Table 2 shows the results of the logistic regression analyses that predict smoking, drinking, and the co-occurrence of smoking and drinking. Models 1a-3a were designed to examine whether there was significant family structure variation in adolescent risk behaviors. First, we expected that adolescents who reside with two biological married parents were least likely to smoke, drink, or both. Second, we anticipated that adolescents who live in cohabiting stepfamilies were more likely to smoke, drink, or both than those in any of the three other family structures. Models 1b-3b included maternal socialization and modeling measures as well as indicators of the family economic context to test the extent to which family processes accounted for family structure differences in adolescent risk behaviors. The table reports both the unstandardized coefficients and the odds ratios ((Exp(b)). Odds ratios less than one indicate a lower likelihood than the reference category whereas odds ratios greater than one signify a higher likelihood.
