On stats and chatroom paedophiles

Letter Colin Z. Robertson questions my grasp of statistics:

In "Watch out! There's a chatroom paedophile about" you wrote:

"According to US research, cited by Donald Findlater, a therapist involved in the rehabilitation of convicted paedophiles, as many as a third of child pornography consumers are, concurrently child sexual abusers. Which means that, conservatively, more than 2,000 children were being sexually abused in 1999 by the 7,000 British subscribers."

uhm, no.

The phrase "as many as a third" means that one third is the upper limit. So, *at most*, 2333 of those 7000 subscribers were also abusers.

Now, we haven't been told what the lower limit is. It could be 0% of consumers for all we know. So, of those 7000 subscribers, the actual number of abusers could be anywhere between 0 and 2333.

Note that this doesn't tell us *anything* about the number of children abused. You seem to have assumed that each abuser abuses roughly one child on average. This may be true. The number may be larger than that. If abusers collaborate to abuse a child then the number may be lower than that. We don't know.

So for you to say that more than 2000 children have been abused by these 7000 subscribers is really quite a stretch. Furthermore, to claim that this is a conservative estimate is flat out wrong. The most conservativeestimate you could make from this data is that 0 children were abused.

I'm not claiming that that's actually the case. What I am claiming is that on the basis of the data provided in your article we simply don't know what the numbers are.

I know we're dealing with an unpleasant topic here, but that's no reasonto ignore the basic rules of statistics.

Also, it would really help if you'd provide a reference to this "US research". Anyone can wave some figures in the air. It would give yours a lot more authority if you could *show* that they were backed up by research.

colin.

Sponsored: 10 ways wire data helps conquer IT complexity