I mean... David's colleague used to go on about something 3% - 5% of the population being child abusers (which presumably led to people accusing him of secretly being one, ouch).

He probably did this to try to get people to "take crime seriously". This kind of tactic is very patronizing, when you're trying to protect civil rights (a common one being privacy).

That particular number probably comes from surveys with tortured proxy variables, ambiguous questions, poor / inconsistent definitions, and selection bias (people with certain traits being more prominent than in the general population). Also, I've seen data, which if you were to interpret the same way, would produce 20%+, which is obviously nonsense.

Another possibility is that someone counts every "report" on the Internet as a "person". This is also a dodgy assumption.

Or it might rely heavily on participants from forensic / clinical / clinical adjacent settings? Let's suppose that a clinical setting already does research, participating in yet another study might be easier, in that case. I'm leaning more with the other two though.