Evaluating the inter-rater reliability of the Violence Risk Scale-Sexual Offense Version ( VRS-SO ) in a community-based treatment setting 1

Methods to assess risk of reoffending and guide rehabilitation efforts for offenders have been evolving over several decades, resulting in the development of a number of specialised tools. Evaluating the validity and reliability of these tools across contexts is imperative, in order to have confidence in the accuracy of risk assessments and the decisions they inform. One such tool is the Violence Risk Scale-Sexual Offense version (VRS-SO), designed to assess both risk of reoffending and change across treatment among sex offenders. Although validated on several incarcerated samples, to date the validity of this tool in community based settings has not been evaluated. This study reports on an exploration into VRS-SO inter-rater reliability, carried out as part of a wider community validation study. Clinician-scored pre-treatment VRS-SO ratings (n = 8) were gathered over an eight-month period, and each case was also rated independently by a researcher using file information. Rater consistency was analysed using the intraclass correlation coefficient (rICC). Unstructured interviews of clinicians were also undertaken to obtain valuable user perspectives. Our results indicate strong preliminary support for VRS-SO inter-rater reliability in a community treatment setting (e.g., ICCc,2 = .98, p < 0.001for total scores), and add to the existing evidence base further given the use of prospectively collected clinical ratings. Item-level analyses suggest that modifications for the community setting and/or for clients whose offences were internet-based could further enhance reliability. The findings inform applications of the VRS-SO with community-based clients as well as potential future modifications for this context.