Applying multilevel analytic strategies in adolescent substance use prevention research

Raymond F. Palmer, John W. Graham, Erica Lynne White, William B. Hansen

Research output: Contribution to journalArticlepeer-review

41 Scopus citations


Background. School-based drug prevention programs have been criticized on methodologic grounds because the unit of analysis is often not the unit of randomization, thus increasing the likelihood of Type I errors. Application of multilevel analytic strategies appropriately corrects this biasing tendency. This study demonstrates the practical use of such analysis. Methods. Data from 2,370 seventh-grade students participating in a substance use prevention trial were analyzed using a multilevel strategy. We examined the effectiveness of a social pressure resistance training and a normative education (NORM) intervention against an information-only control group. Results. The NORM condition revealed 1-year program effects for cigarette and marijuana use with individuals as the unit of analysis and only marginal effects with classroom as the unit of analysis. No program effects were found using school as the analysis unit. A multilevel strategy revealed program effects for cigarettes and marijuana with both class and school as grouping levels. The effect for alcohol use was significant at the 2-year follow-up. Conclusions. Interventions establishing conservative drug use norms in classrooms may be an effective strategy in reducing substance use onset among adolescents. Utilization of appropriate analytic strategies is important in the analysis and interpretation of data containing nested structures.

Original languageEnglish (US)
Pages (from-to)328-336
Number of pages9
JournalPreventive Medicine
Issue number3
StatePublished - May 1998

All Science Journal Classification (ASJC) codes

  • Epidemiology
  • Public Health, Environmental and Occupational Health


Dive into the research topics of 'Applying multilevel analytic strategies in adolescent substance use prevention research'. Together they form a unique fingerprint.

Cite this