Testing the LS/CMI for Predictive Accuracy: Does Age Matter?

Abstract

The Level of Service/Case Management Inventory (LS/CMI) is one of the most widely used instruments for assessing recidivism risk and treatment needs in correctional settings. The predictive validity of the measure and its predecessor (LSI-R) has been established in meta-analytic studies and research finds that the scale's accuracy is largely independent of sex, race, and ethnicity. Whether the LS/CMI works equally well for different age groups remains in question. The current study assessed the predictive accuracy of the LS/CMI from a sample of 14,940 adults in custody (AIC) released from an Oregon prison between 2011 and 2017 for three age groups: 18 to 39, 40 to 54, and 55 and older. Study findings indicate that age is an important factor to consider when assessing an AICs risk for recidivism. Older adults in custody (55+) recidivated at significantly lower rates than other age groups. Mean LS/CMI score differences were found by age group. The LS/CMI was found to be equally accurate considering age, as higher risk scores were associated with higher recidivism across ages. However, findings also indicated that going beyond accurate scoring is important. LS/CMI cut-points may not be age-responsive, as similar LS/CMI scores produced differing levels of recidivism by age group. The LS/CMI may overclassify the risk of recidivism for older AICs and LS/CMI cut-points may need to be adjusted considering age. A discussion of the findings and how they can inform local policy and correctional assessment practices more broadly is provided.