Predicting women’s recidivism

Abstract

Despite the growing number of women involved in the criminal justice system, most risk assessment tools used with this population were developed on male offenders, fuelling debate about whether these tools should be used with women. This study investigated the predictive validity of one such dynamic tool À the Dynamic Risk Assessment for Offender Re-entry (DRAOR) À with a sample of female and male parolees. Unexpectedly, the DRAOR subscales predicted recidivism for women, but not for men, and the acute risk subscale independently contributed to recidivism prediction. Furthermore, the DRAOR incrementally predicted women's recidivism above Downloaded from static risk. These results support the use of the DRAOR with women offenders, and more generally, a 'gender-neutral' approach to risk assessment tools for women.

Challenges for the theory and application of dynamic risk factors

“…One recent metaanalysis found that LS tools in fact had greater predictive validity for female offenders (mean effect size = 0.53) than for male offenders (mean effect size = 0.39), for both adult and youth populations (Andrews et al, 2012). Support for the gender-neutral assumption has also been found with risk assessment tools developed for offenders in community settings, again with researchers finding that the predictive validity of the Dynamic Risk Assessment for Offender Re-entry (DRAOR; Serin, Mailloux, & Wilson, 2012) was greater for female offenders than for male offenders (Yesberg et al,2015).The mixed results in research on gender and risk assessment highlight the need for further research in this area. Although overall results support the view that risk assessment tools are equally valid for males and females, it is important to investigate whether the inclusion of gender-specific items would further improve the predictive ability of risk assessment measures for females (Yesberg et al, 2015).…”

“…Support for the gender-neutral assumption has also been found with risk assessment tools developed for offenders in community settings, again with researchers finding that the predictive validity of the Dynamic Risk Assessment for Offender Re-entry (DRAOR; Serin, Mailloux, & Wilson, 2012) was greater for female offenders than for male offenders (Yesberg et al,2015).The mixed results in research on gender and risk assessment highlight the need for further research in this area. Although overall results support the view that risk assessment tools are equally valid for males and females, it is important to investigate whether the inclusion of gender-specific items would further improve the predictive ability of risk assessment measures for females (Yesberg et al, 2015). It is also important to acknowledge that the generalisation of risk assessment measures to other minority offender populations, such as different ethnic groups, has also been challenged by the results of various studies that suggest a reduction in the accuracy of risk-assessment tools for non-White populations within North America (Chenane, Brennan, Steiner, & Ellison, 2015;see Shepherd, Luebbers, & Dolan, 2013, for review).…”

“…Research focussing on the identification of important dynamic risk factors, and the development and validation of related risk assessment tools, is overwhelmingly conducted using samples of adult male offenders (Reisig, Holtfreter, & Morash, 2006;Schwalbe, 2008) under the general assumption that dynamic risk factors are broadly consistent across gender and age (for discussion of the theory underlying this, see Andrews, Bonta, & Wormith, 2006). This view has been described by Yesberg and colleagues as the "gender-neutral" assumption (Yesberg, Scanlan, Hanby, Serin, & Polaschek, 2015) and is currently the predominant view within the area of offender risk assessment. However, a number of scholars have rejected the genderneutral assumption, instead asserting that there are a number of important differences between men and women which challenge the idea that dynamic risk factors -and therefore measures of dynamic risk -informed solely by male populations are equally valid for females.…”

“…However, a number of scholars have rejected the genderneutral assumption, instead asserting that there are a number of important differences between men and women which challenge the idea that dynamic risk factors -and therefore measures of dynamic risk -informed solely by male populations are equally valid for females. These includes differences in motivations for offending (Veysey & Hamilton, 2007), experiences of victimisation (Reisig et al, 2006), base rates of offending (Odgers, Moretti, & Reppucci, 2005) and pathways to crime (Belknap, 2014; also see Yesberg et al, 2015). A meta-analysis that investigated the predictive validity of the Level of Service InventoryRevised (LSI-R; a widely-used dynamic risk measure in North America) for both male and female samples provided support for this view (Holtfreter & Cupp, 2007).…”

Measurement of Change in Dynamic Factors Using the START: AV

“…The acute dynamic risk factors also significantly contributed to the prediction of recidivism, but yielded better predictive accuracy when averaged over a 6-month period, suggesting that the acute dynamic factors were more stable than anticipated (Hanson et al, 2007). However, the research on risk assessment in juvenile justice on acute and stable dynamic factors is scant; such investigations appear limited to adult sex offenders (Beech & Ward, 2004; Blacker, Beech, Wilcox, & Boer, 2011; Hanson & Harris, 2000; Hanson et al, 2007) or adult offender/parolee populations (Yesberg & Polaschek, 2015; Yesberg, Scanlon, Hanby, Serin, & Polaschek, 2015). …”