Those individuals who could be located were sent a letter and a consent form requesting their participation in a follow-up to their earlier participation in the DARE evaluation. Those individuals who returned the signed consent form were mailed a questionnaire that took approximately 30 to 45 min to complete. Of the available sample, 5 had died, 176 refused to participate, 83 could not be located, and 163 were contacted but did not return the survey. For their time and effort, participants were paid $15 to $50.

Similar to the earlier data collection, participants were asked questions about their use of alcohol, tobacco, marijuana, and other illegal drugs. For each drug category, participants were asked to report how often they had used the substance in their lifetime, during the past year, and during the past month. In addition, participants were asked a variety of questions concerning their expectancies about drug use. For each drug, respondents reported how likely they believed using that drug would lead to five negative consequences (e.g., "get in trouble with the law" and "do poorly at school or work") as well as how likely they believed using that drug would lead to eight positive consequences (e.g., "feel good" and "get away from problems"). Negative and positive expectancy scores were formed for each drug at each age. Two potential mediators of the DARE intervention, peer-pressure resistance and self-esteem, were also assessed. Participants responded to nine items designed to assess the ability to resist negative peer pressure (e.g., "If one of your best friends is skipping class or calling in sick to work, would you skip too?"). Finally, participants responded to the 10-item Rosenberg Self-Esteem Scale ( Rosenberg, 1965). All scale scores had acceptable reliabilities (alphas ranged from .73 to .93, with an average of .84).

A complete description of the experimental and comparison interventions is contained in the Clayton et al. (1996) study. Twenty-three elementary schools were randomly assigned to receive the DARE intervention, whereas the remaining 8 schools received a standard drug-education curriculum. The DARE intervention was delivered by police officers in 1-hr sessions over 17 weeks. The focus of the DARE curriculum is on teaching students the skills needed to recognize and resist social pressures to use drugs. Additionally, the curriculum focuses on providing information about drugs, teaching decision-making skills, building self-esteem, and choosing healthy alternatives to drug use. The control condition was not a strict no-treatment condition but instead consisted of whatever the health teachers decided to cover concerning drug education in their classes. The drug education received by students in the control condition cannot be described in detail because of the considerable latitude on the part of teachers and schools in what was taught. Nonetheless, in many instances, emphasis was placed on the identification and harmful effects of drugs, peer pressure was frequently discussed, and videos using scare tactics were often shown. These drug education units lasted approximately 30 to 45 min over a period of 2 to 4 weeks.

Pre-DARE levels of lifetime alcohol use and positive and negative expectancies about alcohol use were significantly related to their counterparts 10 years later. DARE status was unrelated to alcohol use or either kind of alcohol expectancy at age 20.

Pre-DARE levels of past-month marijuana use and negative expectancies about use were significantly related to their counterparts 10 years later. Similar to the findings for cigarettes, respondents' sixth-grade positive expectancies about marijuana use were not significantly related to marijuana expectancies at age 20. DARE status was unrelated to marijuana use or either kind of marijuana expectancy at age 20.

Finally, the number of illicit drugs (except marijuana) used in the past year was examined. Because no measures for these items were obtained during the initial baseline measurement, we estimated a means-as-outcomes HLM using no Level 1 predictors and only DARE status as a predictor at Level 2. The results show that DARE had no statistically significant effect on the variety of illicit drugs used.

The results for peer-pressure resistance were similar to previous results. Pre-DARE levels of peer-pressure resistance were significantly related to peer-pressure resistance levels 10 years later, whereas DARE status was unrelated to peer-pressure resistance levels.

Finally, pre-DARE levels of self-esteem were significantly related to self-esteem levels at age 20. Surprisingly, DARE status in the sixth grade was negatively related to self-esteem at age 20, indicating that individuals who were exposed to DARE in the sixth grade had lower levels of self-esteem 10 years later. This result was clearly unexpected and cannot be accounted for theoretically; as such, it would seem best to regard this as a chance finding that is unlikely to be replicated.

Our results are consistent in documenting the absence of beneficial effects associated with the DARE program. This was true whether the outcome consisted of actual drug use or merely attitudes toward drug use. In addition, we examined processes that are the focus of intervention and purportedly mediate the impact of DARE (e.g., self-esteem and peer resistance), and these also failed to differentiate DARE participants from nonparticipants. Thus, consistent with the earlier Clayton et al. (1996) study, there appear to be no reliable short-term, long-term, early adolescent, or young adult positive outcomes associated with receiving the DARE intervention.

