The effects of family stressors on substance use initiation in adolescence

Smoking and drinking are critical problems in adolescence that have long-term adverse impacts on health and socio-economic factors. We examine the extent to which family stresses influence the timing of initiation of smoking and drinking. Using national panel data from the National Educational Longitudinal Study (NELS) we capitalize on the survey design and use school-level fixed effects that control for the local environments, including prices of cigarettes and alcohol. In addition, we narrow our control group to classmates who will experience a similar stressor in the future. We find that a composite measure of family stressors when young increases the likelihood of initiating tobacco and alcohol use, with much of the impact attributable to parental divorce. In our baseline estimates, the composite stress measure is associated with a 30% increase in the likelihood of smoking and a 20% increase in drinking. When we control for multiple sources of confounding, the impact shrinks and remains significant for smoking but not for drinking. We conclude that studies which do not control for confounding are likely to significantly overestimate the impact of family stress on substance use. Our approach helps to move the literature forward by separating causal results from spurious associations.