Finally, a series of multiple group SEMs was conducted to examine possible gender moderation of the longitudinal relationships under investigation. First, the SEM depicted in Figure 1 was estimated as an unconstrained model across gender groups; only the factor loadings of the latent variables were constrained to equality to establish consistent measurement for girls and boys. Next, all path coefficients among the measures of alcohol use, adverse consequences, subjective well-being, and alcohol problems were constrained to equality across the two groups. Results from a chi-square difference test using the difftest option in Mplus indicated that these constraints did not result in a statistically significant decrease in model fit when comparing the unconstrained and constrained models, χ2 (4, N = 109,99) = 2.67, p = .62. This suggests that the longitudinal relationships were similar for boys and girls.
