This paper looks at the differences between three groups of male adolescent offenders: sexual offenders, assaultive offenders, non-violent offenders. A group of non-delinquent adolescents were used as a control. Their primary finding was that while assaultive offenders hang out with a peer group which is also delinquent, the family relations and behavioral functioning sexual offenders, on the other hand, were much like the nondelinquent adolescents, the main difference being that they are more isolated.
One interpretation of this data is that assault is a behavior which is strongly influenced by a peer group, while sexual offences are more driven by internal processes. This interpretation is consistent with most of the experience that I am familiar with. Of course, there are other valid interpretations. For example, it could be that the same characteristics that create a tendency to assault also increase the tendency to associate with other violent offenders.
I found the paper itself fairly hard to read, couched as it is in thick social science language. However, the conclusions are interesting, especially because they both challenge some of the accepted wisdom and fit with some of the clinical evidence, and because the study was fairly carefully done.
Peace,
Somebody Somewhere