4 97 REPLICATION AND COMPARISON NETWORKS 4.3.1.3 Networks including violent recidivism (including sexual contact) Again, all interrelationships between dynamic risk factors were positive in networks with violent recidivism (including sexual contact) in both independent samples (Figure 4.3a and 4.3c). General social rejection/loneliness, lack of concern for others, impulsive acts, and poor cognitive problem-solving were among the four dynamic risk factors with the highest estimated strength centralities. In contrast, capacity for relationship stability and significant social influences were relatively weak connected with other factors in both the DSP and BC networks (Figure 4.3c and 4.3d). In both the original (DSP) and the replication (BC) sample, the spinglass and the walktrap algorithms revealed a distinct community consisting of sexual preoccupation, sex as coping, deviant sexual interests, and emotional identification with children in the networks with violent recidivism (including sexual contact). The spinglass algorithm showed for the DSP sample a second community consisting of general social rejection/ loneliness, capacity for relationship stability, and negative emotionality/hostility. Supplemented with third community containing significant social influences, impulsive acts, poor cognitive problem-solving, and violent recidivism (including sexual contact). And, completed with and a fourth community formed by hostility toward women, lack of concern for others, and cooperation with supervision. Spinglass community analyses of the BC sample revealed a second community of general social rejection/loneliness and capacity for relationship stability. A third community formed by negative emotionality/ hostility, hostility toward women, lack of concern for others, cooperation with supervision, and significant social influences. And, a fourth community consisted of impulsive acts, poor cognitive problem-solving, and violent recidivism (including sexual contact). The walktrap community network estimated on the DSP sample showed two additional communities, one containing general social rejection/loneliness and capacity for relationship stability and the other consisting of the remaining eight dynamic risk factors. For the BC sample the walktrap community analysis showed a second community of the remaining 10 dynamic risk factors. Supplemental Material B provides additional results of the community analyses, including graphs, ranges, means, and modes of the communities. 4.3.2 STATISTICAL NETWORK COMPARISON To statistically compare the three networks based on the data of the original (DSP) and replication (BC) sample, global strength (sum of all the edges in the network), and network structure (the maximum difference in edge weights in the network) were compared using the NCT. In addition, the similarity of the network structures was also assessed by correlating the adjacency matrices of the networks and by correlating the rank of node’s strength centrality across networks (see Table 4.3).
RkJQdWJsaXNoZXIy MjY0ODMw