Study
The study by Flannery and colleagues (2003b) also evaluated the impacts of the PeaceBuilders program for students in grades 3–5. The examination of the program’s effects on students in grades 3–5 used the same eight schools in Pima County, Arizona, as described in Study 1. However, as the study authors separated students in grades K–2 and 3–5 into two different study sample groups and used different data collection methods for each group, the studies have been treated separately. For students in grades 3–5, the authors used self-report surveys to collect responses, compared with face-to-face interviews conducted with students in grades K–2.
Of the student sample for grades 3–5, there were 954 students in the PeaceBuilders (treatment) program group and 943 in the control group. As in Study 1, 51 percent of the students were Hispanic, 28 percent were white, 13 percent were American Indian, 6 percent were African American, and 1.5 percent were Asian American. There were no statistically significant baseline differences between groups.
The program was implemented for 2 full school years (fall and spring) starting in 1994 (baseline), with a 1-year follow up in spring 1995 (Time 2). Additional follow ups (with the control group receiving the intervention) were conducted in fall 1995 (Time 3) and spring 1996 (Time 4). The CrimeSolutions review of the program focused on outcomes at the Time 2 follow up.
Outcomes of interest for grades 3–5 included teacher-rated child aggression and social competence, and child self-reports of aggression and prosocial behavior. Teachers for grades 3–5 used the same measures as the K–2 teachers (see Study 1). The students’ self-reported aggression was measured using 9 items, invented by the study researchers, that evaluated aggressive behavior through a 3-point scale ranging from “no” (1) to “a lot” (3) for scale items such as “ I hit someone” or “I put down other kids.” Child self-reported prosocial behavior was measured through a 16-item instrument that assessed empathy, caring, helpfulness, and support of others (i.e., “I helped adults at school without being asked” and “I helped other kids”) on a 3-point scale: “no” (1), “a little” (2), and “a lot” (3). A hierarchical linear model was used to measure group differences between baseline and the spring of year 1 (Time 2) follow up. This study authors did not conduct subgroup analyses.
Study
Flannery and colleagues (2003a) conducted a randomized controlled trial in Pima County, Arizona, to study the effects of the PeaceBuilders violence prevention program for grades K–2. Eight schools were selected based on reports of their high numbers of juvenile arrests and histories of suspensions and expulsions. The eight participating schools were from two large school districts in the central city and on the outskirts. Schools were grouped into four matched pairs primarily based on geographic proximity and the percentage of ethnic students; the percentage of students eligible for free or reduced-price lunch; and the percentage of students in English as a Second Language (ESL) classrooms. The matched pairs were randomly assigned to the PeaceBuilders (treatment) intervention (n = 4) or the control (no treatment) condition (n = 4).
It is important to note that data collection instruments and methods for the child self-report measures varied across grade level; face-to-face interviews were conducted with students in grades K–2, while teachers and students in grades 3–5 filled out bubble scan sheets to provide responses. These methods are considered best practice for the two age ranges. Results from the two forms of the instruments could not be combined, thus the study authors reported the outcomes separately for students in grades K–2 and students in grades 3–5. Thus, this CrimeSolutions review considered the results as two separate studies: Study 1 pertains to the results for students in grades K–2; Study 2 (described below) pertains to the results for students in grades 3–5.
A total of 4,128 students were included in the study sample (grades K–5). For grades K–2, 438 students in the prevention (treatment) group and 405 in the control group were included in the program analysis. Of the total sample, 51 percent were Hispanic, 28 percent were white, 13 percent were American Indian, 6 percent were African American, and 1.5 percent were Asian American. There were no statistically significant baseline differences between groups.
The program was implemented for 2 full school years (fall and spring) starting in fall 1994 (baseline), followed by a posttest in spring 1995 (Time 2). Additional follow-up periods (with the control group receiving the intervention) were in fall 1995 (Time 3) and spring 1996 (Time 4). This CrimeSolutions review focused on outcomes at the Time 2 follow up.
Outcomes of interest for students in grades K–2 included teacher-rated child aggression and social competence, and child self-reported aggression and prosocial behavior. Teacher-rated child aggression was measured using the 25-item Teacher Report Form (TRF), which rated child behavior on a 3-point Likert scale (0 = not true, 1 = somewhat true, 2 = very true). Teacher-rated child social competence was measured using the 19-item Walker-McConnell (W-M) Scale of Social Competence and School Adjustment (elementary school version; grades K–6) which included three scales: 1) School Adjustment (7 items); 2) Peer-Preferred Behaviors (7 items); and 3) Teacher-Preferred Behaviors (5 items), such as cooperation, self-control, and social behavior. Teachers rated the frequency of such behaviors on a Likert scale ranging from never (1) to frequently (5).
Data for children in grades K–2 were collected through individual face-to-face interviews, which were approximately 5 to 8 minutes in duration. Child self-reported aggression was measured by children answering five items through “yes” or “no” responses to questions that examined whether the child got into trouble at school, whether the child ever got into fights, and whether the child ever cut in line. Child self-reported prosocial behavior was measured through six questions based on sharing, helpfulness, saying “thank you,” and saying “I’m sorry,” and answers that included “yes,” “sometimes,” or “no, not really.” A hierarchical linear model was used to measure group differences between baseline and the spring of year 1 (Time 2) follow up. The study authors did not conduct subgroup analyses.