The mean effect size across all of these measures was 0.24. Tobler’s most recent analysis examined 120 programs of school-based drug prevention programs between 1978 and 1990 (Tobler and Stratton, 1997). This study showed median effect size of 0.14 on measures of tobacco, alcohol, and other drug use across all programs.
Results from the most recent meta-analysis of school-based drug prevention programs (Gottfredson et al., forthcoming) documents effect sizes slightly smaller than those from previous meta-analyses.2 This study found that across 88 relevant published treatment-control comparisons, the mean effect size for school-based prevention activities on measures of alcohol and other drug use (but not tobacco use) is statistically significantly different from zero. The mean effect size was 0.054, which (assuming a control group prevalence rate of 50 percent) translates into about a 2.7 point difference between the prevention and control groups in the percentage of students who report using a substance. Although the average effect size across all studies is small, the range of average effect sizes observed from study to study is broad (–.44 to .54) and varies by type of prevention program.
In contrast to these small effects of substance abuse prevention programs are the larger effects found on a wider array of outcomes of psychological, behavioral, and educational interventions. Lipsey and Wilson (1993), summarizing effect sizes from 302 reviews of such studies, reported an average of the average effect sizes across these reviews of 0.50 with a standard deviation of 0.29. Thus, relative to a much broader set of social and behavioral outcomes, substance use is more difficult to alter, at least through the types of prevention strategies that have been studied.
The limited evidence available suggests that some forms of prevention activities are effective for reducing some measures of substance use. Some studies produce a substantial effect, and others no effect or negative effects. This section takes a closer look at the available evidence in order to highlight its limitations as a basis for policy decisions.
Gottfredson et al. (forthcoming) recalculated effect sizes based on the entire population that received any of the program whenever possible. Also, if multiple effect sizes were available for different measures of substance use, these multiple measures were averaged to obtain one effect size per study. These practices, as well as the exclusion of effects on tobacco use, may explain the slightly lower estimates of the magnitude of effects found in this study.