Ith grade. No systematic associations were observed between agentic goals and alcohol use (6th grade: r=.02, 7th grade: r=.17, 8th grade: r=.04, 9th grade: r=.11) and the strength of the association between communal goals and alcohol use decreased with grade (6th grade: r=.22, 7th grade: r=.13, 8th grade: r=.04, 9th grade: r=.-.03).Alcohol Clin Exp Res. MG-132 web Author manuscript; available in PMC 2016 December 01.Meisel and ColderPageMultilevel ModelsAuthor Manuscript Author Manuscript Author Manuscript Author ManuscriptThe gender interaction terms did not significantly improve model fit (2 [8, N=386]=5.16, p>.05), and were not considered further. However, the first-order effect of gender was included as a statistical control variable in models testing grade interaction terms. A nested chi-square test comparing a model with and without the hypothesized interaction terms with grade suggested that model fit improved with the inclusion of twoway (2 [8, N=386]=18.25, p<.05) and three-way (2 [4, N=386]=11.21, p<.05) interactions. As shown in Table 1, significant three-way interaction terms were found for grade ?descriptive norm ?communal goals (B =-0.33, p=.03), grade ?injunctive norms ?communal goals (B =0.30, p=.03), and grade ?descriptive norms ?agentic goals (B=0.24, p=.04). The grade ?injunctive norms ?agentic goals three-way interaction term was not statistically significant (B =-0.15, p=.30). To facilitate interpretation of the three-way interaction terms, simple slopes of norms by Aprotinin site levels of social goals were plotted for an early (6th variables predicting 7th grade alcohol use) and late (9th grade variables predicting 10 grade alcohol use) cross-lag (see Figure 1). Descriptive Norms Descriptive Norms and Agentic Goals As seen in Panel A of Figure 1, for adolescents in the 6th grade, descriptive norms were not found to significantly predict 7th grade alcohol use for adolescents with high or low levels of agentic goals (OR=0.86 and 1.71, respectively, both ps>.05). High levels of descriptive norms in the 9th grade were associated with increased probability of alcohol use in the 10th grade for adolescents with high (OR=2.43 p<.05), but not low (OR=1.09, p>.05) levels of agentic goals. This pattern provides partial support for the hypothesized interaction between descriptive norms, agentic goals and grade. That is, there was a shift in the moderating role of agentic social goals with grade, such that descriptive norms became a predictor of alcohol use for youth characterized by strong agentic goals, but only in later grades. Descriptive Norms and Communal Goals High levels of descriptive norms in the 6th grade were associated with increased probability of alcohol use in the 7th grade for adolescents characterized by high (OR=2.07, p<.05) but not low (OR=0.72, p>.05) levels of communal goals. As seen in Panel 2 of Figure 1, in later grades, this pattern reversed itself, such that 9th grade descriptive norms were not associated with 10th grade drinking for adolescents high in communal goals (OR=0.72, p>.05), but they were associated with 10th grade drinking for adolescents low in communal goals (OR=2.58, p>.05). Although descriptive norms were not hypothesized to interact with communal goals, these findings suggest a developmental shift such that in early adolescence, descriptive norms influence alcohol use for those characterized by strong communal goals whereas in later adolescence descriptive norms influence alcohol use for adolescents character.Ith grade. No systematic associations were observed between agentic goals and alcohol use (6th grade: r=.02, 7th grade: r=.17, 8th grade: r=.04, 9th grade: r=.11) and the strength of the association between communal goals and alcohol use decreased with grade (6th grade: r=.22, 7th grade: r=.13, 8th grade: r=.04, 9th grade: r=.-.03).Alcohol Clin Exp Res. Author manuscript; available in PMC 2016 December 01.Meisel and ColderPageMultilevel ModelsAuthor Manuscript Author Manuscript Author Manuscript Author ManuscriptThe gender interaction terms did not significantly improve model fit (2 [8, N=386]=5.16, p>.05), and were not considered further. However, the first-order effect of gender was included as a statistical control variable in models testing grade interaction terms. A nested chi-square test comparing a model with and without the hypothesized interaction terms with grade suggested that model fit improved with the inclusion of twoway (2 [8, N=386]=18.25, p<.05) and three-way (2 [4, N=386]=11.21, p<.05) interactions. As shown in Table 1, significant three-way interaction terms were found for grade ?descriptive norm ?communal goals (B =-0.33, p=.03), grade ?injunctive norms ?communal goals (B =0.30, p=.03), and grade ?descriptive norms ?agentic goals (B=0.24, p=.04). The grade ?injunctive norms ?agentic goals three-way interaction term was not statistically significant (B =-0.15, p=.30). To facilitate interpretation of the three-way interaction terms, simple slopes of norms by levels of social goals were plotted for an early (6th variables predicting 7th grade alcohol use) and late (9th grade variables predicting 10 grade alcohol use) cross-lag (see Figure 1). Descriptive Norms Descriptive Norms and Agentic Goals As seen in Panel A of Figure 1, for adolescents in the 6th grade, descriptive norms were not found to significantly predict 7th grade alcohol use for adolescents with high or low levels of agentic goals (OR=0.86 and 1.71, respectively, both ps>.05). High levels of descriptive norms in the 9th grade were associated with increased probability of alcohol use in the 10th grade for adolescents with high (OR=2.43 p<.05), but not low (OR=1.09, p>.05) levels of agentic goals. This pattern provides partial support for the hypothesized interaction between descriptive norms, agentic goals and grade. That is, there was a shift in the moderating role of agentic social goals with grade, such that descriptive norms became a predictor of alcohol use for youth characterized by strong agentic goals, but only in later grades. Descriptive Norms and Communal Goals High levels of descriptive norms in the 6th grade were associated with increased probability of alcohol use in the 7th grade for adolescents characterized by high (OR=2.07, p<.05) but not low (OR=0.72, p>.05) levels of communal goals. As seen in Panel 2 of Figure 1, in later grades, this pattern reversed itself, such that 9th grade descriptive norms were not associated with 10th grade drinking for adolescents high in communal goals (OR=0.72, p>.05), but they were associated with 10th grade drinking for adolescents low in communal goals (OR=2.58, p>.05). Although descriptive norms were not hypothesized to interact with communal goals, these findings suggest a developmental shift such that in early adolescence, descriptive norms influence alcohol use for those characterized by strong communal goals whereas in later adolescence descriptive norms influence alcohol use for adolescents character.
