Using Latent Class Analysis to Identify Need Typologies and Recidivism Likelihood Among Women Who Perpetrate Violence

Abstract

Understanding the heterogeneity of women who engage in violence is critical to provide effective treatment and reduce the likelihood of recidivism. Existing typologies of women who engage in violence have been created using mixed methodological approaches; the field would benefit from replication using a quantitative clustering method—latent class analysis (LCA)—as it is arguably more objective than methods used to date. A LCA was conducted using archival data involving 3,773 justice-impacted women in Western Canada to identify unique subgroups of women who perpetrate violence. Three distinct profiles emerged: (a) intimate partner violence (IPV)-only (40.5%), wherein almost all women reported only perpetrating domestic violence and had zero or only one previous violent conviction, (b) patterned (19.1%), wherein violence was perpetrated toward domestic partners and unknown victims, and the majority had two or more previous violent convictions, and (c) isolated (40.4%), wherein very few perpetrated domestic violence, some perpetrated violence toward unknowns, and the majority had either zero or only one previous conviction for a violent offense. Need profiles and recidivism outcomes were further analyzed as a function of group membership. As hypothesized, the group with the greatest criminal history and use of violence reported the greatest needs. Recidivism also increased as the number of dynamic needs increased. Notably, 80.9% of the sample was predominantly low risk/low need and were identified as IPV-only or isolated women. Implications of these findings may be used to inform risk classification, treatment targets, and treatment intensity required to reduce the likelihood of recidivism among women who perpetrate violence.