The relative utility of fixed and variable risk factors in discriminating sexual recidivists and nonrecidivists

This study compared the relative utility of fixed and variable risk factors in discriminating between recidivist and nonrecidivist sexual offenders. Subjects were 95 adult male offenders released from the Canadian federal correctional system between 1988 and 1992. Risk factors from the Sexual Violence Risk--20 (SVR-20; D. P. Boer, S. D. Hart, P. R. Kropp, & C. D. Webster, 1997) were coded from prerelease institutional records; sexual and nonsexual violent recidivism was coded from postrelease police and correctional records. SVR-20 risk factors were categorized as fixed (static) or variable (dynamic) markers according to the criteria of H. C. Kraemer et al. (1997); the fixed risk markers were further divided into offense history and psychosocial factors. Hierarchical Cox regression survival analyses were conducted to compare the relative contribution of fixed offense history, fixed psychosocial, and variable psychosocial risk markers in accounting for any violent recidivism and sexually violent recidivism. Analyses indicated that fixed psychosocial factors added little to the models comprised fixed offense history factors alone. There was some evidence that variable psychosocial factors had incremental validity when added to predictions made on the basis of fixed factors, particularly in the prediction of sexual violence. The individual factors that were included in the final models are consistent with previous findings, and support the use of sexual deviance and antisocial lifestyle variables in the prediction of recidivism among sexual offenders.