This paper examines whether a national anti-drug advertising campaign is associated with a change in adolescents' drug use behavior. Specifically, the primary objective of the paper is to investigate the relationship between adolescents' recall of exposure to anti-drug advertising and their probability of trying marijuana, cocaine and crack, as well as their volume of drug use given trial. We use a version of an attribute-based discrete choice framework based upon work by Hanemann (1984) to generate a model of drug use behavior, and estimate the model using a modified sequential probit framework. We estimate the model using data from the Partnership for a Drug-Free America (PDFA) Partnership Attitude Tracking Survey (PATS) over the years 1987 through 1990. Since PDFA advertising did not begin until after the first set of data were collected, we had a form of "natural experiment" where we could first address the factors influencing drug use on a control group (respondents interviewed prior to the commencement of PDFA advertising). We were then able to test hypotheses about the effectiveness of anti-drug advertising over the first three years of the program (1988-1990). The results support the hypothesis that anti-drug advertising lowers the probability of both marijuana and cocaine/crack trial by adolescents. Specifically, the results suggest that the cumulative impact of anti-drug advertising was to lower the probability of marijuana trial by 9.25% and cocaine trial by 3.6% at the individual level in year three of the program (1990). However, our results suggest that recall of anti-drug advertising did not generally affect adolescents' decisions regarding how much marijuana or cocaine/crack to use for those already using these drugs. Thus, although we find strong evidence suggesting that anti-drug advertising is associated with decreased drug trial, our results suggest that its impact on the volume of consumption of harder drugs by existing drug users is minimal.
