Study 1
Bradshaw, Mitchell, and Leaf (2010) conducted a longitudinal-group, randomized-effectiveness study of School-wide Positive Behavioral Interventions and Supports (SWPBIS) using 37 public elementary schools from suburban (48 percent); urban fringe (41 percent); and rural districts in Maryland, of which 49 percent received Title I support. Between 2 and 13 schools participated from each of five districts that varied by size. After the schools were matched on demographic factors, 21 were randomized to receive SWPBIS, and 16 were assigned as comparison schools that did not receive SWPBIS. Thirty-seven schools participated in the project over 2 consecutive years.
The Maryland State Department of Education (MSDE) provided school-level suspension rates for the baseline year through year 4 of the study. The percentage of students suspended was calculated by dividing the total number of suspension events for a given school year by the student enrollment for that year, multiplied by 100. The researchers used the Wilcoxon signed-rank test, in which a Z score was computed separately for the two conditions, to determine whether there was a significant reduction in suspension rates for the SWPBIS and comparison schools between the baseline year and year 4 of the study. Data on Office Discipline Referrals (ODR) was collected using the Internet-based School-Wide Information System; however, outcomes on ODR are not included here because ODR data was collected only from the SWPBIS schools and not from the comparison schools.
Study 2
Horner and colleagues (2009) conducted a randomized, wait-list controlled effectiveness trial between 2002 and 2006 with groups of elementary schools in Illinois and Hawaii. Thirty schools from each state were randomly assigned to either a treatment or control/delayed group. The schools were self-nominated by school building administrators, had not previously received training in SWPBIS, and had demonstrated state capacity to provide whole-school team building in SWPBIS. Following design adjustments, the study included 33 schools in the treatment group that received SWPBIS training at Time 1 of the study, and 30 schools in the combined control/delay group that received the training a year later, at Time 2.The only statistically significant difference between the two groups was the larger enrollment in the control/delay schools than in the treatment schools. Participating schools had an average enrollment of 471 students (ranging between 131 and 969); 61 percent non-white ethnicities (ranging between 2–100 percent); and 51 percent qualifying for free or reduced lunch (ranging between 0–99 percent).
The School Safety Survey (SSS) Risk Factor Score was used to measure the level of perceived school safety in the participating schools. The survey provides a summary rating of at least five individuals serving four varying roles within schools, including an administrator, a supervisory staff member, a classified staff member, and at least one teacher. The Risk Factor Score, an index examining design of space, crowding, perceived caring, perceived sensitivity to cultural differences, student bonding in the school, the quality of student–adult interactions, and the level of adult supervision, was used to determine the extent to which the overall school culture was perceived as a safer and more socially supportive environment. The researchers computed a partial correlation coefficient (r) to estimate the effect size.