Can science spot a paedophile?

If a psychological test could distinguish potential or actual child sex offenders from those with natural inclinations, would it be ethical to use it to screen applicants for jobs that involve contact with children? Many of us would surely answer yes to this question, and as a first reaction I would be among the assenters.

But there is a problem with a diagnostic devised by social psychologists at Radboud University who report that a combination of two tasks for implied sexual associations can distinguish with 90% accuracy paedophilic men from others with a sexual preference for adult women.

In their study, Matthijs van Leeuwen and his colleagues found that, in both the paedophile and control groups, a picture association test revealed positive and negative correlations with sex with either children under 12 years or adult women, respectively. What this means, say the researchers, is that paedophiles not only have positive associations with children and sex, but also negative associations with adult women and sex. In the control group the opposite is found.

The thing about picture association tests and similar is that they can so easily be assimilated into professional practice by people who haven’t a clue about anything much. I am thinking in particular about recruitment consultants and human resources executives.

Such tests have the imprimatur of scientific authority, and in the popular imagination they must be correct. What of the accuracy of the test? While a 90% confidence level may sound impressive to a layman, in scientific terms it is not a sufficiently strong correlation. A positive correlation which may be improved with further research, but still not good enough to make categorical statements about implied sexual associations.

To be fair on the researchers, they include a caveat concerning the serious ethical implications of their work. Still, I take issue with their claim that the implied sexual associations test is accurate as it stands. It isn’t, and it would be unethical to develop it into a diagnostic.

There is an understandable desire to prevent child sex abuse from occurring, but identifying potential offenders is fraught with moral and scientific difficulties. Mis-identification of healthy individuals as paedophiles-in-waiting would have disastrous consequences for them personally.

Is there an acceptable level of false-positive test results? If so, why and what? In other words, do the ends justify the means, and how many innocent adult lives is it OK to destroy in order to save children from sexual predators?