Results from the multivariate SEM model predicting adolescent risky drinking and conduct problems are presented in Figure 1. The model demonstrated mediocre fit to the data (χ2 (105) = 720.54; p < .001; CFI = .77; RMSEA = .07). Although the chi-square value was statistically significant (which is common for models with large sample size) and the CFI below typically accepted standards of “acceptable” fit (i.e., .90), the RMSEA was indicative of acceptable model fit (Kline, 2010). We note that relatively low values of the CFI tend to occur in complex models where there are small correlations among variables. Given the small magnitude of several associations in the model (e.g., mothers’ ADS were not related to fathers’ and their own parenting), a somewhat low CFI is not unexpected. Kenny (2014) suggests that since the CFI is an incremental fit index, CFI may not be a useful indicator of model fit in cases where a null model (no associations specified among the study variables) produces RMSEA values < .158; ours was .102. As such, despite the relatively low CFI, the overall model fit statistics suggest acceptable fit.
