Examining pedophilic interest as a risk factor for sexual re-offending

By Ian McPhail (@IanVMcPhail on twitter)

One of the central preoccupations of forensic psychology is identifying what psychological and social characteristics may contribute to re-offending.  This concept is straightforward and important.  When someone commits an offence, is caught and convicted, and becomes involved in the criminal justice system, we want to understand what separates those who return to criminal behaviour and those who do not.  When we understand this, people working with individuals who have committed sexual offences can help them limit the influence of these problems in their lives and increase the chance that they will desist from crime.

Research trying to identify these so-called risk factors will typically look to theoretical accounts of what causes crime and then will examine whether theorized psychological and social characteristics are indeed linked to people continuing to offend after being caught and punished. If a risk factor predicts future re-offending, then its status as empirically supported is bolstered. If a risk factor does not predict, then it moves closer to the waste bin for interesting, but ultimately unsupported ideas.

The best way to do this research is to measure psychological or social characteristics in a group of criminal justice-involved people, wait for a period of time after they are released from prison into the community, then see who in this group has re-offended and whether the characteristics are higher in those who re-offend compared to those who do not re-offend.  We call this a prospective research design.

The next best way to do this research is to look back into the files of a group of criminal justice-involved people, measure the characteristics of interested in from the file information available, and see if those who have re-offended have higher levels of the characteristics.  This is what we call a retrospective research design.

Pedophilic interest as a risk factor for sexual re-offending

For a long time, clinicians and researchers have theorized that having a sexual interest in children, or pedophilic interest, will make it more likely that someone who has been caught and convicted of a sexual offence will commit another sexual offence after returning to the community.  This probably aligns with our common sense: Someone who is sexually interested in children and has offended, if they continue to be sexually interested in children, then they are more likely to re-offend compared with another person with a sexual offence history who does not have this sexual interest.

In science, it is important to examine whether common sense is true and to expose hypotheses to more and more severe tests.  When I examine the available literature, an interesting and counter-intuitive finding emerges: Being diagnosed as having pedophilic interest does not necessarily confer risk to sexually re-offend.  How can that be?  As with a lot of science, the devil is in the nuances.

Using latent structure research to light the way

In my last post on nextgenforensic, I reviewed some research that has examined the latent structure of pedophilic interest. I also provided some of the findings by my colleagues. To briefly recap, my colleagues and I found that pedophilic interest is categorical: there are distinct classes (or, “taxa”) of people in terms of their sexual interest in prepubescent children. The classes we identified were those who are not pedophilic at all, those who were non-preferentially pedophilic, and those who were preferentially pedophilic.

When we looked at the sexual recidivism rates in these three classes, when found that the sexual recidivism rates were 29%, 27.5%, and 50% for the non-pedophilic, non-preferential pedophilic, and preferential pedophilic classes, respectively. This pattern of results indicates that men who are not pedophilic and those who are non-preferentially pedophilic re-offend at similar rate, while preferentially pedophilic men re-offend at much higher rates.

[T]his pattern of re-offence rates was a striking finding and suggests that it is not pedophilic interest per se that is a risk factor, it is … the preference for prepubescent children that is the risk factor.

Now, there are a number of reasonable explanations for this pattern of results in our sample. Maybe the non-pedophilic men had problems with other risk factors that were not measured in the study, contributing to their re-offence rate being roughly equal to the non-preferentially pedophilic class. Or, the sample size of preferentially pedophilic men was so small that their high rate of re-offend was an anomaly that won’t replicate.

However, this pattern of re-offence rates was a striking finding and suggests that it is not pedophilic interest per se that is a risk factor, it is the exclusivity of pedophilic interest or the preference for prepubescent children that is the risk factor. Given this possibility and the uncertainty in our initial findings, I wanted to follow-up on our finding to see if this pattern emerges in other studies examining pedophilic interest as a predictor of sexual recidivism.

Testing whether non-preferential pedophilic interest is a risk factor for sexual recidivism

Let’s take a look at some of the research that has looked at rates of sexual recidivism in non-pedophilic sexual offenders and non-preferentially pedophilic sexual offenders. Broadly speaking, someone who is non-preferentially pedophilic has a somewhat equal level of interest in prepubescent children and adults. Someone who is non-pedophilic is exclusively attracted to adults. The results from five studies examining whether non-preferential pedophilia is related to sexual re-offending are presented in the table below.

Study Assess-ment method N d Non- Pedophilic Offenders Non-preferentially Pedophilic Offenders
Moulden et al. (2009) DSM 164 –0.007 ~38% ~38%
Moulden et al. (2009) PPG 157 0.161 ~38% ~38%
Eher et al. (2010) DSM 118 –0.040
Wilson et al. (2011) PPG 45 0.152 8.8% 9.1%
Wilson et al. (2011) DSM 97 0.315 10.0% 16.4%
Stephens et al. (2017) PPG 443 0.062 9.1% 9.8%
McPhail et al. (2018) PPG 235 –0.013 29.1% 27.5%
Meta-analytic average 1,057 0.029

Here’s a short guide for the results table.  The first thing to point out is that the men in these studies have all been convicted of a sexual offence. The studies examined rates of sexual re-offending after these samples of men were released from custody after serving a sentence for a sexual offence.

