4 101 REPLICATION AND COMPARISON NETWORKS Fifth, network structures and metrics of networks of dynamic risk factors might differ across subsamples of individuals with a history of sexual offenses (e.g., intimate partner violence vs. nonpartner violence; Borsboom et al., 2017; van den Berg et al., 2020). However, we were not able to differentiate between networks based on offense type (e.g., sexual abuse of children, rape, exhibitionism, or sexual violence in intimate relationship) or whether participants received mandated treatment as these characteristics were not available. Given Canada’s correctional programming, it would be expected that most (if not all) would have received some form of treatment. Sixth, in this study, networks are undirected regarding relationships between dynamic risk factors due to the use of cross-sectional data (i.e., one-time measurement of dynamic risk factors). This makes it impossible to determine whether the activation of a dynamic risk factor causes or is caused by the activation of another dynamic risk factor, or whether the causal impact is bidirectional. 4.4.2 FUTURE RESEARCH DIRECTIONS Despite the above limitations - assuming the estimated network structures indeed describe causal interactions, and that “risk of sexual reoffending” is the result of such interactions - our study provides guidance for future research and theoretical work. First, the community analyses revealed four distinct communities of dynamic risk factors concerning sexual self-regulation, emotionally intimate relationships, antisocial traits, and self-management. A network-based theoretical account of risk of sexual reoffending should include these communities. Not the least because they are largely congruent with risk domains previously identified in the empirical literature (Hanson & Morton-Bourgon, 2005; Stinson & Becker, 2013; Stinson et al., 2008, 2016; Thornton, 2002, 2013) and in line with factor analyses of risk factors derived from demographic information (e.g., ever lived in a legal intimate relationship), criminal history (e.g., number and type of offenses), and victim information (e.g., sex, relative or not) of adult men with a history of sexual offenses (e.g., Brouillette-Alarie et al., 2016, 2022). Above and beyond this alignment with previous research on risk domains, our findings can also be used to further develop the SRA model by specifying interrelations among the model’s four domains. More specifically, based on our findings, the following dynamic risk factors may be considered candidates for connecting these domains: general social rejection/ loneliness, lack of concern for others, poor cognitive problem-solving, and impulsive acts. Furthermore, incorporating interrelationships among (domains of) dynamic risk factors could provide forensic workers with more accurate and extensive guidance on the risk assessment and counseling of men with a history of sex offenses. However, it should be emphasized that further empirical validation of these interrelations within the SRA model is needed. Second, based on the centrality analyses, we hypnotized that general social rejection/loneliness, lack of concern for others, poor cognitive problemsolving, and impulsive acts have a stronger influence on other dynamic risk factors
RkJQdWJsaXNoZXIy MjY0ODMw