All statistical analyses were performed using SAS software version 9.1 (SAS Institute Inc., Cary, NC) and accounted for the complex sampling design of HBSC. Because we hypothesized gender differences in the associations between parent-child communication and adolescent substance use, we conducted the analyses separately for males and females. To summarize the characteristics of our study sample, we produced descriptive statistics (frequencies, percentages, mean, and standard errors) including all the demographic, independent and dependent variables of interest. To evaluate potential gender differences in all categorical variables, we conducted a series of chi-square tests. To test our three hypotheses, we conducted six weighted logistic regressions (three for males and three for females), with smoking, drinking and marijuana use as independent outcomes, and father and mother communication as the predicting variables. In these regression models, we controlled for demographic variables including ethnicity/race, socioeconomic status and family structure.