Although one can never prove the null hypothesis, the present study appears to overcome some troublesome threats to internal validity (i.e., unreliable measures and low power). Specifically, the outcome measures collected exhibited good internal consistencies at each age and significant stability over the 10-year follow-up period. For all but two measures (positive expectancies for cigarettes and marijuana), measurements taken in sixth grade, before the administration of DARE, were significantly related to measurements taken 10 years later, with coefficients ranging from small ( b = 0.09 for positive expectancies about alcohol) to moderate ( b = 0.24 for cigarette use). Second, it is extremely unlikely that we failed to find effects for DARE that actually existed because of a lack of power. Thus, it appears that one can be fairly confident that DARE created no lasting changes in the outcomes examined here.

Advocates of DARE may argue against our findings. First, they may argue that we have evaluated an out-of-date version of the program and that a newer version would have fared better. Admittedly, we evaluated the original DARE curriculum, which was created 3 years before the beginning of this study. This is an unavoidable difficulty in any long-term follow-up study; the important question becomes, How much change has there been? To the best of our knowledge, the goals (i.e., "to keep kids off drugs") and foci of DARE (e.g., resisting peer pressure) have remained the same across time as has the method of delivery (e.g., police officers). We believe that any changes in DARE have been more cosmetic than substantive, but this is difficult to evaluate until DARE America shares the current content of the curriculum with the broader prevention community.

One could also argue that the officers responsible for delivering DARE in the present study failed to execute the program as intended. This alternative seems unlikely. DARE officers receive a structured, 80-hr training course that covers a number of topics, including specific knowledge about drug use and consequences of drug use, as well as teaching techniques and classroom-management skills. Considerable emphasis is given to practice teaching and to following the lesson plans. Although we did not collect systematic data on treatment fidelity in the present study, a process evaluation by Clayton, Cattarello, Day, and Walden (1991) attested to the fidelity to the curriculum and to the quality of teaching by the DARE officers.

Finally, advocates of DARE might correctly point out that the present study did not compare DARE with a no-intervention condition but rather with a control condition in which health teachers did their usual drug-education programs. Thus, technically, we cannot say that DARE was not efficacious but instead that it was no more efficacious than whatever the teachers had been doing previously. Although this is a valid point, it is unreasonable to argue that a more expensive and longer running treatment (DARE) should be preferred over a less expensive and less time-consuming one (health education) in the absence of differential effectiveness ( Kazdin & Wilson, 1978).

This report adds to the accumulating literature on DARE's lack of efficacy in preventing or reducing substance use. This lack of efficacy has been noted by other investigators in other samples (e.g., Dukes et al., 1996; Ennett et al., 1994; Wysong, Aniskiewicz, & Wright, 1994). Yet DARE continues to be offered in a majority of the nation's public schools at great cost to the public ( Clayton et al., 1996). This raises the obvious question, why does DARE continue to be valued by parents and school personnel ( Donnermeyer & Wurschmidt, 1997) despite its lack of demonstrated efficacy? There appear to be at least two possible answers to this question. First, teaching children to refrain from drug use is a widely accepted approach with which few individuals would argue. Thus, similar to other such interventions, such as the "good touch/bad touch" programs to prevent sexual abuse ( Reppucci & Haugaard, 1989), these "feel-good" programs are ones that everyone can support, and critical examination of their effectiveness may not be perceived as necessary.

A second possible explanation for the popularity of programs such as DARE is that they appear to work. Parents and supporters of DARE may be engaging in an odd kind of normative comparison ( Kendall & Grove, 1988), comparing children who go through DARE with children who do not. The adults rightly perceive that most children who go through DARE do not engage in problematic drug use. Unfortunately, these individuals may not realize that the vast majority of children, even without any intervention, do not engage in problematic drug use. In fact, even given the somewhat alarming rates of marijuana experimentation in high school (e.g., 40%; Johnston, O'Malley, & Bachman, 1996), the majority of students do not engage in any drug use. That is, adults may believe that drug use among adolescents is much more frequent than it actually is. When the children who go through DARE are compared with this "normative" group of drug-using teens, DARE appears effective.