An ideal data source to test an application of MLT would provide population-based data on labeled and unlabeled persons with AUDs, and the requisite measures of PAS, social network variables, and psychiatric outcomes (see Fig. 1). Such data are available within the National Epidemiologic Survey of Alcohol and Related Conditions (NESARC), a representative survey of the United States general population. While NESARC provides longitudinal assessments, several key variables (e.g., PAS) were only assessed at the follow-up interview, which precluded our ability to establish temporal precedence. Hence, we conducted cross-sectional analyses of NESARC using structural equation modeling (SEM) to test two basic hypotheses of MLT: (1) the relationship of PAS with past-year AUD and past-year internalizing psychiatric disorders would be positive and mediated by individuals’ social network involvement and perceived social support, and (2) the mediated relationship would exist for labeled, but not unlabeled persons. While MLT assumes perceived stigma is inconsequential for unlabeled individuals, we deemed it worthwhile to evaluate this empirically for alcohol stigma.
