Understanding the Classroom Context and Its Influence on Intervention Implementation

Introduction: Research has shown that teacher ratings of student behavior vary as a function of their own experiences and attitudes (e.g., Pas & Bradshaw, 2014). Such teacher experiences and attitudes have also been highlighted as being associated with school-based implementation of preventive interventions (e.g., Han & Weiss, 2005; Domitrovich et al., 2015). Less is known, however, about how classroom context relates to implementation, particularly when that context is informed by both student behavior and teacher characteristics. The purpose of this study is to model classroom context in the multilevel framework, incorporating student-level information along with teacher characteristics to create a more holistic picture of classroom context. Further, we will explore the impact that classroom context has on intervention implementation, specifically how it relates to the implementation of the Good Behavior Game (GBG), a behavioral classroom management preventive intervention. Method: Data come from 2,797 teachers in 18 Maryland elementary schools. All teachers taught in schools that were randomized to the intervention condition of a RCT testing GBG versus an integration of GBG with the PATHS (Greenberg & Kusché, 2011) social emotional curriculum. Teachers completed a student rating form for each student in their classroom (i.e., the Teacher Observation of Classroom Adaptation; TOCA), prior to intervention implementation. Specifically, they rated student social competence, positive peer relations, emotion regulation, hyperactivity, aggression and impulsivity, and academic achievement. Teacher factors included items from the Behavior Management Self-Efficacy Scale (Main and Hammond, 2008), the Maslach Burnout Inventory (Maslach et al., 1997), and the Organizational Health Inventory for Elementary Schools (Hoy and Feldman, 1987). A multilevel latent profile analysis (LPA) approach was used to identify student behavioral profiles (level 1) and classroom context was modeled at level two with information from student profiles at level 1 along with teacher factors from level 2. The association between classroom context and implementation were then explored (i.e., total number of games). Results: The multilevel LPA revealed four latent student behavior profiles at level 1; those rated as highly competent (26.3%), competent (22.6%), displaying mild behavioral issues (27.6%), and aggressive (23.5%). These findings extend previous research on this sample demonstrating a significant association between teacher factors (e.g., fit with teaching style, emotional exhaustion) and intervention dosage (Domitrovich et al., 2015). Discussion: Findings will be discussed regarding the relationship between teacher factors, student behavioral ratings, and implementation. Implications for using multilevel modeling in universal intervention datasets will be discussed.