Study 1
Thompkins and Chauveron (2010) used a quasi-experimental design to determine the impact of The Leadership Program’s Violence Prevention Project (VPP) on student attitudes toward and use of aggressive behavior. Approximately 160 middle and high schools in New York, N.Y., implemented the program for 4 years, starting in the 2005–06 school year. Every year, a smaller subset of the treatment schools, those implementing VPP, were selected at random and compared to students who did not receive VPP lessons (the comparison group). Due to different school implementations of the program, proper random assignment to treatment and control groups was not possible. Students in the comparison group were matched with students in the treatment groups according to grade and academic level. Comparison group students did not receive any VPP lessons throughout the 4 years of the study.
This large study actually consisted of two smaller studies conducted on the different populations that VPP is designed to address: middle school and high school students. The middle school sample consisted of 3,264 students in sixth, seventh, or eighth grade from 24 New York City schools. The treatment group (n = 1,688 students) received the VPP curriculum administered in classrooms. The comparison group (n = 1,585 students) did not receive VPP or any other conflict-resolution training. The overall sample was almost evenly split on gender—51 percent boys—and was largely Hispanic (47 percent) and African American (36 percent). The treatment group and the comparison group differed significantly with regard to age, with the treatment group being younger. The treatment group also had a greater proportion of Hispanic students than Asian students. There were no other discernable differences between the two groups.
The high school sample consisted of 1,112 students in the 9th and 10th grades, drawn from 13 New York City schools. There were no significant differences between the treatment group (n = 587 students) and the comparison group (n = 517 students) in terms of age or racial composition. The treatment group had a slightly lower attrition rate than the comparison group, 25 percent versus 31 percent, respectively. The overall sample was again 51 percent boys, and largely Hispanic (42 percent) and African American (41 percent).
In both the middle and high school student samples, baseline surveys were administered in class before the implementation and teaching of VPP. Usually, the VPP curriculum began the week after baseline measurement. The surveys used — both pre- and posttest — gathered data on academic self-concept, peer support, normative beliefs about aggression, conflict-resolution skills, and avoidance behavior. Survey items consisted of Likert-type scales (e.g., ratings of “strongly disagree” to “strongly agree”), composite behavioral scales, and vignettes to assess students’ beliefs about aggression and the types of conflict-resolution behavior they may have used. All scales used showed moderate to strong measures of validity and reliability.
The impact of VPP on student behavior was assessed with hierarchical linear modeling. This analytic plan is more appropriate for school-based studies, as it takes into account the clustering of students within schools, as well as multiple data points for each individual subject. This allows for the influence of different school environments to be captured, as well as the influence of an individual on themselves over time: repeated measures of an individual cannot be considered independent observations, as the individual’s past behavior will have an influence on their current behavior. A three-level model was used for all analyses of behavior and attitudes. The first level represents the growth trajectory or change in behavior for each student. The second level represents the variation in growth between students within a school, and the third level represents variation between schools. The study authors did not conduct subgroup analyses.