The model demonstrated excellent fit, χ2(df = 72) = 190.01, p < 0.001; CFI = 0.98; TLI = 0.95; RMSEA = 0.04. An earlier AO (p < 0.001), shorter delay (p < 0.001), positive family history (p = 0.02), and higher impulsivity (trend p = 0.06) were associated with more problems during senior year of college. Consistent with the heavy drinking model, male gender (p = 0.01), a later AO (p < 0.001), and/or a longer delay (p < 0.001) was associated with steeper increases in problems over time. Figure 2 depicts the model with standardized regression coefficients (significant paths in bold). Also mirroring the heavy drinking model, the indirect effects of AO operating through delay were in the opposite direction of the main effects for the intercept (β = 0.09, p < 0.001) and slope (β = −0.10, p < 0.001). The net effects of AO were significant (β = −0.14 for intercept; β = 0.22, p < 0.001 for slope), but considerably smaller than the direct effects (β = −0.23 for intercept; β = 0.32 for slope; Fig. 3).
