Study 1
Taylor and colleagues (2011) used an experimental design with random assignment to determine the effect of the Shifting Boundaries intervention on adolescent peer violence and sexual harassment victimization and perpetration. During September 2009 to June 2010 the research team randomly assigned New York City middle schools to one of four conditions: 1) the classroom-based intervention only, 2) the schoolwide intervention only, 3) both classroom and schoolwide interventions, and 4) a control (no intervention) group. Thirty schools completed the study and implemented their assigned condition as planned. The assignment to condition was as follows: six schools and 23 classrooms received only the classroom-based intervention; eight schools and 30 classrooms received only the schoolwide intervention; seven schools and 28 classrooms received both the school and the classroom intervention; and nine schools and 36 classrooms were the control group. The control group went through its normal class schedule and did not receive any of the elements of the classroom intervention or the schoolwide intervention. The CrimeSolutions review of this study focused on the comparison of students who participated in both the classroom and schoolwide interventions (the third intervention condition) with students in the control (no intervention) group.
The ethnic breakdown of the study sample (all four groups) was 34 percent Hispanic, 31 percent Black, 16 percent Asian, 13 percent White, with the remainder in the “other” racial category. The sample was evenly split between sixth and seventh grade students, with 1,266 students (48 percent) in sixth grade and 1,388 students (52 percent) in seventh grade. Participating students ranged in age from 10 to 15, with 94.5 percent falling in the 11-to-13 age range. Fifty-three percent of the overall sample was female.
In addition, baseline measures show that 40 percent of the study sample had prior experience with a violence prevention educational program. Nearly half of the sample (48 percent) reported at least one experience of being in a dating relationship that lasted 1 week or longer. About one in five respondents (19.4 percent) reported having been the victim of any physical or sexual dating violence at some point in time. Two thirds of the sample (66 percent) reported having been the victims of any physical or sexual peer violence at some point in time. One in five respondents (20 percent) reported having perpetrated any physical or sexual dating violence at some point in time. Nearly three out of five (57 percent) reported having perpetrated any physical or sexual peer violence at some point, and nearly half (45.8 percent) reported having sexually harassed someone at some time.
Comparison indicated there were no statistically significant differences between students in all four groups at baseline in age, gender, prior experience with dating violence–prevention programs, the length of their prior dating relationships, any peer violence victimization in their lifetimes, and any dating violence victimization in their lifetimes.
Data collection and measurement occurred at three different times: immediately before group assignment (baseline), immediately after the six lessons were completed, and 5 to 6 months after the conclusion of lessons (follow-up). The student surveys took about 40 minutes to complete and were administered during one classroom period. Student surveys were divided into five sections, measuring 1) knowledge, 2) attitudes, 3) behavioral intentions, 4) actual behavior, and 5) demographic information. The knowledge measures included questions about state rape laws, definitions of abuse and sexual harassment, resources for help, rape myths, and skills such as conflict resolution. Attitudinal measures covered the acceptability of violent, abusive, and harassing behaviors. The behavioral intentions were measured from responses about the willingness to intervene in harmful situations and interrupt harassment. Actual behavior was measured by self-reports of perpetration and victimization of dating violence and sexual harassment. The frequency (number of occurrences of a repeated event) was measured for the victimization and perpetration outcomes. The demographic portion of the survey covered standard age, gender, race, and ethnicity questions and prior attendance at an educational program about sexual assault, harassment, or violence, and prior history of dating.
Because of the clustered nature of the data (individuals within classrooms within schools), specific statistical procedures were used to account for the clustering. Dichotomous outcome variables were analyzed with logistic regression that included a robust variance estimate that adjusted for within-cluster correlation. Negative binomial regression with a robust variance estimate was used for all count data (e.g., “how many times were you victimized?”). Ordinary least-squares regression with a robust variance estimate was used for the normally distributed/linear outcomes such as the knowledge, attitude, and intention measures. No subgroup analyses were conducted.