We conducted multigroup analyses to examine whether the patterns of associations differed across sex and age groups by removing sex from the model and then comparing a model with the coefficients of interest constrained to equality with another model that had all coefficients freely estimated across females and females. Multigroup analyses across age groups were conducted by removing age from the model and then comparing a model with the coefficients of interest constrained to equality with another model that had all coefficients freely estimated across age groups (emerging adulthood vs young adulthood vs middle adulthood). A statistically significant Wald chi-square test of parameter equalities would indicate significant differences in regression coefficients across groups. Full information maximum likelihood estimation method was used to account for missing data and clustering within families was accounted using the CLUSTER command in Mplus. To account for the fact that we tested the associations independently in two ancestral groups, we used Bonferroni corrected p < .025 to evaluate statistical significance.
