Saving My Life: Dynamics of Peer and Staff Corrections Among Therapeutic Community Residents

ABSTRACT

Background: Therapeutic communities (TCs) maintain order and encourage behavior change through a system of peer corrections. This study is the first quantitative analysis of the dynamics of the exchange of peer corrections at TCs. Objectives: We applied longitudinal social network analysis to compare the reactions of TC residents to peer versus staff intervention, while analyzing dynamics of correction exchange among residents. Method: The data consisted of a large database of staff and peer affirmations and corrections at four therapeutic community units that occurred between the years 2006 and 2008. We modeled the data as a directed temporal social event network, using a generalized linear mixed effects model to analyze predictors of corrections among residents. Results: Residents were more likely to send a correction following peer affirmations and corrections than following staff affirmations and corrections. Residents reciprocated corrections to individual peers. Autocorrelation was evident in both sending and receiving corrections and residents were more likely to send a correction after having sent an affirmation. Residents who arrived at roughly the same time were more likely to exchange corrections. Residents tended to send and receive more corrections in the middle 3 months of their treatment. European American residents and those with higher scores on the LSI-R were more likely to receive corrections than others. Conclusions: TC residents respond more strongly and more positively to peer than to staff intervention. The pattern of exchange of peer corrections in TCs is complex. This suggests possible paths to improved outcomes.

KEYWORDS:

• Therapeutic communities
• cooperation
• evolutionary psychology
• social network analysis
• mutual aid
• longitudinal networks
• unit atmosphere
• substance abuse treatment
• systems science

Acknowledgments

The authors would like to acknowledge valuable conversations with George De Leon, Ian Hamilton, and Jessica Linley.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the article.

Funding

The authors would like to acknowledge support from NIH grant R21DA023474.

Notes

1 This is a subtly different interpretation of a p-value than typically allowed with frequentist approaches. A frequentist would have to interpret p as the probability of observing the difference we observed assuming that the true difference between the parameters was zero. However, the Bayesian approach we use treats parameters and derived quantities as random variables, the distribution of which can be used to quantify the probability that the parameter sits in a particular range, or the probability that one parameter does not exceed another.