What Individual Attributes Predict Friendship in a Prison?

Yiwen Zhang

Contact: yuz427@psu.edu

Prison is a unique social institution that incapacitates criminals in a confined space. It is also known as “the society of captives,” as confinement provides an environment for incarcerated individuals to form an array of relationships through their daily interactions. However, we know very little about prisoners’ networks beyond what is portrayed within mass media. Research in this area is scant due to limited access to prisons and their residents. This study, using data from the Therapeutic Community Prison Inmate Network Study (TC-PINS), aims to understand whether individual attributes are predictive of friendship in a prison unit. TC-PINS is a novel data set that surveys prison inmates about their peer relationships. Inmates from within the TC-PINS data have the unique characteristic as all of them are individuals with substance use disorders: two out of every three U.S. prison inmates have substance abuse disorders. The unit in which they reside is a therapeutic community, which is a substance use treatment program that applies group-based cognitive behavioral therapy. This program’s length is four months. Inmates in the therapeutic community are encouraged to interact with their peers through daily activities; they do not have contact with inmates in other units outside this community. The design of this program provides a suitable setting conducive to the study of prison networks and the effect of relational processes on behavioral changes within a defined social boundary. This study uses a sample of the TC-PINS data (n=61) to examine whether and which individual attributes shared between two inmates predict the formation of prison friendship. Individual attributes considered in this study are age, race, education, type of used substance, severity of substance use, severity of offense, and treatment phase. Network methods include centrality measures, subgroup analyses, and logistic regression quadratic assignment procedure (LR-QAP). To visualize subgroups by shared individual attributes, findings of the Girvan-Newman community detection algorithm show that a friendship tie is likely to form between two inmates in the therapeutic community who share the same race or are classified at the same offense severity level. LR-QAP tests reveal that being at the same treatment phase is positively associated with the likelihood of a friendship tie (p<0.001), and that using the same type of substance is negatively associated with friendship formation (p<0.01). The correlations of friendship with race and offense severity level are not statistically significant in LR-QAP results. This study does not find predictive effects of the other individual attributes on prison friendship.

← Schedule