Study
Brown and colleagues (2011) conducted a clustered randomized controlled trial to determine the impact of Steps to Respect® on reducing bullying behaviors and improving school climate and connectedness. In this study, the researchers emailed district contacts to establish a selection pool of schools to participate, using the following criteria: 1) had socioeconomic and racial/ethnic diversity, 2) had an established liaison with the Committee for Children, 3) expressed a strong need or desire for school bullying prevention, and 4) were not currently implementing a school bullying prevention program. Schools considered as day schools, alternative schools, or private/parochial schools were excluded. To minimize cost, the study researchers focused on schools in three geographic areas located in north–central California. From these three geographic areas, 45 of the 96 districts expressed interest in the study; thus, schools within these districts were considered for inclusion.
The study researchers selected 34 schools to participate in the study. These schools were matched based on school and student characteristics (i.e., size, change in student enrollment from 2006 to 2007, number of teachers, percentage of students who spoke English as a second language, and percentage of students receiving free or reduced-price lunch) and were randomly assigned to either the intervention group or the waitlist control group. Between the random assignment periods and the start of the intervention program, two schools withdrew from the study. The study researchers were immediately able to replace one school (selecting a school from the original pool of eligible schools) but not the other. Thus, 33 schools were included in the analyses.
All students in third through sixth grade in participating schools were invited to participate in the study (n = 3,119). Of these students, 2,940 students were included in the study sample. Fifty percent of schools were from suburban areas, 25 percent were from rural areas, 15 percent were in midsized cities, and 10 percent were from small towns. About half (49 percent) of students in intervention schools were male, 52 percent were white, 7 percent were Black, 6 percent were Asian American, and 35 percent were of other or mixed race. In regard to ethnicity, 43 percent of students in intervention schools identified as Hispanic. For students in control schools, 52 percent were male, 53 percent were white, 6 percent were Black, 6 percent were Asian American, and 35 percent were of other or mixed race. In regard to ethnicity, 41 percent of students in control students identified as Hispanic. There were no statistically significant differences in student demographics (i.e., race, ethnicity, age) between students in intervention and control schools.
Data were collected at pretest and posttest (i.e., immediately following the end of the intervention) periods, using student and teacher surveys. Outcomes of interest included teacher-reported student social competency, teacher-observed physical bullying, teacher-observed nonphysical bullying, student-reported bullying perpetration, and student-reported bullying victimization. The study researchers developed the Teacher Assessment of Student Behavior Survey, a brief online survey to measure all teacher-reported and observed outcomes. Teachers were instructed to respond to survey items using the time frame “since the beginning of the school year,” which included items such as “Gets along with classmates,” “Pushed, shoved, or tripped a weaker student,” and “Spread rumors about another student.” Teacher responses were on a five-point Likert scale ranging from 1 (“never”) to 5 (“always”). Student-reported bullying perpetration and bullying victimization were measured using a revised version of Colorado’s Trust’s Bullying Prevention Initiative Student Survey, which included items such as “I teased or said mean things to certain students” and “A particular student or group of students spread rumors or told lies about me.” Student responses were on a four-point scale ranging from 1 (“strongly disagree”) to 4 (“strongly agree”).
Hierarchal linear model analyses were used to determine differences between students in intervention and control schools. The study authors did not conduct subgroup analyses.
Study
Low, Frey, and Brockman (2010) conducted a clustered randomized controlled trial to evaluate Steps to Respect®’s impact on reducing forms of playground relational aggression (such as gossiping), using the same study sample as in Study 1 (Frey et al. 2005). This study examined student playground observation data that were collected in the fall (October to December; pretest) and the spring (April to June; posttest), from the same six matched-pair elementary schools (intervention group = 3 schools; control group = 3 schools), located in the Pacific Northwest.
As in Study 1, the total observation sample consisted of 544 students, with 296 students in intervention schools and 248 students in control schools. Student demographics in this study (ethnicity and gender) were also the same as in Study 1. There were no statistically significant differences between intervention and control students in rates of perpetrated gossip or targeted gossip at pretest (i.e., baseline). However, there were gender and grade differences in rates of perpetrated gossip and rates of targeted gossip at baseline. Because these differences were statistically significant, covariates for gender and grade were entered as fixed level-1 variables.
