Study
Shope and colleagues (1992) examined the effectiveness of the Alcohol Misuse Prevention Study (AMPS) curriculum for fifth and sixth grade students using a randomized pre–post, experimental-control design. Forty-nine schools (with 213 classrooms) from six school districts southeastern Michigan were randomly assigned to one of three conditions: the curriculum, the curriculum plus booster, or the control condition. Only fifth grade classes were assigned to the curriculum plus booster so that they could receive the booster in sixth grade.
Pretesting was done in fall 1984. The curriculum was implemented in winter 1985. The first posttest was done in spring 1986. The booster curriculum was implemented in winter 1986. The second posttest was done in spring 1986. And the third posttest was done in spring 1987.
The study reported on the outcome results 26 months after the initial program was implemented. The pretests conducted in fall 1984 included 5,356 fifth and sixth grade students. The final posttest in spring 1987 included 3,833 students when they were in seventh and eighth grades. The study did not provide information on the demographics of the students.
Data was collected from students who completed a confidential, self-administered questionnaire that covered topics such as alcohol use and misuse, understanding of the curriculum content, susceptibility to peer pressure, and health locus of control. Alcohol use was self-reported by students in separate items in response to questions asking about the frequency and quantity of beer, wine, and distilled spirits use during the previous 12 months. Alcohol misuse was measured by 10 items asking about overindulgence, trouble with peers, and trouble with adults experienced as a result of alcohol use during the previous 12 months. Understanding the curriculum material was measured by 17 items that assessed knowledge of alcohol effects, pressures to use alcohol, and perceived ability to resist pressure. Two indices measuring susceptibility to peer pressure and internal health locus of control were created from the results of factor analysis.
Because only the fifth graders experienced the booster curriculum, the effects of the AMPS curriculum were assessed for fifth and sixth grade students separately. The CrimeSolutions review examined the differences between treatment and control group fifth graders at the 26-month follow up. Repeated measures analyses of variance were used to test for significance in changes on the dependent variables. Although schools were the unit of assignment to the experimental conditions, individual students were the unit of analysis. No subgroup analyses were conducted.
Study
In a follow-up to the 1992 study in which the researchers randomly assigned fifth and sixth grade classes from 49 schools, Shope and colleagues (1996) examined the effect of a high school–level adaptation of AMPS by concentrating exclusively on a cohort of students who had previously participated in the AMPS project as sixth, seventh, and eighth graders. The design restricted the cohort to students in nine high schools in four school districts and defined the sample as being established in 10th grade. Earlier exposure to the sixth grade prevention curriculum was viewed as a background factor. Pretesting occurred in late fall of 10th grade, and the curriculum was implemented in winter of the same school year. The 10th grade posttest occurred 2 months after intervention in spring, and the 12th grade posttest was administered in spring of 12th grade.
In the initial sixth grade sample, there were 2,024 eligible students. In 10th grade, 1,100 (54.3 percent) of these students took the 10th grade pretest survey, as did as 931 new students, resulting in a baseline sample of 2,031. Of the 2,031 students at baseline, 1,613 (79.4 percent) completed the posttest questionnaire in 10th grade and 1,185 (58.3 percent) completed the posttest questionnaire in 12th grade. Overall, a total of 1,041 students (51.3 percent) completed all three questionnaires and provided the data that are used herein.
The study collected data on alcohol misuse prevention knowledge, refusal skills, alcohol use, alcohol misuse, and drinking after driving. Alcohol misuse prevention knowledge was measured by 31 items regarding alcohol facts and effects, application of that information to typical alcohol-related situations, pressures to use alcohol, and perceived ability to resist pressure. Each student’s alcohol refusal skill was rated by both male and female raters, in addition to a self-rating. Students self-reported alcohol use and misuse. Frequency and quantity of alcohol were assessed separately for beer, wine, and liquor. Alcohol misuse was measured by 10 items assessing the frequency of types of negative consequences experienced as a result of alcohol use during the previous year. Finally, driving after drinking was assessed by a single questionnaire item asking students, “During the past 12 months, how many times did you drive after drinking?”
To evaluate the effectiveness of the 10th grade curriculum on refusal skill ability, analysis of variance (ANOVA) by treatment and gender was used. To evaluate the effectiveness of the curriculum on other dependent variables, repeated measures ANOVAs were used. Significant treatment by occasion interactions were of primary interest and would indicate differential rates of change between the treatment groups, which, if in the desired direction, would support the effectiveness of the AMPS curriculum. Subgroup analyses were conducted to examine the impact of gender on program outcomes.
Study
In another follow-up to the 1992 randomized study, Shope and colleagues (2001) looked at the long-term effects of the AMPS curriculum on students 7 years after they had obtained their drivers’ licenses. The study used the same sample of students from the 1992 study but restricted the analyses to those who participated in the 10th grade pretest and who had obtained a Michigan driver’s license by June 1997, resulting in 1,820 subjects who received the AMPS curriculum in the 10th grade and 2,815 control subjects. These students were clustered into 254 classrooms. The total sample was 50.5 percent male, 85.1 percent white, with an average age of 16.4 years at the time of licensing. The treatment group was slightly more likely to be white (86 percent, versus 84 percent) and older at licensure (16.5 years, versus 16.4 years), compared with the control group. The analysis did not control for students before their participation in the AMPS curriculum, so the study essentially was looking at the effectiveness of the program offered to 10th graders and not at the effectiveness of the AMPS curriculum offered in fifth and sixth grade with follow-up in the 10th grade.
Students were asked about their alcohol use and misuse. Alcohol use was measured by separate items asking about the frequency and quantity of beer, wine, and liquor use. Two questions were used to assess alcohol misuse: 1) during the past 12 months, how many times did you get drunk? and 2) during the past 12 months, how many times have you had five or more drinks in a row? Ten additional questions were used to construct an alcohol misuse index.
The primary outcomes of interest were serious offenses. Serious offenses included those that met any of the following criteria: 1) involved use of alcohol; 2) were classed as “serious” by the Secretary of State’s office (e.g., reckless driving, vehicular homicide); 3) resulted in 3 or more points assigned to a driver (e.g., speeding in excess of 15 mph over the speed limit; 4) involved nondriving drug offenses. Information about traffic offenses and reported crashes between 1986 and 1997 was obtained from Michigan’s driver history files for all participating students who obtained a driver’s license in the state by June 1997. Offense data was available only for those offenses that resulted in convictions. In addition, only crashes reported to the police were included.
Poisson regression was used to model the number of incidents per year as a function of treatment group, duration of license, and potential confounders such as sex, race, alcohol use and misuse, relicense offenses, age at licensure, family structure, and parental attitudes toward alcohol. Statistical significance was assessed by using likelihood ration tests for nested Poisson regression models and Wald tests for generalized estimating equations. No subgroup analyses were conducted.
