This report is based on the first epidemiological investigation of clusteri
ng of tobacco, alcohol, inhalant, and other drug involvement within individ
ual schools using data from Panama's 1996 National Youth Survey on Alcohol
and Drug Use. Clustering was estimated with the Alternating Logistic Regres
sion method. Adjusted estimates of pair-wise cross-product ratios (PWCPR),
a measure of clustering, show modest clustering (i.e. PWCPR > 1.0) at the s
chool level for tobacco smoking (PWCPR = 1.41; 95% confidence interval, CI
= 1.22-1.64), alcohol consumption (PWCPR = 1.33; 95% CI = 1.22-1.45), use o
f inhalants, (PWCPR = 1.35; 95%CI = 1.07-1.69), and other drug use (PWCPR =
1.38; 95%CI = 1.14-1.68). These findings provide preliminary evidence that
the odds of drug use among school-attending youths increase when another y
outh in the same school uses drugs, and suggest a new line of research on w
ithin-school diffusion that should include the identification of school-lev
el factors that contribute to student drug use. (C) 2000 Elsevier Science I
reland Ltd. All rights reserved.