Outcomes of interest included observed rates of exclusionary gossip perpetration and victimization. Both outcomes were measured using a coding scheme developed by study researchers (based on reviewing videotapes of playground behaviors in Toronto, Ontario, and in vivo observations on Pacific Northwest playgrounds). Gossip perpetration was defined as a student who was involved in slander/malicious exclusionary gossip (e.g., “Is the cootie girl in your class?”). Gossip victimization was defined as a youth who had experienced or was a target of derogatory talk or labels applied to a third party (e.g., “Did you hear Dan cheated on the exam?”).
Hierarchical linear model analyses were used to determine differences in rates of gossip perpetration and victimization between students in intervention and control schools. The study authors did not conduct subgroup analyses.
Study
Frey and colleagues (2005) conducted a clustered randomized controlled trial to determine the impact of Steps to Respect® on reducing school bullying problems in six elementary schools located in the Pacific Northwest. Participating schools were located within two suburban districts and matched on school characteristics (i.e., size, ethnic breakdown, and percentage of students receiving free or reduced-priced lunch). Matched-pair schools were randomly assigned to the intervention (n = 3) or control condition (n = 3), with a matched pair in one district starting in the 2000–2001 school year, and the other two matched pairs in another district starting in the 2001–02 school year. Students and teachers assigned to the intervention condition received the Steps to Respect® program. Students and teachers in the control condition did not receive any type of training or classroom intervention. All third through sixth grade students in participating schools were invited to participate in the study (n = 1,126). Of this initial study sample, 70.0 percent were white, 12.7 percent were Asian American, 9.0 percent were Black, 7.0 percent were Hispanic American, and 1.3 percent were Native American.
Students were eligible for the study sample if they were observed for at least 40 minutes during pretest playground activity. Five hundred forty-four students (intervention group = 296 students; control group = 248 students) met this criterion and were included in the study analysis. Of this sample, 50.7 percent of students were male. There were no statistically significant differences between students in the total sample and students in the observation sample, based on baseline characteristics (such as ethnic background and English proficiency; however, the study authors did not provide the breakdown of baseline characteristics for the observed study sample, only for the initial study sample). The study authors reported that common reasons for incomplete data were students’ moving to another school or missing recess.
Data were collected through survey and observation methods. Observation data were collected across 2.5 months between October and December (pretest) and between April and June (posttest), using handheld devices. Student self-report survey data were collected in the fall around mid-November (pretest) and in the spring about 1 year after program implementation in late April or early May (posttest). Outcomes of interest included student-reported bullying behaviors (including direct bullying/aggression, indirect bullying/aggression, and bullying victimization) and observed playground behaviors (including aggressive bullying behavior and nonbullying behavior).
The CrimeSolutions review of this study focused only on the student-reported measures of direct and indirect bullying/aggression, student-reported victimization, and observed behaviors of bullying and nonbullying aggression. Student bullying behaviors were measured using items from the Student Experience Survey: What School Is Like for Me, which defined direct bullying/aggression behavior as a student participating in explicit bullying behavior (e.g., “I called kids names at school”), indirect bullying/aggression as a student participating in bullying behavior that could create harmful social effects (e.g., “I told my friends to ignore kids I was mad at”), and victimization as a student having been bullied by other students (e.g., “A group of kids at school called me mean names”). Observed playground behaviors were measured by the following five behavior categories: 1) bullying, 2) encouragement, 3) nonbullying aggression, 4) agreeable social behavior, and 5) argumentative social behavior. Bullying behavior was defined as physical, verbal, or indirect aggression involving either a discernible power imbalance between an aggressor and a target (or aggressors—that is, a group of children aggressing against a single child), or repeated aggression, during the same observation session, by a child toward a nonretaliating peer. Nonbullying aggression behavior was defined as physical, verbal, or indirect aggression that did not involve a discernible power imbalance or repeated nonreciprocal aggression.
Hierarchical linear models were used to determine differences in bullying behaviors between students in Steps to Respect® intervention and control schools. Chi-square analyses were used to determine observational differences between students in intervention and control schools. Subgroup analyses were conducted to examine grade and sex differences between students in intervention and control conditions.
