Study
St. Pierre and colleagues (2005) used a randomized, two-cohort longitudinal design to assess the impact of Project ALERT on substance use beliefs and behavior in eight Pennsylvania middle schools. The study included 1,649 seventh grade students, who completed a pretest and four waves of posttests over the 2-year program and 1-year follow-up (baseline, postimplementation in seventh grade, before and after implementation in eighth grade, and follow-up survey in ninth grade). Half the students (50.5 percent) were male. Students were primarily white (81.4 percent); the remainder were African American (5.4 percent), Native American (2.2 percent), Hispanic (1.3 percent), Asian American (1.1 percent), and "other" (8.5 percent). The socioeconomic level of the schools varied, ranging from 10 percent of students eligible for free/reduced lunch to 60 percent of students eligible.
Classrooms of seventh graders were randomly assigned to one of three groups: the revised Project ALERT curriculum implemented by adult program leaders; the revised Project ALERT curriculum implemented by adult program leaders assisted by high school teen leaders; and a control group. The CrimeSolutions review looked at the differences between the treatment group implemented by adult program leaders and the control group. Measures included use of cigarettes, alcohol, and marijuana and mediating variables of substance use (e.g., future expectation to use, friends’ approval of substance use). Analysis used a three-level hierarchical linear model to account for a nested design. The authors conducted subgroup analyses on gender, risk, school, and implementation quality.
Study
Ellickson and colleagues (2003) conducted a randomized trial in South Dakota, where 48 school clusters were randomly assigned to two treatment groups and one control group. In one treatment group, participants received the revised Project ALERT curriculum. In the second treatment group, participants received the revised Project ALERT curriculum and booster sessions in 9th and 10th grades. For this analysis, these two treatment groups were analyzed as one group. In the control schools, students were not exposed to Project ALERT but continued to receive whichever drug information programs their schools offered. The schools encompassed urban, small-town, and rural communities. The study ran 18 months from fall 1997 to spring 1999. Group equivalence was increased by using blocking by geographic region and community size and stratification by community size and type.
The study included 4,276 students from 55 middle schools (21 control schools, n = 1,723 students; 34 treatment schools, n = 2,553 students). Half the students were female, 12.5 percent were nonwhite, and about 70 percent of the students lived with both biological parents. Teachers were trained to deliver the curriculum.
Data was collected at baseline, before program delivery, and at 18 months after baseline. Researchers collected saliva samples from 654 students to test cotinine concentrations. Data was collected on substance use beliefs and on alcohol, cigarettes, and marijuana use at three points: ever used, used within the past month, and used within the past year. Students were classified in one of three risk groups (baseline nonusers, experimenters, and users) per substance. Multiple controls helped rule out alternative explanations of treatment effects. All analyses were adjusted for attrition (about 9 percent for all groups) and clustering of students within schools. The authors conducted subgroup analyses to look at the impact of risk (i.e., baseline nonusers, experimenters, and users) on the outcomes.
Study
Ellickson, Bell and McGuigan (1993) used an experimental pre–post design with random assignment to place 30 California and Oregon schools in either a control or one of two treatment conditions. Seventh and eighth graders in 20 of the schools went through the Project ALERT curriculum. Adults taught the classes in 10 of the schools, while older teens drawn from nearby high schools assisted the adults in 10 others. In the remaining 10 schools, students were not exposed to Project ALERT but continued to receive whichever drug-information programs their schools offered. The CrimeSolutions review looked at the differences between the adult led program and the control group, where students were not exposed to Project ALERT. The schools encompassed urban, suburban, and rural communities. Nine schools had a minority population of 50 percent or more. Eighteen drew from neighborhoods with household incomes below the state median. Researchers used three methods to increase pretreatment equivalence between the three conditions (blocking by district, restricted assignment, and randomized assignment of schools).
Researchers followed participants for a 6-year period, from grade 7 through grade 12. To establish a baseline before the program began, the researchers surveyed 6,500 seventh graders about substance use and attitudes toward drugs. Over the next 5 years, the team conducted six follow-up surveys with nearly 4,000 of these teens as they moved through grade 12 (at baseline, then at 3, 12, 15, 24, 36, and 60 months after baseline). The surveys compared students’ drug use and related attitudes before, during, and after their exposure to Project ALERT’s curriculum, with similar data from students who had no contact with the program. Trained data collectors administered student surveys in all schools before and after program lessons. Self-reported drug use was validated by testing saliva samples collected from students and by consistency analyses over time. Logistic regression was used to analyze substance use outcomes as a function of treatment and baseline covariates. Multiple controls helped rule out alternative explanations of treatment effects. All analyses were adjusted for attrition and clustering of students within schools. The authors conducted subgroup analyses to look at the impact of risk (i.e., baseline nonusers, experimenters, and users) on the outcomes.
