Predicting Success in the Montgomery County Pre-Release Center: The Actuarial Efficacy of the Selection Suitability Scale

The rising costs of incarceration and a renewed interest in rehabilitation has prompted a resurgence of interest in community corrections. A major concern is determining which offenders are appropriate for community corrections without compromising public safety. The Montgomery County Pre Release Center (PRC) is a work release facility that offers comprehensive services designed to assist offenders with transitioning back to the community after a period of incarceration. The PRC uses the "Selection Suitability Scale" (SSS), a structured instrument created by PRC staff over 20 years ago, to ascertain which offenders are appropriate for admission to the institution. The SSS quantifies criteria believed to influence the applicant's probability of success in the PRC, and classify their level of risk to the community. Criteria include measures of criminal history, employment history, residential stability, as well as mental health and substance abuse. Those with higher scores on the SSS are hypothesized to be more likely to succeed in the institution. This study assessed whether the instrument predicted an offender's performance using three outcome measures, and whether the SSS, the total scale score and disaggregated by sub-category component score, predicted the applicant's performance above and beyond demographic and criminal history information easily obtained from institutional records. Using multivariate regression, three outcome measures of success were examined. These include whether the resident incurred an infraction, was discharged in good standing, and a composite scale score of 13 performance areas assessed by the staff during the resident's last month of program participation. Study subjects included 600 male (n=427) and female (n=173) residents from 2001 to 2004. The SSS performed as expected - those with higher scores on the scale perform better than those with lower scores. Further, the total SSS score provided a small improvement over demographic and criminal history factors alone. Likewise, several SSS component scores, depending on the outcome examined, are predictive. The general conclusion is despite the modest predictive power of the SSS, this should not chill additional experimentation either with this or other predictive tools. Study limitations, including that these results were not cross-validated and future research plans are explicated