The four Baron and Kenny criteria evaluated were as follows: Baseline depression group must be significantly AG-014699 associated with the dependent variable. To test Criterion 1, susceptibility to initiate smoking at 18 months was regressed onto baseline depression group in a model controlling for treatment group, gender, ethnicity, and baseline susceptibility to initiate smoking. Results indicated that a higher proportion of students with CES-DC scores ��16 were susceptible to initiate smoking at 18 months: F(1, 962) = 4.50, p = .03; odds ratio (OR) = 1.4, 95% CI = 1.03�C1.89. Baseline depressive symptoms must be significantly associated with the hypothesized mediator. To test Criterion 2, self-efficacy at 18 months was regressed onto baseline depression group in a model controlling for baseline self-efficacy, treatment group, gender, and ethnicity.

Results indicated that those in the low-depression group (CES-DC < 16) had higher self-efficacy scores: F(1, 973) =10.68, p = .001 (see Table 2 for means and SDs). Table 2. Demographic Characteristics by Baseline Depression Group The hypothesized mediator must be significantly associated with the dependent variable. To test this criterion, the susceptibility to initiate smoking at 18 months was regressed onto self-efficacy scores at 18 months in a model controlling for baseline self-efficacy, baseline susceptibility to initiate smoking, treatment group, gender, and ethnicity. Indeed, those who were susceptible to initiate smoking at 18 months had lower self-efficacy scores: F(1, 962) = 25.83, p < .0001; OR = 1.04, 95% CI = 1.

02�C1.05. Those who were susceptible to initiate smoking had mean self-efficacy scores of 39.8 (SD = 11.4), while those who were not susceptible had mean self-efficacy scores of 44.7 (SD = 11.7). The effect of baseline depressive group on the dependent variable must be meaningfully reduced when including a hypothesized mediator in the model. To test Criterion 4, susceptibility to initiate smoking at 18 months was regressed onto baseline depression group in a model controlling for baseline self-efficacy, baseline susceptibility to initiate smoking, treatment group, gender, ethnicity, and self-efficacy at 18 months, the mediator variable. The association between baseline depression group and the outcome must be meaningfully reduced in the presence of the mediator.

When baseline depression group and self-efficacy at 18 months were both added to the model predicting susceptibility to initiate smoking at 18 months, baseline depression group no longer significantly predicted susceptibility : F(1, 959) = 3.17, p = .075; OR = 1.3, 95% CI = 0.097�C1.82. However, self-efficacy at 18 months continued to predict susceptibility at 18 months: AV-951 F(1, 959) = 24.37, p < .0001; OR = 1.04, 95% CI = 1.02�C1.05 (see Table 3). Table 3.