Evaluation of the LSI-R as a Risk Assessment Tool in Utah
Study Dates: 2013 Study Description: This report is divided into two major sections. Section one describes the need for accurate, valid risk and needs assessments, and provides a literature review of the most well-known, validated and commonly used instruments as well as a review of tools that are relatively new but show promise. The second section was originally intended to validate the LSI-R in Utah’s population and within demographic subgroups (i.e., race, ethnicity and sex). It was also intended that the research would examine whether Utah’s potential adoption of the newer and broader LS/CMI, would improve, reduce or simply maintain the current tool’s ability to predict recidivism. While the issue of validation within Utah will remain an important area of future research, preliminary analyses of LSI-R data provided to UCJC revealed problems that precluded a fair and accurate assessment of the validity of either instrument. Problems included both data quality issues (i.e., invalid tests due to too many missing items, total scores that did not match the sum of item-level responses, prohibited combinations of item-level responses) and administrative issues (i.e., inadequate training of administrators, inherent difficulties in using the LSI-R, and ambiguities on the LSI-R response form). Rather than using inaccurate data in attempting to validate the instrument(s) for use as recidivism risk prediction tool(s) in Utah, the present research necessarily altered focus to examine the extent of the problems.