Table 3

Odds ratio of smoking* in the intervention group compared with the control group at the first follow up measurement, adjusted for smoking at baseline; obtained by multilevel logistic regression (n=1669)

Unadjusted OR95% CIAdjusted OR†95% CI
*Smoking is defined as all students who experiment with smoking or who smoke daily or weekly.
†Adjusted for the following factors: ethnicity, age, religion, and gender (all three at the class and individual levels). At school level, the size of the school was included. OR, odds ratio; CI, confidence intervals.
‡In multilevel logistic regression, random variance on the school and class levels represents the relation of the explained variance between the two levels. These variances cannot be compared with the random variance at the individual level due to the error estimate included in the latter variance.
Control group1.001.00
Intervention group0.600.40 to 0.910.620.43 to 0.90
Random variance‡ Variance95% CIVariance95% CI
School level0.109−0.06 to 0.280.028−0.09 to 0.14
Class level0.131−0.08 to 0.340.140−0.07 to 0.35
Individual level0.9590.89 to 1.030.9750.91 to 1.04