Friday, November 7, 2014

Doing the numbers: How many pedophiles offend?

I've often said that <many pedophiles do not offend against children>. It is the foundation of much of the rest of our message.

It is possible to address this quantitatively. An average person's eyes are likely to glaze over, so let me give you the bottom line:

"Between 2% and 45% of pedophiles abuse a child at some point in their lives."

It's a vital enough question for us that I believe it's worth doing the analysis even if the results cover such a wide range.

Here's the method:

First you figure out how many sex offenses against children were committed in a given year. This is how many were reported, then multiplied by a factor to adjust for the percentage of such crimes that go unreported.

You then reduce it by the number of crimes that were committed by people who are not pedophiles but instead situational offenders (primarily attracted to adults). You reduce it further to account for those pedophiles who abuse multiple children in a year. The result is the number of pedophiles who abuse at least one child in a given year.

You then consider the population of the jurisdiction in question, the number of men, and the percentage of men who are thought to be pedophiles, and compute the number of pedophiles in the population.

Divide the first by the second and you get the percentage of pedophiles who commit an offense in a given year.

You then expand this to account for the many years in a pedophile's life. There are two extreme ways to do this. First, you can assume that each pedophile only abuses during one year, so the number of offending pedophiles is the number who abuse in a year multiplied by some large number like 50 (to cover age 15 through 65?). At the other extreme, you can assume that once an abuser, always an abuser -- an abuser abuses every year, so the number in one year is the same as the lifetime number. The second seems much closer to reality.

I made a little spreadsheet program to do all those calculations. So we have the theory. What about some data?

One of the German members of Virtuous Pedophiles [see note below] pointed us to <this article>. Given numbers in that article plus some he estimated and some others I estimated, we can turn the crank:

14,877 assaults on children reported in a recent year (2013?)
If we assume only 1 in 5 cases of sexual assault is reported, we get 74,385 total assaults on children.
If we assume 40% are by pedophiles, we get 29,754 assaults on children by pedophiles.
If we assume each pedophile abuser has 2 victims on average in a year, we get 14,877 pedophiles making at least one assault.
The male population of Germany is 40,000,000.
If we assume 1% are pedophiles, that is 400,000.
The percentage of pedophiles committing a crime in a year is 3.719%.
If we assume a factor of 3 for the lifetime adjustment, 11.16% of German pedophiles have at least one hands-on offense against a child in their lifetime. This means that 88.84% never abuse a child!

But as the old saying goes, "People fooling with statistics fool people with statistics."

The lifetime adjustment factor is particularly uncertain. But if instead we go back one step the percentage committing a crime in a given year (3.719%) is itself a meaningful statistic. Yet that number is also misleading for it counts among the non-offending pedophiles both small boys who will one day turn into pedophiles and the men who are too disabled to have an opportunity to offend.

Some of those numbers above are quite solid (number of assaults reported, male population of Germany). Most others have a range of variation but it won't affect the result beyond say doubling it or halving it (whether it's 20% of assaults by pedophiles or 60%, or whether the average pedophile has 1.3 victims per year or 4.0). Others have plausible values that would have a dramatic effect. These include notably the percent of pedophiles in the population (0.1% or 10%?), the number of unreported cases for every case that is reported (1, or 50?), and the lifetime adjustment (2 or 30?).

If we change the above analysis to say there are 10 unreported cases of child sex abuse for every one reported, and that only 0.5% of men are pedophiles, then it is 15% abuse in a given year, and the lifetime percentage is 45%.

You might also expect that sexual abuse by family members (mostly non-pedophiles) is less likely to be reported than abuse by people with a more distant relationship, so a lower factor of 3 would be reasonable there. Some have suggested that only 20% of child sex abuse is by pedophiles, and a high-end respectable estimate of pedophiles in the population is 3%. Turning the crank with those assumptions yields 0.37% in a given year (only 1 out of 250!). For such low numbers a higher lifetime adjustment makes sense, and using 5 yields 1.86% over a lifetime.

The above examples make clear that this kind of analysis depends critically on the values of many variables which are not known with any certainty. Citing a range between 2% and 45% of pedophiles who offend in a lifetime reflects that uncertainty, but I'd hardly be confident saying it has to lie in that range, or that the chances are 95% that it lies in that range. There is just too much uncertainty.

If a number as low as 0.37% of pedophiles abuse a child in a given year, you could ask how that differs from the number of males overall who will abuse. This number we can give with more confidence. It is 74,385 divided by 40,000,000, and if we assume just one victim per abuser we get 0.19%. It would be quite a revelation if pedophiles were only twice as likely to abuse a child as men in general!

