Estimates of intragroup dependence for drug use and skill measures in school-based drug abuse prevention trials: an empirical study of three independent samples.

Abstract

Group-randomized drug abuse prevention trials customarily designate schools as the unit of assignment to experimental condition; however, students within schools remain the unit of observation. Students nested within schools may show some resemblance based on common (peer) selection or school climate factors (i.e., disciplinary practices, group norms, or rules). Appropriate analyses of any treatment effects must be statistically correct for the magnitude of clustering within these intact social units (i.e., intraclass correlation coefficient [ICC]). There is little reported evidence, however, of variation in ICCs that might occur with studies of racially or geographically diverse populations. The purpose of this study was to generate estimates of intragroup dependence for drug use and psychosocial measures (hypothesized mediators) from three separate drug abuse prevention trials. Clustering for the drug use measures averaged .02 across study and age-groups (range = .002 to .053) and was equivalently small for the psychosocial measures (averaging .03 across studies and age-groups; range = .001 to .149). With few exceptions and across different samples, clustering decreased in magnitude over time. Clustering was largest for peer smoking and drinking norms among white, suburban youth and smallest for alcohol expectancies among urban black youth. Findings are discussed with respect to the influence of social climate factors and group norms in the design and analysis of school-based, drug abuse, prevention programs.