4 100 CHAPTER 4 4.4 DISCUSSION The objective of this study was to better understand the interrelationships among dynamic risk factors predicting future sexual reoffending of adult men with a history of sexual offenses. We conducted a study to examine to what degree the findings of our original network analysis (van den Berg et al., 2020) could be replicated using a larger and independent sample. Our hypothesis that the networks estimated using the data from two independent samples would be similar could not be rejected. Although, statistical network comparison by the NCT demonstrates that networks based on the original (DSP) and replication (BC) data sets without recidivism, with sexual recidivism, and with violent recidivism (including sexual contact) revealed differences in network structures, there were no meaningful differences in global strength. Further, both the correlation of the adjacency matrices of the networks and the correlation of the rank of the node’s strength centrality estimates across networks were strong. Besides, similar communities of dynamic risk factors were found in networks estimated using data from both independent samples. 4.4.1 LIMITATIONS Some limitations of this study should be acknowledged. First, the unequal sizes of the original DSP and the replication BC samples might have affected our statistical network comparison (van Borkulo et al., 2022). However, given the robustness of the estimated networks and the similarity in variance of total score of the STABLE-2007, a bias caused by the difference in sample sizes is unlikely. Second, although the STABLE-2007 is empirically and theoretically driven and previous studies revealed no differences in predictive propensities between North America and several European countries (Brankley et al., 2021), the STABLE-2007 captures a limited number of dynamic risk factors. Adding an additional dynamic risk factor to a network analysis may impact the network construction and strength centrality of all dynamic risk factors. Third, although all individual dynamic risk factors were related to recidivism, each risk factor was assessed using a single item. It would be valuable to evaluate the impact on network construction when dynamic risk factors are, or could be, based on multiple items/scale scores. Fourth, our findings are based on two North American routine correctional samples of adult males charged or convicted for a sexual offense. Although both samples could generally be considered a reflection of the use of the STABLE-2007 by correctional services, generalization to other (sub)groups of offenders is not possible (e.g., women, juveniles, men from different cultures or cultural contexts). This could in particular apply to men who have not entered the criminal justice system, as social isolation - operationalized as general social rejection/loneliness by the STABLE-2007 and found to have a high-strength centrality in the present study - has been found to be less predictive of sexual violence in non-incarcerated samples (Malamuth et al., 1991).
RkJQdWJsaXNoZXIy MjY0ODMw