Abstract / Summary
Adolescent cannabis use is associated with family and peer contexts, but whether these associations vary across historical periods is uncertain. We analyzed 12 repeated cross-sectional waves of the Chilean National School Population Survey on Drug Use (2001-2023; 636,008 records). Wave-specific Spearman correlations were estimated using complete cases, with sampling variances and within-wave covariance matrices obtained from 5,000 stratified course-cluster bootstrap replicates. Random-effects meta-analysis and meta-regression used Hartung-Knapp small-sample inference; subgroup interactions used wave-delete jackknife tests. Complete-series unweighted estimates were primary because final expansion weights were unavailable for three waves; supported-wave weighted analyses were sensitivity checks. Caregiver knowledge of whereabouts was inversely associated with cannabis use frequency (pooled ρ = -.200, 95% CI [-.212, -.189]), while perceived regular cannabis use among friends was more strongly and positively associated (pooled ρ = .427, 95% CI [.408, .445]). Neither overall historical trend was significant under small-sample inference (knowledge: β = .0014, p = .065; friends: β = -.0027, p = .109). Weighted sensitivity supported attenuation for knowledge (β = .0027, p = .007). Peer-use trends differed by sex and family structure, whereas knowledge trends did not. These repeated cross-sectional findings indicate strong, persistent associations but limited evidence for a uniform historical change. The subgroup variation and sensitivity dependence emphasize cautious interpretation.