Abstract
Previous studies found lower substance use in schools achieving better examination and truancy results than expected, given their pupil populations (high value-added schools). This study examines whether these findings are replicated in West Scotland and whether school ethos indicators focussing on pupils' perceptions of schooling (environment, involvement, engagement and teacher-pupil relations) mediate the associations. Teenagers from forty-one schools (S2, aged 13, n = 2268; S4, aged 15, n = 2096) previously surveyed in primary school (aged 11, n = 2482) were surveyed in the late 1990s. School value-added scores were derived from standardised residuals of two regression equations separately predicting from pupils' socio-demographic characteristics (1) proportions of pupils passing five Scottish Standard Grade Examinations, and (2) half-day truancy loss. Outcomes were current smoking, monthly drinking, ever illicit drug use. Random effects logistic regression models adjusted for potential pupil-level confounders were used to assess (1) associations between substance use and school-level value-added scores and (2) whether these associations were mediated by pupils' perceptions of schooling or other school-level factors (school roll, religious denomination and mean aggregated school-level ethos scores). Against expectations, value-added education was positively associated with smoking (Odds Ratios [95% confidence intervals] for one standard deviation increase in value-added scores were 1.28 [1.02-1.61] in S2 and 1.13 [1.00-1.27] in S4) and positively but weakly and non-significantly associated with drinking and drug use. Engagement and positive teacher-pupil relations were strongly and negatively associated with all substance use outcomes at both ages. Other school-level factors appeared weakly and largely non-significantly related to substance use. Value-added scores were unrelated to school ethos measures and no ethos measure mediated associations between value-added education and substance use. We conclude that substance use in Scotland is more likely in high value-added schools, among disengaged students and those with poorer student-teacher relationships. Understanding the underpinning mechanisms is a potentially important public health concern.