In this study we use a large nationally representative sample of adolescents at the ages of 12–17 years, drawn from multiple years of a repeated cross-sectional survey, to address these limitations.