Here's a question from a Pew Research Center, Pew Forum on Religion and Public Life survey, Oct. 15-19, 2003:
Society should not put any restrictions on sex between consenting adults in the privacy of their own home. Do you completely agree, somewhat agree, somewhat disagree, or completely disagree with this statement?
Completely agree - 61%
Somewhat agree - 19
Somewhat disagree - 5
Completely disagree - 8
Mixed opinion (vol.) - 1 Don't know/Refused - 5
So can we confidently say that 61% really do completely agree with the statement -- to the point of supporting the legalization of prostitution? I'm pretty sure the answer is no: I doubt that most respondents would have realized that prostitution is covered (though it clearly is).
OK, can we at least that 61% is the rough fraction of people who completely agree with the statement as to those matters about which the statement is most often made -- homosexual activity, nongenital sex, and the like? Not really: The number may well be higher, but some people who believe that might have said "somewhat agree" or even "somewhat disagree" precisely because they realized that prostitution is covered. What that number is, we have no idea.
I do think we can guess that probably at least about 60% of the public oppose bans on homosexual activity and nongenital sex; I'd think that when most respondents heard the "consenting adult" language, they'd probably have thought of those things. But we can't even be confident of that, it seems to me.
Surveytaking is necessarily imprecise, partly because people don't accurately report what they really believe (or how they'll really vote). But questions like this, it seems to me, just exacerbate the likely inaccuracy.