An Evaluation of Risk/Need Factors for Adult Women with Violent Offenses

The present study examines the validity, reliability, and psychometric properties of forensic risk [Violence Risk Scale (VRS), Psychopathy Checklist-Revised (PCL-R), Saskatchewan Primary Risk Assessment (SPRA), Ontario Domestic Assault Risk Assessment (ODARA)) and protective tools (Structured Assessment of Protective Factors(SAPROF)] as well as the importance of including gender informed variables (co-offending, suicidality, substance use) in risk assessment for adult justice involved women with histories of violence. The total sample consisted of 246 provincially sentenced women in Saskatchewan who had perpetrated a violent index offence in 2018 or 2019. Of the 246, 188 cases had enough information available within corrections files to be rated on the forensic tools and gender informed variables. Recidivism information was collected after file-based ratings were made and was available for the entire sample (N = 246) over a mean follow-up time of approximately five years. The forensic measures and their component/factors demonstrated good inter-rater reliability (ICC = 0.81 – 0.98). Generally, there was also strong convergent validity amongst the forensic tools, with correlations in the large range. Predictive accuracy evidence was obtained for all five tools with respect to violent, non-violent, and general recidivism at fixed follow up times and for total recidivism. Area under the curve (AUC) values varied based on the tool and the type of recidivism, but generally fell in the moderate (AUC = 0.64 – 0.70) to large (AUC = 0.71 – 0.83) range for the total sample. Calibration analyses further supported the ability of the VRS and SPRA to differentiate between low, medium, and high-risk justice involved women (depending on recidivism type). Depending on recidivism and the tool it was compared against, the gender informed variables uniquely predicted recidivism suggesting the importance of considering these variables when examining risk for justice involved women. There were high rates of intimate partner and familial violence, and gang affiliation was significantly related to higher rates of recidivism. The present study provides vital evidence for the use of these forensic tools and gender informed variables to inform risk, needs, and responsivity variables in justice involved women. Limitations and future directions are discussed.