Glenn D. Walters, David K. Marcus, John F. Edens, Raymond A. Knight, Glenn M. Sanford

Recent research has suggested that a qualitatively distinct subtype of psychopathic sex offender can be identified via taxometric analyses (Harris et al., 2007). In this study we attempted to replicate the hypothesized psychopathic sexuality taxon in a group of 503 male sexual offenders using data from the Psychopathy Checklist-Revised (PCL-R:Hare, 2003) and five coercive and precocious sexuality items. Ambiguous to dimensional results were obtained when, in a replication of the Harris et al. (2007) study,dichotomized indicators were analyzed with summed input maximum covariance (MAXCOV). Clearly dimensional results, however, were obtained when higher correlating and more valid quasi-continuous indicators were analyzed with traditional (input variables not summed) MAXCOV, and both dichotomous and quasi-continuous indicators were analyzed with mean above minus below a cut (MAMBAC) and latent-mode factor analysis (L-Mode). These results suggest that Harris et al. (2007) may have mistaken the random fluctuations of weakly correlating and poorly differentiating indicators for a taxon. Consistent with the vast majority of earlier research,our results suggest that psychopathy (with or without coercive and precocious sexuality) is a dimensional construct.