Several reviews point out the extreme variability in effects across studies of the same type of program and recommend studies of potential moderators of program effectiveness, including the risk level of the population to which the prevention program is delivered, the timing of program delivery in the life-cycle, the duration of the program, and role, skill level, and background of the trainer.

Little is known about the optimal timing of delivery. Because many of the risk factors for later substance use appear at an early age (e.g., impulsive temperament, poor social skills, school failure, peer rejection) or with the behaviors of the parents (e.g., parental drug use and attitudes about drugs, family management practices), it can be argued that prevention should start very early in the life course and target parents as well as children. Many prevention efforts, however, are designed to be delivered during early adolescence, when social influence to use increases. Few studies have assessed effects on later substance use of attempts to alter risk factors that appear at an earlier developmental stage, but they have generally produced positive results. The anger-coping program of Lochman (1992) was an elementary school intervention that reduced substance abuse during adolescence.

The Good Behavior Game is another example of a successful early preventive intervention studies using a randomized control group design (Dolan et al., 1993; Kellam et al., 1994; Kellam and Anthony, 1998). This group-based behavior management program, in which small student teams are formed in each classroom, rewards the teams for achieving behavioral standards. Because the team reward depends on the behavior of each member of the team, peer pressure is used constructively in this program to achieve positive behavior. Early results suggest that the program reduced aggressive behavior, especially among the most aggressive subpopulation. Results from the first follow-up study showed a statistically significant reduction in teacher-rated aggression at 6th grade for certain subgroups. The most recent reports from the project show that by age 14, boys in the classrooms with the program had a lower risk of starting to smoke tobacco than boys in the control classrooms (Kellam and Anthony, 1998).

Other examples of early prevention efforts that have produced positive effects on measures of substance use have been studied using quasi-experimental research designs. They include the Child Development Project (Battistich et al., 1996) and the School Development Program (Comer, 1985). The Child Development Project targets elementary schools and includes the following components, all aimed at creating an environment to support positive youth development: