These are very valid points and thgere is no way to completely exclude the
"experimental effect", though I think it can be minimized by the way in which
you solicit the visitor's cooperation and also the point in his visit at which
you do the soliciting. With respect to those who will not participate (i.e.,
provide contact information), I think there are two things that can be done.
The first is to count the proportions of those who will participate and those
who will not. The proportion of the latter may (or may not) determine the
relative credibility of the information collected from the other group. I think
you also have to ask why visitors would not participate and this may also help
you determine the extent to which you must allow for the effects of
non-response.
I have not noticed any great problem to date with visitors refusing to provide
contact information, though I must say that we have not had that many occasions
when we've tried to collect it.
Harry Needham