The Predictive Accuracy of the LSI-R in Female Forensic Inpatients—Assessing the Utility of Gender-Responsive Risk Factors

Abstract

Female reoffending has long been a neglected research interest. Accordingly, risk assessment instruments were developed based on the criminological knowledge of male recidivism. While feminist researchers have repeatedly criticized the failure to incorporate gender-responsive risk (GR) factors, opinions on the gender neutrality of existing instruments remain inconsistent. In order to substitute the existing literature, while extending the scope to mentally disordered offenders, the aim of the given study was the prediction of general recidivism in a sample of 525 female forensic inpatients who had been discharged from forensic psychiatric care in Germany between 2001 and 2018. Primarily, ROC analysis was conducted to assess the predictive accuracy of the LSI-R. Subsequently, separate binary logistic regression analyses were performed to determine the predictive utility of GR factors on recidivism. Lastly, multiple binary logistic regression was used to assess the incremental validity of the GR factors. The results showed that the GR factors (i.e., intimate relationship dysfunction, mental health issues, parental stress, adult physical abuse, and poverty) significantly contributed to the prediction of recidivism, while a mixed personality disorder, a dissocial personality, an unsupportive partner, and poverty added incremental validity to the predictive accuracy of the LSI-R. However, given that the added variables could only improve classification accuracy by 2.2%, the inclusion of gender-specific factors should be cautiously evaluated.