June/July. I have tried to forsee problems, such as people not being prepared to spend time sorting out encryption, which may just not be a priority for them or may be beyond their computer literacy.
People are notoriously unwilling to make simple changes to their workflow that better protect them or their confidentiality (or even their incomes!) from 3rd-party observation or interference.
If anyone on the AoIR list has experience in this area and has the time to review my Encryption Guide, I'd be most grateful for comments and suggests on this draft. Also, you can use the guide to set up encryption for your own IM use, and you are welcome to test your set-up with me anytime (this will help prepare me for the interviewees!). You can download it from here: http://www.savefile.com/files/585220
Your guide looks fairly straightforward; remember, though, that you may have interviewees who have a serious aversion to our computer jargon. ;) With regard to email encryption - the "mixmaster" suite of anonymous remailer tools are pretty good, and useful in cases such as these. [These are the same flavor of tools (different generation, of course) that were once used to run the (fairly infamous) anon.penet.fi remailer service.] --elijah