Preliminary development of a dynamic risk assessment tool for women offenders: an examination of gender neutral and gender specific variables

Risk assessment measures have largely been developed using samples of male offenders. When applying these measures to women offenders there is a risk in misrepresenting women’s needs and over-classifying their levels of security and required intervention. As such, there is a need to develop risk assessment tools that are specific to women and reflective of women’s diverse risks and needs.

The goal of the current study was to develop a dynamic risk assessment scale specific to women offenders that could be used to monitor progress in the community and predict risk of recidivism upon release.

Gender-informed, criminal risk variables were examined for their association with recidivism in a sample of 309 released federal women offenders. Recidivism was defined as any new conviction within a two year period after release. Logistic regression procedures were then used to develop a scale that could be used to predict recidivism amongst the women. A second cohort of women released the following year (n = 256) was used to examine the generalizability and validity of the predictive model.

Eight of the candidate variables contributed to the final predictive model. These included: previous adult convictions, previous revocation of conditional release, average number of major institutional incidents per year, unstable employment history, previous break and enter conviction, contraband charge during current sentence, previous assault conviction, and previous youth convictions.

Alternative, simpler scales were also considered using these variables and all three scales were checked for their predictive accuracy between the two samples of federal women offenders. Results revealed a substantial decline in the model’s predictive ability between the two samples.

Ultimately, a gender-informed, dynamic risk scale could not be validated. The variables that contributed to the initial model were neither dynamic in nature, nor were they those variables considered to be traditionally gender-informed. This could partly be due to a number of limitations that occurred throughout the study. These included: the second source nature of the data collected, the small sample of women offenders, and the decline in the model’s predictive accuracy.

These results demonstrate that further research initiatives are required in attempts to develop a risk assessment tool specific to women offenders. Possible future endeavors could include: collecting more in-depth data (e.g., through interviews) concerning risk factors salient to women; considering the implementation of a static risk measure; and, a validation study that examines an already existing tool for its possible use with women.

Zakaria, D., Allenby, K., Derkzen, D., & Jones, N. (2013). Preliminary development of a dynamic risk assessment tool for women: An examination of gender neutral and gender specific variables. Research Report R-280. Ottawa, ON: Correctional Service of Canada.

To obtain a PDF version of the full report, or for other inquiries, please e-mail the Research Branch or contact us by phone at (613) 995-3975.

You can also visit the website for a full list of research publications.

Prepared by: Kim Allenby