The level of service inventory and female offenders : addressing issues of reliability and predictive ability
Brews, Albert Lawson
The legitimacy of classifying female offenders in the correctional system has been disputed (especially the application of male-normed risk assessment tools), and yet, there is a need to accurately determine the risk of re-offending and the criminogenic needs of the offender along with general and specific issues (i.e., responsivity) that will encourage successful program delivery. The Level of Service Inventory – Ontario Revision (LSI-OR; Andrews, Bonta & Wormith, 1995) is an assessment tool used throughout Ontario’s probation services and provincial institutions. Although the first edition of the LSI was based primarily on a male sample, later revisions included norms for female offenders based on samples spanning three continents (Blanchette & Brown, 2006). Although its reliability and predictive validity has been demonstrated across many field settings and offender populations, few studies (e.g., Rettinger, 1998) have addressed the question of predictive validity on a sufficiently large sample of female offenders to convince the skeptics of the LSI-OR’s applicability to women (Blanchette & Brown). The current study examined internal consistency, the ability to discriminate recidivists from non-recidivists with t-tests, and the capacity to predict recidivism with correlation and receiver operating characteristic analysis. The sample consisted of 2831 female offenders who were either released from a provincial correctional facility, completed a conditional sentence in the community, or completed a sentence of probation in Ontario during a one year period (2002/2003). Special consideration was given to female offenders from different disposition groups, with different racial backgrounds, with mental health issues and with prior victimization. The LSI-OR had very strong internal consistency and was able to distinguish offenders who committed a re-offence from those who did not commit a re-offence; both when considering the scale as a whole and when considering individual subscales. The LSI –OR was also found to predict recidivism for all female offenders. It also predicted recidivism for all subgroups with the exception of female offenders released on a conditional sentence and who had been previously victimized. While the use of the LSI-OR to assess provincial female offenders is supported, however, new risk levels are suggested to increase the predictive ability and reduce the potential for over-classification.
DegreeMaster of Arts (M.A.)
SupervisorWormith, J. Stephen
CommitteeWoods, Philip; Presse, Cindy; Farthing, Gerald
Level of Service Inventory