The SEM depicted in Figure 1 was estimated. Although not illustrated, all variables (excluding latent variable indicators) were regressed on the 5 covariates, which themselves were allowed to covary. Prior to the SEM, a confirmatory factor analysis (CFA) was conducted, in which the measurement model and all covariances among the 11 study constructs were estimated. Note that unstandardized factor loadings for the same indicator at the two waves were constrained to equality to ensure consistent measurement over time. In addition, per standard practice, covariances between the residuals of the same indicator over the two measurement occasions were estimated. According to current guidelines (e.g., a CFI of close to .95 or greater and an RMSEA of .06 or less; Hu & Bentler, 1999), the fit between the data and the CFA was acceptable, χ2 (5, N = 208) = 6.09, p = .30, CFI = .99, RMSEA = .03. Standardized factor loadings ranged from .66 to .99. Estimated correlations among the constructs are reported in Table 2. In general, measures of alcohol use, adverse consequences, and problem alcohol use had statistically significant positive associations with one another, and statistically significant negative associations with subjective well-being.
