Sample size estimation

The sample size of this study was estimated using the Rosner’ formula (n = [zα√pq(1 + 1/k) + zβ √p1q1 + p2q2/k]2/Δ2) [33]. In order to achieve 80 % power to detect a group difference of 13 % [34] with a two-sided 5 % significance level, 165 female students in each arm were required. On the basis of a predicted attrition rate of 20 %, the goal was to randomly assign 198 female students in each intervention and control groups. Of those who initially agreed to participate in the study, 26 dropped out for variety of reasons (e.g. medical illness, unwillingness to participate, moving and schedule conflict). As a result, 186 female students in the intervention group and 184 female students in the control group completed the study in 12 months (Fig. 2).