My dissertation is now available! (It's even had a flattering review.)
The main part of the dissertation is based around a survey of 430 or so people on whether they knew about self-archiving and whether they accessed journal articles online from sources other than the official publishers' website. Over 70% of respondents did, a higher proportion than in previous surveys, though it depended a great deal on subject area. People in the field of medicine had hardly heard of it at all and were very unwilling to trust self-archived material. Why should this be?
There's a possibility of a sample bias, but other studies have found a similar tendency in medicine (there is much less self-archived material available in medicine than in other fields, as well). I suspect two things: firstly, that many researchers in medicine spend a lot less time immersed in academia than people in other subjects, and the concept of open access hasn't spread beyond academia yet; and secondly, that there is a much greater need for papers in medicine to be seen as authoritative, as determining the trustworthiness of articles is both more important and more difficult. More difficult, because the trials and experiments described are often expensive, long-term, or subject to stringent ethical stipulations; more important, because of the consequences if, say, a doctor acts on incorrect advice.
Whether these attitudes will change as self-archiving increases in popularity (as I believe it will) remains to be seen.