The names of the authors of the study are in the first column; which are hyperlinked to the cited studies.  The second column provides the measurement method used to diagnose pedophilic interest. I included here studies that used two methods of diagnosing pedophilia: the Diagnostic and Statistical Manual of Mental Disorders (labelled “DSM” in table) and phallometric testing (labelled “PPG” in the table).  These two methods are routinely used to identify pedophilic interests and represent reasonable measures of this sexual interest.

In the third column, the total sample size is given.  In the fourth column, the standardized mean difference is provided (i.e., d).  This statistic is a way of quantifying the difference in rates of re-offending between the non-preferentially pedophilic and non-pedophilic men (the recidivism rates are provided in the last two columns).  To help interpret this statistic, a positive d value indicates higher sexual re-offence rates by non-preferentially pedophilic men.  A d of 0 means no difference in rates of sexual re-offence.  And a negative d indicates higher sexual re-offence rates by the non-pedophilic men. If non-preferential pedophilic interest is a risk factor that predicts sexual recidivism, we expect the d to be positive.

The consistent finding is that non-preferential pedophilia is generally not related to sexual re-offending. When I compute a meta-analytic average that combines the ds for these five studies (the last row of the table), the average difference in sexual re-offending between these two groups is almost nil (d = 0.029). Even though the overall rate of recidivism was somewhat different across the studies (e.g., the overall rate was ~9% in Wilson et al., but it was 38% in Moulden et al.), almost all the d values are around 0 and the re-offence rates in both groups are similar.

Making sense of the data

Researchers have proposed a rubric for identifying when a characteristic can be considered an “empirically supported risk factor” or when a characteristic becomes an “unsupported risk factor”. A so-called empirically supported risk factor is one that has been studied 3 or more times, using the research designs I mentioned above, and reasonably differentiates between re-offending and non-re-offending people. In statistics speak, “reasonable” means that when a meta-analysis of these studies is performed, the aggregate standardized mean difference (d) is about |0.15|.

A so-called unsupported risk factor has been studied 5 or more times and this characteristic does not differentiate those who re-offending from those who do not.  That is, the meta-analytic average standardized mean difference is less than |0.15|.  While this rubric represents a consensus agreement between three researchers (and I will be honest by saying I am not completely convinced of its value), I will use it as a general heuristic to interpret the data on whether non-preferential pedophilia is a risk factor for sexual re-offending.

[T]he empirical evidence accrued to date tells us that non-preferential pedophilia is unsupported as a risk factor for sexual re-offending.

When I compare the meta-analytic average, taken from 5 studies, to the criterion for an unsupported risk factor, the average d = 0.029 is well under the suggested d = |0.15|.  In fact, an average effect size this small suggests there is almost no difference in sexual re-offence rates in non-preferentially pedophilic and non-pedophilic men.

Taken together, the empirical evidence accrued to date tells us that non-preferential pedophilia is unsupported as a risk factor for sexual re-offending.  To me, this finding is striking because it goes against common sense assumption that those with pedophilic interests are at higher risk to re-offend. It also speaks against non-preferential pedophilia as a core theoretical construct in our understanding of why some people continue to commit sexual offences.  This finding should prompt a reconsideration of non-preferential pedophilia as a risk factor for sexual re-offending.

Why might non-preferentially pedophilic men not be at elevated risk to re-offend?  Is it because being non-preferentially attracted to children means people have some significant degree of attraction to adults? And it is this ability to be attracted to adults that helps these men pursue satisfying relationships with adults which help them to avoid re-offending?  These are all interesting questions that require further deliberation and study.

Another reason why this finding is interesting is that some uses of pedophilia as a marker of elevated risk for re-offending is not supported by the available evidence.  To use non- preferentially pedophilia in this way is to practice in a manner inconsistent with what the evidence tells us.  I imagine this argument is likely to ruffle some feathers.  However, the above evidence does not support non-preferential pedophilia as a risk factor and has potentially large ramifications for clinical practice.

Wrapping up

In this post, I used a recent study my colleagues and I did, which examined the latent structure of pedophilic interest, as a guide to examine some of the existing science regarding the status of pedophilic interest as a risk factor for sexual re-offending. The findings I report here partially support the three-class structure to pedophilic interests we found. The data I presented also reinforce the need to follow-up latent structure studies with research examining practical applications of latent structural results.

In my next post on nextgenforensic, I will continue to follow-up our recent study and examine the question: what does the available evidence say about exclusivity of pedophilic interests as a risk factor for sexual re-offence?  Stay tuned…


Suggested citation:

McPhail, I. V. (2018, December 1). Examining pedophilic interest as a risk factor for sexual re-offending[Weblog post].  Retrieved from


Want to submit your own post? Click here to find out how!

%d bloggers like this: