PREDICTORS OF TOBACCO SALES TO MINORS

Background. Our purpose was to determine which characteristics of buye rs, stores, and store clerks predicted successful tobacco sales to min ers. Methods. Seventeen miners visited randomly selected retail outlet s in the Austin, Texas, area and attempted to purchase tobacco product s. We used logistic regression modeling to determine independent predi ctors of successful purchase attempts. Results. Overall, 101 of 165 at tempts (61.2%) were successful. Although legally required, only 25.3% of stores posted warning signs about underage purchase of tobacco, and stores with signs were no less likely to sell to miners than stores w ithout signs. Successful attempts were more common from vendors in met ropolitan locations and from vendors with no alcohol products. Only 8. 1% of attempts succeeded when buyers were questioned (usually about ag e), compared with 95.6% of attempts when no questions were asked (P < 0.001). The best predictor of a successful purchase attempt was failur e to question buyers (adjusted odds ratio = 1,850; P < 0.001). Conclus ions. Warning signs had no effect on vendors' compliance with the stat e miners' access law, and failure to question miners about their age s ubstantially increased the odds of a successful purchase. Laws prohibi ting tobacco sales to miners should be enforced by requiring vendors t o obtain proof of buyers' ages for persons who appear to be <30 years of age.