An Exploratory Analysis of Centres for Men Authors of Violence in Apulia Through Bayesian Graphical Modelling

Rehabilitation programmes for perpetrators of violence against women are a key component of prevention policies; however, assessing their effectiveness remains challenging due to complex psychological and social dynamics and limited data availability. This study presents an exploratory analysis of individual-level data from centres for men authors of violence in the Apulia region (Italy), covering accesses from 2022 to 2024. We employ a Bayesian Graphical model approach to investigate conditional dependence patterns among dimensions such as self-reported types of violence, denial, programme dropout, and operator-assessed victim risk levels. This framework enables the identification of a sparse set of relevant associations and provides a measure of the evidence supporting each link (edge) between violence dimensions and key programme-related variables. Results highlight several significant associations, with denial emerging as a central factor associated with programme dropout and co-occurring with more severe violence profiles and higher victim-risk assessments.