| Previous research has suggested that child molesters are quite heterogeneous with regard to their personality functioning and offending style and that the Minnesota Multiphasic Personality Inventory (MMPI) may have limited utility in differentiating patterns among them. However, cluster analysis, which is used to empirically develop taxonomies, has been used infrequently with child molesters. This study cluster analyzed the MMPI data of 310 child molesters who were being evaluated on an outpatient basis for a variety of purposes. A hybrid clustering method was used in which the data was initially analyzed with a group average cluster analysis. A K-means procedure was then conducted, using mean values generated from the group average cluster analysis as starting values. This analysis resulted in ten distinct profile types, which were cross-validated across two randomly selected subsamples of the entire sample. This hybrid clustering method appeared sensitive to issues of profile shape, as well as overall profile elevation. The level of psychopathology, as indicated by the mean MMPI profile, ranged from slight to extreme across the clusters. The ten profile types were then found to differ with regard to: demography; history of childhood victimization; use of denial; intelligence; deviant physiological arousal to pedophilic audiotape stimuli; deviant sexual functioning, as measured by the Multiphasic Sex Inventory and by interview report; use of cognitive distortion and justification; and social skills, as measured by the Interpersonal Behavior Survey. A number of offense-specific variables were examined (e.g., age and sex of the victim, relationship to the victim), but were not found to differ across the different profile types. The results were thought to indicate that the MMPI is useful in research with child molesters and, with cluster analytic methods, can be used to develop a replicable taxonomy, as well as examine the ovariation of pertinent factors within offenders. The results of this study were compared to previously published cluster analyses of sexual and nonsexual criminal offenders. Theoretical, clinical, forensic and methodological implications of the research were discussed. Suggestions for future research were made. |
