Multivariate Prediction of Cigarette Smoking among Children in Grades Six, Seven and Eight

Abstract

Cigarette smoking among the young has not declined in a manner similar to that in adults and is a cause for concern. While several studies have been carried out that demonstrate short-term effects of educational programs on the prevalence of youthful cigarette smoking, no longitudinal studies have been conducted with the goal of identifying characteristics associated with high-risk for smoking. The purpose of this study was to relate variables measured in 1975 to smoking status in 1978 using multivariate statistical procedures. Sociodemographic characteristics, smoking environment, school performance, attitude toward smoking and knowledge about the effects of smoking were included. Results were available for 1,082 children who were tested in 1975 in grades four through six and again in 1978 when they were in grades six through eight. Cross-sectional analysis of variables measured in 1978 revealed that six variables accounted for 37 per cent of the variance in smoking status using a multiple regression model. Five variables that were measured in 1975 in the longitudinal analysis accounted for 32 per cent of the variance in 1978 smoking status. The utility of these results for prediction and for development of educational programs is discussed.