It was unexpected that we found no direct relationship between PAS and social network involvement, as well as no mediation through this variable. In a prior NESARC study, the relationship between PAS and SNI was weak but significantly associated (Glass et al., 2013). Nonetheless, we chose to analyze a measure of SNI in the current study because we sought to conduct a comprehensive test of MLT, which posits that treatment receipt results in social labeling and a subsequent reduction and/or weakening of social ties owing to stigma (Link et al., 1989). In the current analysis, the specification of a path between the social network and perceived social support variables (which was necessary to test the mediation model) attenuated the association between PAS and SNI. It is possible that one’s perceptions of their social support may be more important than the simple presence of interactions with others. More specifically, the SNI measure does not assess the quality of social relationships. Additionally, we note that the ISS and SNI measures do not tap into aspects of relationships that may be particularly important
