Several sensitivity analyses were warranted. Because several items in the Perceived Devaluation-Discrimination scale ask about perceived stigmatization of people who have previously obtained alcohol treatment (see Supplement S2), and because 18% of the sample had obtained treatment in their lifetime, we examined whether including prior alcohol treatment as a covariate during latent class estimation altered the taxonomy of the comorbidity subtypes. The comorbidity classes remained unchanged. Next, we evaluated adjusted mean differences in perceived alcohol stigma across the latent classes when modeling stigma as a latent factor because our prior work suggested that using a latent factor may reduce the scale’s measurement error (Glass et al., 2013a). In this sensitivity analysis we also adjusted for prior alcohol treatment when comparing stigma means cross the latent classes. Patterns of statistical significance for the adjusted mean differences were consistent with our primary analysis. Sensitivity analysis results are included in Supplement S3.
