A Mediation Analysis of Modified Labeling Theory to Examine Potential Mechanisms of Alcohol Stigma in the United States General Population
Method: We analyzed data from 3,142 respondents of the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC) who completed Wave 1 (2001-2002) and Wave 2 (2004-2005) interviews and met criteria for past-year DSM-IV AUD at Wave 2. NESARC Wave 2 included the 12-item alcohol-adapted Perceived Devaluation-Discrimination (PDD) scale to assess PAS (higher scores indicate more stigma). Validated perceived social support, social network involvement, and mental health functioning scales were also included in NESARC (higher scores indicate more social support, social involvement, and higher mental health functioning). We used structural equation modeling to conduct mediation analysis, where PAS was the independent variable, social network involvement and social support were consecutive mediators, and mental health functioning was the dependent variable. The product of coefficients approach was used to assess the total indirect effect of PAS on mental health functioning through the two mediating variables.
Results: Significant direct effects were found for all pathways in the model, in which PAS was inversely associated with social network involvement, social support, and mental health functioning. Social network involvement, social support, and mental health functioning had positive associations with one another. The standardized total effect of stigma on mental health functioning was -0.18 (p<0.001), which represents the direct effect and effects through both mediators. The standardized total indirect effect was -0.09 (p<0.001), providing evidence for mediation through social network involvement and social support.
Conclusions and Implications: In these cross-sectional analyses, the inverse relationship between PAS and mental health functioning was partially explained by the mediators of social network involvement and social support. These findings suggest that the modified labeling theory of mental illness may generalize to populations of persons affected by AUDs. Longitudinal data are needed to further unravel these relationships. To be more theoretically meaningful, the broad measures of social network involvement, like the one used in this study, should be replaced by measures assessing the nature of one’s social relationships (e.g. instrumental support vs. drinking buddies). If stigma reduction interventions were to target affected individuals, behavioral activation including the scheduling of social activities might be indicated. However, stigma is a social phenomenon, so interventions would be better targeted at persons who hold stigmatizing beliefs in order to educate them about the personal experiences and lives of those affected by problems.