Predicting Smoking Behaviors among Junior High School Students in Ghana

Despite the rising rate of smoking in sub-Sahara African countries, measures to control the tobacco epidemic have been limited to developed countries. The purpose of the present study was to recommend predictive models for determining predictors of smoking tendencies among junior high school students in Ghana. The 2009 Global Youth Tobacco Survey (GYTS) served as the data source. The GYTS is a school-based survey designed to enhance the ability of countries to monitor tobacco use among youth and to guide the implementation and evaluation of tobacco control and prevention programs. Logit model and forward selection were used to choose predictive variables for smoking tendencies and behaviors. Receiver Operating Characteristic (ROC) curve, Area under the curve (AUC) and C-Index were validation tools used to assess the predictive power of recommended models. Results showed promising potential for different predictive models: where students smoked, having friends who smoked, having people smoke in their presence, chewing tobacco products, and a student’s sex significantly predicted their smoking tendencies.