The analyses lacked causal ordering for PAS, the mediators, and outcomes. Without longitudinal data, reverse causality is an alternative explanation for the mediated pathway, so that we cannot rule out that low levels of social support lead to increased PAS, and both are associated with AUD and internalizing disorders. However, longitudinal studies of non-substance-related psychiatric stigma have not revealed a consistent direction in the stigma-social support relationship (Mueller et al., 2006; Perlick, 2001; Roeloffs et al., 2003). Our study may have also been limited by the use of a general measure of SNI, which does not make distinctions between out-ofhome and in-home social involvement (Link et al., 1989; Perlick, 2001), does not probe whether absence of social interaction was due to stigma, does not determine whether individuals avoided contact with their social network or vice versa, and importantly does not characterize support in terms of its alcohol-positive or - negative nature. In addition, it is possible that variables related to stigma, social support, and AUD were excluded from the analysis, which could result in bias. Finally, PAS and social support
