The Level of Service Inventory-Ontario Revision: Risk/need assessment and recidivism.

The current research represents the first longitudinal study using the LSI-OR, which is a modified version of the LSI-VI, on a sample of 630 adult male offenders, namely, 454 inmates and 176 probationers. Subjects were administered the LSI-OR at intake, pretreatment, or for the purposes of internal programming and were followed for an average of 2.6 years. Any recidivism was defined as any reconviction for a new offence. Overall, the findings support the use of the LSI-OR as the risk/needs measure for institutional, community offenders as well as special offender groups, including females and young offenders. From a variety of psychometric analyses, the LSI-OR demonstrated more than adequate reliability estimates that were superior to those reported for the LSI-VI and the PCL-R. Adult male offenders from the institutional sample were more likely to recidivate both generally (61.9% vs 35.2%) and violently (27.1% vs 16.5%) compared to adult males who were under community supervision. Institutional offenders were also known to reoffend sooner than their community counterparts as demonstrated through survival analyses. The General Risk/Need Factor section which samples the major risk/need predictors correlated with general recidivism very significantly ( r = .39). The largest association with general recidivism was produced by the Criminal History subscale (r = .40). Special offender groups, namely, a mentally disordered, domestic violence, and sex offender group, were identified from LSI-OR items, and were subjected to several analyses. The General Risk/Need Factor total score and the Specific Risk/Need Factor section total scores differentiated recidivists and nonrecidivists for the special offender groups. Several stepwise multiple regression analyses were examined using the General Risk/Need Factor and the Specific Risk/Need Factor subscales, the LSI-OR Factors, and all the sections of the LSI-OR as the independent variables in separate analyses. Recidivism as the dependent measure was used as a dichotomous and continuous variable. The recidivism factors were also used in the regressions. The general pattern of predictors for general recidivism included criminal history, procriminal attitudes, and companions for the combined sample with a multiple R of .43. For the prediction of violent recidivism, the Antisocial Pattern and the History subscale contributed uniquely to the regression equation for the combined sample with a multiple R of .39. (Abstract shortened by UMI.