Parental Income and Smoking Participation in Adolescents: Implications of misclassification error in empirical studies of adolescent smoking participation

In adults, the negative relationship between smoking and income is well established. However divergent results have been reported on the impact of parental socioeconomic status on adolescent smoking. In this study we investigate the extent to which misclassification errors in self-reported smoking affects estimates of the impact of parental income on smoking in adolescents aged 11-15 years old. We use the Household Survey for England (HSE) which contains both a self-reported smoking component and an objective measure of smoking obtained through cotinine assays. Smoking participation is modelled using self-reported smoking and cotinine-validated smoking as binary dependent variables in two separate probit models. We compare marginal effects of parental income (and other independent variables) in both models. Our results suggest that self-reported smoking is misreported leading to biased estimates of the effect of parental income on adolescent smoking. Income-related inequality in smoking (the concentration index) is also underestimated when misclassification errors vary across income quintiles.