Examining the latent structure of pedophilic interest

By Ian McPhail (@IanVMcPhail on twitter)

In the last five years, there has been a series of attempts by forensic and sex scientists to examine and elucidate the latent structure of sexual interests in prepubescent children, or, pedophilic interest. In this blog post, I will discuss what latent structure is and what the recent science has been finding. In two upcoming blog posts, I will examine the ramifications of this recent latent structure research.

What is latent structure in psychological constructs?

When scientists talk about latent structure, we typically mean how some characteristic shows up in people. One possibility is that a characteristic is either present for some people and absent for others (and so has a “categorical” structure). Another possibility is that all of us have the characteristic to varying degrees, with some of us having a lot of the characteristic and some of us having very little of it (and so it has a “dimensional” structure).

Scientists can use a variety of statistical techniques to identify which kind of structure, categorical or dimensional, likely fits the characteristic under study. This research can help us reach conclusions about whether we should think about a characteristic as categorical or dimensional. Without such data, we are left to simply make an assumption about a characteristic’s structure and hope for the best.

Latent structure in pedophilic interest

Why is it important to scientifically examine latent structure and what does this have to do with pedophilic interest?

Let’s take the example of diagnosing pedophilic interest in clinical practice. If a clinician makes a diagnosis like, “This client has pedophilic interest”, they are assuming categorical latent structure: people are either pedophilic or they are not pedophilic. Without data and empirical examinations of the latent structure of pedophilic interest, these decisions are based on a theoretical assumption. The problem is that, without empirical data to support it, this assumption could be wrong and we wouldn’t know it is wrong. In forensic contexts, basing clinical decisions on a wrong assumption has large ramifications for the person being assessed and for public safety.

Without such data, we are left to simply make an assumption about a characteristic’s structure and hope for the best.

Data on latent structure can help weed out the incorrect assumptions and identify more correct ones. One implication is that improving our understanding of latent structure will lead to more accurate and fair decisions in clinical practice.

What does the science say?

To date, there have been three robust empirical examinations of the latent structure of pedophilic interest. Schmidt and colleagues found that pedophilic interest is dichotomous: people in their sample had pedophilic interests or they did not. Stephens and colleagues found that pedophilic interest is dimensional: people in their sample showed varying degrees of pedophilic interest.

My colleagues and I recently found that pedophilic interest was categorical, but it was trichotomous: some people in our sample did not have pedophilic interests, while some people did. However, in those who did experience pedophilic interests, there were two distinguishable categories. Taken together, this suggested that the latent structure of pedophilic interests has 3-classes of people. When we did some further analyses, we found that these three classes could be characterized, broadly speaking, as: those interested in adults, those who were non-preferentially pedophilic, and those who were preferentially pedophilic.

I think this is a situation where scientists’ oft-repeated phrase, “Further research is needed,” is applicable.

What does this research mean?

We currently have three empirical examinations of pedophilic interests. Two studies reached a categorical conclusion: one finding two categories, the other finding three. One study reached a dimensional conclusion. Obviously, this is not the clearest picture we could have after three investigations. I think this is a situation where scientists’ oft-repeated phrase, “Further research is needed,” is applicable. To be confident in our conclusions, we need to conduct a few more empirical examinations of the latent structure of pedophilic interest.

One other way forward is to start testing these models to see which produce consistent and meaningful results. Because we have three models that have received support, if one model fits the latent structure of pedophilic interest better, this model should produce better results in research. I intend to examine the ramifications of the past results in two follow-up blog posts on nextgenforensic. For the moment, I will set up the background for the upcoming posts.

In our study on latent structure, there was an interesting pattern to the sexual recidivism rates across the 3 classes of pedophilic interest. (Sexual recidivism is when a person who has been convicted of a sexual offence is released from custody after serving a sentence for that offence, then is caught for committing a new sexual offence.) When the sample was grouped according to the non-pedophilic, non-preferentially pedophilic, and preferentially pedophilic classes, the sexual recidivism rates in these groups were 29%, 27.5%, and 50%, respectively.

When I examined this pattern of sexual recidivism rates, I noticed that non-pedophilic offenders and non-preferentially pedophilic offenders re-offended at almost the same rate. While the preferentially pedophilic offenders had a much higher rate of re-offending.

This was a striking finding. I say striking because if pedophilic interest is a risk factor for sexual recidivism, we would expect men who experience any pedophilic interests to have higher rates of sexual recidivism.  Indeed, two meta-analyses suggest that peodphilic interest is one of the strongest predictors of sexual recidivism science has found to date (see here; a meta-analysis I co-authored is here). Striking findings are those that go against expectations and this result is one that warrants further examination.

In my next posts on nextgenforensic, I will further examine the ramifications of latent structure. Specifically, I will test whether these rates of sexual recidivism hold across other studies that have looked at pedophilic and non-pedophilic sexual offenders. Stay tuned…

Suggested citation:

McPhail, I. V. (2018, November 17). Examining the latent structure of pedophilic interest [Weblog post]. Retrieved from


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

%d bloggers like this: