Last, we note that the subtypes of psychiatric comorbidity derived in our LCA measurement model support several prior studies that used researcher-specified subgroups or latent factors defined by internalizing and externalizing comorbidity (Dawson et al. 2010; Keyes et al. 2011; Hasin and Kilcoyne 2012). The evidence of subgroup differences in perceived alcohol stigma in this study, and alcohol severity or trauma history in other studies (Dawson et al. 2010; Keyes et al. 2011), provides support in alcohol research for the use of four broad categories of internalizing, externalizing, both internalizing and externalizing, and no comorbidity. We also note that in comparison with other LCA studies, the number and specific nature of latent psychiatric comorbidity classes varies across reports, though the broad subtypes of internalizing, externalizing, high-comorbidity, and/or unaffected classes have been found in addition to other more distinctive subtypes (Kessler et al. 2005; Vaidyanathan et al. 2011; Weich et al. 2011; McCutcheon et al. 2013). Perhaps the difference in the number of classes identified between the current study and the prior studies is due to the fact that our analysis uniquely only included those with DSM-5 AUD rather than focusing on the entire general population.
