Collections > Electronic Theses and Dissertations > A Multilevel Study of Schools' Influences on Adolescent Substance Use

Empirical research suggests that school contexts have significant effects on adolescent substance use. The Theory of Health Promoting Schools (HPS), developed in the United Kingdom, explains the influence of school contextual factors on substance use. Two constructs, school value-added and school ethos, have been used in recent European studies to indicate the health promoting quality of schools as it relates to adolescent substance use. I applied the Theory of HPS to a U.S. context and examined relationships between indicators of school value-added and school ethos and student smoking, drinking, heavy drinking, and marijuana use. Data come from Waves 1 and 2 of the National Longitudinal Study of Adolescent Health (Add Health) conducted with students in grades 7-12 (N=12,915 students, 127 schools). I derived and assessed the validity of two new measures of school context suggested by the Theory of HPS. School-value-added had two dimensions-- school achievement added and school truancy added. School ethos had three dimensions-- school disconnectedness, school academic trouble, and institutional disengagement. There was adequate support for construct validity to continue with modeling. I estimated a series of hierarchical generalized linear models. Pseudo intraclass correlation coefficients (ICCs) ranged from 0.06 to 0.12 for the four outcomes. As hypothesized, increases in school disconnectedness were associated with increased odds of student engagement in heavy drinking and marijuana use. Against expectations, increases in school truancy added were associated with decreased odds of student smoking and drinking, and increases in institutional disengagement were associated with decreased odds of student smoking, drinking, and heavy drinking. Also, increases in school academic trouble were associated with decreased odds of student heavy drinking and marijuana use. Cross-level interactions between the school-level variables and their individual analogs were mostly non-significant. In stratified analyses, ICCs for substance use outcomes were higher among high schools than among middle schools, except for smoking. The findings suggest that additional work is needed to develop more valid and reliable measures of the health promoting qualities of schools. However, the patterns of association also suggest that underlying theoretical assumptions about the influence of the school context should be reconsidered.