Examining Student Moderators for the Effectiveness of School-Wide Positive Behavioral Interventions and Supports When Taken to Scale across a State

Purpose: Positive Behavioral Interventions and Supports (PBIS) is a framework for improving school systems, data use, and practices to guide the selection and implementation of evidence-based practices for promoting positive behavior. The universal, school-wide components (i.e., SW-PBIS) have been widely disseminated to over 25,000 schools in nearly all US states. The current study presents the main effects of SW-PBIS on school-level outcomes as well as student-level main and moderation effects in one state’s scale-up of PBIS. Prior efficacy research indicates that SW-PBIS is most effective for at-risk students.

Method: Propensity score weights were generated for each school. Longitudinal autoregressive models were conducted to examine school-level impacts on suspension, truancy, and reading and mathematics achievement rates. Multilevel models will be utilized to examine repeated measures of these same outcomes at the student level, controlling for student demographics. Weights and SW-PBIS effects will be examined at the school level and cross-level interactions (i.e., moderation) between SW-PBIS and student –level indicators. The sample includes students in all 1,316 public elementary, middle, and high schools in one state.

Results: In a first stage of analyses, school-level models demonstrated significant effect of SW-PBIS training on the suspension rates in elementary schools (i.e., d = 0.17 and 0.18, in 2009-11 and 2010-11) and reading (d = 0.32 in 2006-7, d = 1.00 in 2007-8, and d = 0.30 in 2010-11) and mathematics proficiency (d = 0.63in 2006-7, d = 0.34 in 2007-8, d = 0.31 in 2009-10, and d = 0.23 in 2011-12). SW-PBIS training had significant effects on all suspensions (d = 0.03), truancy (d = 0.43), and achievement (ds = 0.58 for reading and 0.46 for math) in secondary schools 2007-08, as well as for reading (d = 0.53) and math (d = .030) in 2008-09. Student-level main effects are expected to replicate the above-described school-level effects. Specifically, we intend to examine effect moderation by demographic (e.g., gender, race, grade) and prior years’ behavioral and academic indicators. We hypothesize that students with poorer prior behavioral and academic performance will most benefit from SW-PBIS.

Conclusions: This study is responsive to the widespread dissemination of SW-PBIS nationally and internationally, and the state-wide infrastructure to disseminate SW-PBIS in nearly all states in the US. Further, it is the first to allow for causal inference regarding effectiveness in a state-wide scale-up. The results have important implications for implementation science, policies around positive (versus punitive) behavioral approaches in schools, and for educational practices.