The use of tobacco among school children in ten Brazilian cities, by means of statistical techniques such as bivariate and multivariate analysis (Logit Model), was analyzed. Bivariate analysis revealed a positive association between the use of tobacco and the school child's non-academic occupation, evening course attendance, poor school performance, and parents' smoking habits. Multivariate techniques and the Logit Model showed that poor school performance (on the evening course) and parents' smoking habits (on the day-time course) were most significant variables. The importance of the utilization of multivariate technics for the real understanding of the various factors involved in complex behavior patterns such as the use of tobacco, alcohol and drugs, was discussed.
Smoking; Multivariate analysis; Risk