I hope those with a green eyeshade mentality will feel free to comment on my methodology and estimates.

In closing, I'd like to return to the victims of the abuse, who barely made an appearance at the start of the analysis. If it turns out that only 2% of pedophiles abuse children, it in no way suggests that fewer children were abused than we thought. They suffer whether their abuser was a pedophile or not. While accounting for multiple victims reduced the number of pedophiles who abuse, it did not reduce the count or suffering of those victims. Victims are the most important people here, not pedophiles. But we as pedophiles inevitably focus on ourselves and our role. The large proportion of us who never abuse a child should not suffer for the actions of those who do.

6/26/2015. Max Weber is the VP member in question, and he points out that I did not give full credit to him by name or the source of those numbers he estimated, which can be seen in the July 28, 2015 paragraph here. My apologies.


  1. I have also tried to make an estimate of the offending rate. But it always comes down to the fact that we don't know how the abuse cases are distributed to all pedophiles.

    I think your estimate of 3 victims per year is too much. It is known that re-offending rate is 5% to 15%, so the figure would be much less than 1, even if there are some cases that will go on abusing yearly.

    And about the reporting rate, there are good studies made here in Finland. A recent study found that 5.4% of 16-year-olds reported acts that would be criminal under our law, corresponding to 3240 cases yearly. Police statistics show around 1200 cases reported per year. Number on reports have gone up, number of abuse cases down.

    If there were statistics on how many convicted child abusers there are, it would be an easy calculation. But I don't know if there are such statistics available.


  2. Thanks for the further information. I don't think I ever suggested 3 victims per pedophile per year -- rather I had 2. The "3" I used was for reporting factor.

    The re-offending rate is relative to being reported, tried and convicted, right? That is of course a huge disruption to a man's life, and might quite possibly "scare him straight". Those who are caught do often report multiple prior victims (even if the totals in studies like the Butner study are highly exaggerated). Those who are not caught could also be expected to often have multiple victims.

    The Finnish data is very interesting on reporting. You are suggesting using a factor of 3 instead of 5 for ALL abuse that is unreported? I had only used 3 for a best-case scenario for the subset of pedophiles.

    I would think that statistics on the number of convicted child abusers would be comparatively easy to get -- compared to many of the variables that have to be estimated, they at least exist somewhere! But I'm not sure why it's so important. We have the number of reports, and since in many cases there would be no conviction because of "reasonable doubt", the reports are probably a better indicator of how many crimes there actually were.

    It is all complicated, that's for sure.

  3. I don't think it is possible to really estimate. There are too many assumptions that could be wildly inaccurate. What's the expression--castles built on sand?

  4. I agree about the re-offending rate figures. But guessing the lifetime and per-year factors is just guessing anyway. It has no empirical data to back that up. So this kind of numbers are not going to convince anyone who doubts them.

    If I calculated correctly, the numbers seem to indicate about 1/2 of abuse cases being reported in Finland. There are about 60000 9th graders in Finland and 5.4% of them reported having experiences that fit the local law. This means 3240 cases per year. And the latest figure is from 2013, when 1657 cases were reported to police.

    The number of convicted abusers would include how many abuse cases are done my same abusers over their lifetime, i.e. the overlapping effect with re-offenders is accounter for. After correcting for the 40% and ratio of undetected cases (and multiplying by two due to the "triangle effect") we get a number that can be compared to number of pedophiles. There are of course still problems with this approach: we would have to assume similar statistics for the undetected cases and also the reporting rates have not been constant during the lifetime of current convicts. The ratio between reports and convictions can be estimated from empirical data.

    The problem with reports is that there is for sure no statistics for how many people these reports have been made.


  5. This comment has been removed by the author.

  6. Nestori also sent me a very interesting chart. I've never inserted a graphic in a post yet, and I'm enjoying the text-only constraint. BUt I found it in p169 of this document online:$file/Raportteja_110_lapsiuhritutkimus_web.pdf

    Though I don't read Finnish, I think it shows the age when the abuse happened. Not too surprisingly, this happens most often when a girl or boy is a young teen. If I read the 1988 chart properly, 1.3% of girls reported at all ages up through age 11, but an additional 5.7% reported being molested when they were 12, 13, or 14. Over four times as many. The ratio is even more pronounced for boys.

    This shows one reason why pedophiles are responsible for a lot less child sex abuse than you might think. Many of those 13- or 14-year-old girls will be well into puberty, and their abusers are often teleiophiles (ordinary men) and not pedophiles. Statistics will therefore be very sensitive to the top age that counts as "child". The higher the age, the less "child sex abuse" will be due to pedophiles.