Finkelhor (1994) found that internationally, estimates vary from between seven percent and 36 percent for women, and three and 29 percent for men. A random sample of 2,869 18 to 24 year olds in the United Kingdom found that 11 percent reported having been sexually abused before the age of 13 years (Cawson et al. 2000). Price-Robertson, Bromfield and Vassallo’s (2010) summary of Australian prevalence studies estimates that four to eight percent of males and seven to 12 percent of females experience penetrative child sexual abuse and 12 to 16 percent of males and 23 to 36 percent of females experience non-penetrative child sexual abuse.
The first thing that jumps about the statistics you cite is that the estimated ranges are huge. That's always a sign that data are unreliable, and it makes it impossible to know where the true figure lies.
By the way, when I Googled Finkelhor, one of the first results that came up was a 2009 study that followed on from his work, which found the estimated ranges are even more ridiculously wide than those you cite: "Thirty-eight independent articles were identified, corresponding to 39 prevalence studies; these articles report the prevalence of childhood sexual abuse in 21 different countries, ranging from 0 to 53% for women and 0 to 60% for men". In other words, collectively, these studies suggest that the number of child sex abuse victims in the world is somewhere between zero and 4 billion. Very informative. This just serves to highlight my original point - there are no reliable data about unreported offences.
It might be tempting to assume the true prevalence is somewhere in the middle of that enormous range, but that's not a statistically valid approach. The reason is, it's clear that at least some of these studies were flawed (the estimates clearly can't all be accurate). It's very possible - likely even - that the same methodological flaws were repeated in a number of these studies, and when that happens you will tend to see a cluster of similarly flawed studies all yielding similar results. On the face of it, the fact that several independent studies found the same thing seems to validate the results, but of course that's just an illusion if they all got inaccurate results due to the same flawed methods. So, for example, it's possible that all of the studies that concluded the problem is very widespread are quite simply wrong. This means the true prevalence could be anywhere within the range of 0-60%.
Abel et al. (1987) give median figures of 1.3 female and 4.4 male non-incest victims per child sex offender who targets children outside of their family .
The median number of victims per offender does not allow you to draw the conclusion you are trying to draw. Consider a sample of 5 offenders, who had the following number of victims each:
-1
-2
-3
-500
-1000
The median number of victims per offender in that sample is 3. However, the total number of people attacked by those 5 people is 1,506 - that's an average (mean) of over 300 victims per offender. And this isn't just a silly hypothetical example - some high profile sex offenders are known to have abused many, many victims (especially those who held positions of authority over children, which is where most of the non-incest abuse seems to have occurred).
And guess what? I Googled Abel et al 1987, and according to that study, the mean number of male victims per offender was 281. This supports the contention that a very small number of people are responsible for the vast majority of sex crimes - which is the exact opposite of what you were trying to say.
