To examine the relationships between 9/11 exposure and subsequent psychopathology, we used a latent dimensional approach. Model fit indices included Comparative Fit Index (CFI), Tucker–Lewis Index (TLI), and Root Mean Square Error of Approximation (RMSEA). Individuals who had never used the relevant substance were coded as zero for each disorder (since wave 1, 7804 individuals (41.7%) had not consumed alcohol; 1632 individuals (87.2%) had not used nicotine, and 17,947 individuals (95.9%) had not used cannabis). Individuals with psychiatric diagnoses prior to 9/11 were excluded from all analyses. Psychiatric and substance use diagnoses, and family history variables were, analyzed as ordered categorical variables (0, 1) implementing these models with a weighted least squares estimator (WLSMV). Missing data were addressed via full-information maximum likelihood in Mplus version 6.12 (Muthén and Muthén, 2000). Estimates were weighted and standard errors adjusted for the complex sample design of the NESARC survey.
