The first were questions included only in the postintervention survey. In these cases, reference 4 the responses by the control and the intervention groups were compared in a manner similar to the baseline data. The second included questions for which the categorical responses to a question were not dichotomous. For these, chi-square tests for both the control villages and the intervention villages were presented. Both the pre- and postintervention surveys included questions in which respondents were asked for their positions, for example, ��reasons to stop smoking��: health��yes or no, money��yes or no, and so forth. The percentage of ��yes�� responses for each subvariable of the series was evaluated to compare the differences between the groups at both the pre- and postintervention stages.

The statistical analysis also explored changes in responses from before and after for each group. Finally, the general estimating equation, using a binomial family and logit link, was employed to adjust for variables such as age or gender when looking for differences between the control and intervention groups in relation to the target-dependent variable. The adjusted odds ratios were reported for predictor variables in the model. Stata 12 software was used for statistical analysis. RESULTS Study Population In the intervention villages, 88.9% of the randomly selected households were contacted (i.e., residents were present and agreed to talk to the team); the refusal rate among these households was 1.3%. The contact rate in the control villages was 91.3%, with a refusal rate of 0.9%.

The total number of respondents who participated in the preintervention study was 7,657. Each of the households that participated in the preintervention was approached following the intervention. In this phase of the study, there was a contact rate of 98.1% in the intervention villages, and of those households, a 98.6% participation rate. In the control villages, there was a contact rate of 98.6%, and of those households, a 99.0% participation rate. In some cases, there were new families or new members of the household at these addresses; therefore, the study population includes some individuals who participated in the postintervention but not the preintervention. However, for the purposes of this study, the analysis included only the 5,934 individuals who participated in both surveys for an overall 77.

5% retention rate. This same retention rate was also seen in both the control and intervention subgroups. Preintervention Characteristics and Smoking Behavior The demographic profiles and baseline smoking behaviors (Table 1) of both the control and intervention groups were very similar. There was no statistically significant difference in the median age, gender ratio, education, Drug_discovery employment, and marital status, although there were slightly more married respondents in the control group.