So don't try to weed out for high risk people (assuming ytou have a good model.)
Define "high risk". If one boy is molested by an admitted homosexual, do we forbid admitted homosexuals from contact with children, wipe our hands, and consider it a job well-done? What about the 50 or 60, or even just 20 or 5, that are molested by perceived "heterosexuals"?
We'll never know how people behave in private. We can't ask -- they'll lie or, at the very least, keep it private, particularly if it's contrary to the image they're trying to maintain. Until their private behavior becomes public, in which case it's too late, we'll never know just what risk exists in any given situation.
I couldn't find a profile anywhere on the net that would allow a predictive risk-assesment of any type or sort of potential molester. There are profiles of their actions leading up to the molestation, but no profile of the molester -- they're gay, straight, celibate, old, young, middle-aged, rich, poor, and middle-class.
I think you're undercutting you're own arguement. People here are telling you this already.
Uh huh. That's why the 2% figure is still used for comparison vs. behavior.