Core-Periphery Structure and Program Seniority in Three Social Networks of Therapeutic Community Resident Peer Interactions
Methodology: Fifty female residents of an eighty-nine bed corrections-based TC agreed to track the verbal pushups and pull-ups they gave over the course of twelve hours. They also agreed to track compliments that they gave to peers that did not involve programmatic activities. This yielded three networks, one of pushups, one of pull-ups and one of compliments. Each was analyzed for core-periphery structure using UCINET 6.0 (Borgatti, Everett & Freeman, 2002). If core-periphery structure is present, individuals in the core interact primarily with each other, while individuals in the periphery interact primarily with those in the core. Because of the dependency in the data, a Wilcoxon signed-rank test was used to compare time in program for residents in the core and the periphery of the networks.
Results: In the network of pushups, density in the core is .184 while density in the periphery is .027, density of pushups from the core to the periphery is .053 and density of pushups from the periphery to the core is .036. In the network of pull-ups, density in the core is .286 while density in the periphery is .019, density of pull-ups from core to periphery is .035 and density of pull-ups from periphery to core is .048. In the network of compliments, density in the core is .227 while density in the periphery is .031, density of compliments from core to periphery is .053 while density of compliments from periphery to core is .055. As expected, the Wilcoxon test indicated that time in program is positively related to core status for pushups (p = .0005) and for pull-ups (p = .0349) but not for compliments (p = .4980).
Conclusions and Implications: A core-periphery structure, in which members of the core primarily interact with each other while members of the periphery primarily interact with members of the core, appears in all three networks. Time in program was positively related to core status for programmatic interaction networks but not for the network of informal compliments. Thus, programmatic interpersonal interactions occur disproportionately among a core group of senior residents. This may explain the significance of length of stay as a predictor of TC outcomes (Mandell, Edelin et al, 2008); residents who leave early never have a chance to move into the core. Informal compliments, in which the core-periphery structure is not determined by seniority, may play an important role in encouraging residents to continue treatment.