Practice Profile: Schoolwide Positive Behavior Interventions and Supports

Lee and Gage (2020) conducted a meta-analysis to evaluate schoolwide positive behavior interventions and supports on school-level behavior, academic, and organizational outcomes. To find eligible studies, the following electronic databases were searched: Education Resources Information Center, Academic Search Premier, Education Index Retrospective, Education Source, PsycINFO, Psychology and Behavioral Sciences Collection in EBSCO host, and ProQuest Dissertations and Theses Global. Then, a hand search of the Journal of Positive Behavioral Interventions and a forward-and-backward search were conducted. Lastly, authors who were identified in the search were contacted about current, unpublished, or submitted research.

After the removal of duplicated records, 1,711 records were included for review. The abstracts were reviewed for inclusion eligibility using the following criteria: 1) the study was conducted in 12 public schools; 2) the independent variable was implementation of schoolwide positive behavior interventions and supports; 3) the primary outcomes could be assessed at the school level; and 4) studies used group experimental or quasi-experimental design with a true comparison group, excluding case studies (for example, pre-post), single-case design, literature reviews, or meta analyses, and correlational studies (e.g., within a group longitudinal designs). Studies with one-to-one school comparisons also were excluded because school-level variance could not be calculated, as were studies with proprietary models (e.g., Safe and Civil Schools).

Thirty-two studies using either a randomized controlled trial design (n = 6) or a quasi-experimental design with a comparison group that did not receive schoolwide positive behavior intervention and support training (n = 26) were included. Studies were conducted in the United States (California, Connecticut, Florida, Georgia, Hawaii, Illinois, Kentucky, Maryland, Minnesota, Missouri, North Carolina, Oregon, Pennsylvania, South Carolina, and Texas), and one study was conducted in Norway. Sixteen studies were conducted in elementary schools, three in middle schools, two in high schools, and 11 included mixed levels of schools with the majority at the elementary level. The average student enrollment was 698 students per school. All but three of the studies focused on the implementation of Tier 1 schoolwide positive behavior interventions and supports. Two studies evaluated the implementation of all three tiers, while one study evaluated the universal and tertiary tiers.

Outcomes of the studies were coded by behavior, academic, and organizational domains. The CrimeSolutions review of this practice focused exclusively on the behavior domain, problem behavior (measured by office discipline referrals, in- and out-of-school suspensions, and teacher ratings). Most studies reported proportions of students with a behavior outcome, while one included study used the Teacher Observation of Classroom Adaptation checklist, a teacher-completed rating scale.

As most studies reported more than one outcome, a robust variance estimation was used to analyze the behavior outcome. Standardized mean difference effect sizes (g) and the variance of g for all outcomes was calculated. The unit of interest was the school-level effect of schoolwide positive behavior interventions and support; thus, the number of schools was used to calculate the variance of g, even when the study reported student-level outcomes, to ensure that the weights were applied accurately in the meta analysis.