Ties That Bind: Homophily and Endogeneity in the Development of Different Types of Social Networks in a Therapeutic Community for Women

Background and Purpose:

Therapeutic communities (TCs) are programs for substance abuse treatment in which peer interactions constitute the primary clinical tool. Therefore, it is important to understand the formation of different relationships between residents. Previous social network studies of TCs indicated that relationships were more likely to form between residents of similar race, age, education, and entry time into the program, a phenomenon known as homophily. However, these studies focused on programmatically defined interactions among male residents. Here we supplement this body of knowledge by studying informal peer friendship networks among women in a corrections-based TC. We then investigate how friendship influence women’s willingness of discussing programmatic issues.

Method:

We conducted a social network survey of 50 women in a high fidelity corrections-based TC over 12 hours. Participants had a mean age of 32.56 (SD=8.83) and 10.93 (SD=1.65) years of education on average. Eighty-eight percent of them were Caucasians, 10% were African Americans, and 2% were Hispanics. Using Exponential Random Graph Models (ERGMs), we tested how homophily of age, race, years of education, case manager, and time of entry into the program influenced tie formation in friendship networks (spending social time, sharing letters from home, and meals with peers). We controlled for reciprocity (mutuality of ties), transitivity (triad formation), and popularity (reception of ties). We then analyzed whether being in the friendship networks increased women’s chance of forming ties in the network of discussing programmatic issues.

Results:

Reciprocity occurred in all networks. Women tended to form triads in all friendship networks. Women with the same case manager were more likely to share social time than those with different case managers (β=0.46, SE=0.19, p=0.02). Women were more likely to spend social time with those who entered the program at different times (β=0.002, SE=0.001, p=0.01). Women tended to avoid focusing on particular popular peers during the social time (β=-0.46, SE=0.16, p=0.004) and sitting at meals with peers (β= -0.27, SE=0.07, p=0.001). Women did not tend to form triads when talking about issues (β= -0.37, SE=0.31, p=0.22). Women with the same case manager had a higher chance of talking about issues (β= -0.52, SE=0.23, p=0.02). However, after accounting for women’s social time and sharing letters networks, the role of case managers became non-significant when discussing issues. Women who spent social time together were more likely to discuss issues with each other (β= -2.25, SE=0.31, p<0.001) as were those who shared letters (β=1.20, SE=0.39, p=0.002).

Conclusions:

Homophily differed by type of interaction. Reciprocity influenced all informal interactions. Women did not focus on popular peers. Women tended to form triads in friendship networks but not when talking about issues. Homophily by case manager influenced the sharing of social time. Women who shared letters or social time were more likely to talk about issues.

Implications:

In this TC, friendship networks were correlated with a willingness to share issues with peers, and homophily by case manager shaped the social time network. Despite external validity limitations, clinicians can use these findings to enhance cooperation in TCs.