Differentiating Risk of Violence and Treatment Needs Among Women In Prison

Background: Women exiting prison have complex needs for treatment and social support, as well as varying risks related to subsequent engagement in criminal behavior. Most of the risk assessment tools used within criminal justice institutions have been designed for and tested exclusively on male offenders. Beyond the inability to assess the more gender specific needs of women, these instruments often over-classify women's risk levels. However, with increasing attention to women's ‘use of force' and assaultive behavior, more administrators and policy makers are interested in tools that more accurately assess both needs and risks among women. To date, most of the literature has focused on two commonly used instruments (LSI & COMPASS) which have scant evidence predicting women's violent behavior. This study assesses the predictive utility of a less commonly used risk/needs assessment tool, the Self Appraisal Questionnaire (SAQ), in differentiating women's risk of violent behavior as well as their need for services and intervention.

Method: A survey was collected from a random stratified sample of incarcerated women (N=574) in one Midwestern prison. Completion of the survey was voluntary and anonymous. In addition to the SAQ, measures of risk included current/previous convictions and self reported uncaught violence. Measure of need included mental health, substance use and victimization history. T-tests and ANOVA's with post-hoc tests examined differences on SAQ scores across women with various levels of risk. Receiver Operating Characteristic (ROC) analysis tested the effectiveness of the SAQ in distinguishing differences. Marascuillo procedure was used for comparing proportions of women needing treatment. Results: The SAQ did not strongly differentiate risk when comparing women with and without a violent conviction. Similarly, the ROC analysis indicated low accuracy of SAQ in detecting women with violent conviction (AUC=.52). However, when measures of uncaught violent behavior were added to the analyses creating 4 differing risk categories, the SAQ demonstrated a fairly good accuracy in identifying high-risk women in ROC analysis (AUC=.75). Comparisons of the proportions of women who needed treatments among 4 groups showed that a greater proportion of women in the high risk groups needed treatment to alleviate risk factors associated with re-offending. However, 40–60% of the women in the highest risk group did not meet the published cut scores associated with treatment need across all of the subscales.

Conclusions and Implications: In research and practice, assessment tools are often used to classify individuals and determine need for treatment or eligibility, often with little attention to the validity of such tools. Women involved in the criminal justice system are often most at risk for misclassification. Although the SAQ's predictive ability regarding violent behavior was enhanced when adding information other than formal convictions, the tool's ability to classify women for treatment services is poor. Without placement in appropriate services it is unlikely that high risk status will diminish–perhaps increasing lengths of incarceration or failing to equip women with the skills necessary for successful community re-entry. Further investigation is needed to adjust the program assignment scores on SAQ subscales for women in prison and validate its utility.