Contextual Effects of Neighborhood and School Environments on Adolescent Marijuana Use

Introduction: Adolescent marijuana use is common in the U.S. and poses a significant public health problem due to its association with risk behaviors, health problems, and lower lifetime earnings.  Examining the determinants of adolescent marijuana use can inform strategies for primary prevention and targeted interventions. The purpose of this study was to estimate the level of variation in marijuana use across schools and neighborhoods simultaneously, and to examine the relative influence of school and neighborhood socio-economic status (SES) on marijuana use, among a national sample of adolescents. While prior work has generally focused on either school or neighborhood, this analysis examines both contexts simultaneously in addition to individual factors.

Methods: Data on adolescent marijuana use come from Wave I (collected in 1994-1995) of the National Longitudinal Study of Adolescent Health (Add Health) when respondents were in grades 7-12. We used a cross-classified multi-level approach to examine fixed and random effects at the individual, school and neighborhood levels on past 30-day marijuana use. We examined the effects of race/ethnicity, receipt of public assistance, and parental education at the individual, school, and neighborhood levels. At the neighborhood level, we also included percentage of households with female head of household and percentage of owner-occupied housing. Cross-classified multi-level logistic regression models were adjusted for age and gender, and were implemented in MLwiN (v2.29; Birmingham, UK) through STATA (v13.1; College Station, TX).

Results: A total of 18,329 adolescents who attended 128 schools and lived in 2,255 neighborhoods (defined by Census tract) were included in this analysis. Average age in the sample was 15.6 years (SD 1.7) with 51% of participants reporting race as non-Hispanic White. Fourteen percent of adolescents reported past 30-day marijuana use.  In cross-classified multi-level models adjusting for individual, school- and neighborhood-level factors, male gender (OR: 1.33; 95% CI: 1.22, 1.45), older age (OR: 1.19; 95% CI: 1.16, 1.24), and having a parent on public assistance (OR: 1.35; 95% CI: 1.17, 1.55) were risk factors for marijuana use. Racial differences were also observed with Multiracial (OR: 1.25; 95% CI: 1.02, 1.52) and Other race (OR: 2.31; 95% CI: 1.40, 3.67) participants more likely and Black (OR: 0.79; 95% CI: 0.67, 0.92) and Hispanic (OR: 0.40; 95% CI: 0.31, 0.50) participants less likely to use marijuana than Whites.  No school or neighborhood level fixed effects were significantly associated with marijuana use.  However, random effects for both school and neighborhood were statistically significant, indicating residual variation across each context even after adjustment for fixed effects (σ²_school=0.28, σ²_neighborhood=0.07).

Conclusion: There is significant variation in the percentage of students reporting marijuana use across both school and neighborhood contexts. However, our results suggest that individual factors rather than school- or neighborhood-level demographic or SES factors are most strongly associated with adolescent marijuana use. Explanations for the importance of variability observed across settings will be discussed.