The Good Behavior Game Professional Development Models Trial: Preliminary Results Based on Classroom Observations

Introduction: This study is a four-year IES Goal 3 project that was designed to understand the level of professional development (PD) required for teachers to implement and sustain the Good Behavior Game (GBG), a universal preventive intervention. GBG is a classroom-based behavior management strategy aimed at reducing aggressive/disruptive behavior and socializing children into the role of student. GBG, delivered in first and second grades, has been shown to reduce rates of substance abuse and other deleterious outcomes into young adulthood, especially for students that were rated as aggressive or disruptive at entry into first grade by their teachers.

Methods: Two models of PD are being compared with each other and with a control condition, in a randomized field trial.  GBG Basic, which provides group-based pre-implementation training to teachers supplemented by a group-based booster session, and GBG w Coach, which has the same activities under GBG Basic plus a coach who works directly in the classroom with the teacher. The first year implementation cohort includes 18 schools, 71 classrooms, and 1343 students with active parent consent.  The analysis is conducted separately for males and females and the independent variables include the treatment condition and a measure of classroom language status. The dependent student-level variables include measures of off-task and disruptive/aggressive behavior based on classroom observations conducted by independent observers. HLM approach for binary outcomes that acknowledges correlated standard errors was used to analyze the data.   An added complexity is that 36 of the classrooms are bilingual and the interventions effect may vary by classrooms’ language status which affects the statistical power of the study.

Results: Our descriptive analysis shows that the disruptive/aggressive behavior decreases 3.9% between fall and spring in the Standard classrooms, 14.1% in the GBG Basic classrooms and 12.9 % in the GBG w Coach classrooms.  The off-task behavior in the afternoon decreased 2.4% in the Standard classrooms, 12.4% in the GBG Basic classrooms and 6.6% in the GBG w Coach classrooms.  Our preliminary HLM results are promising showing borderline significance (p-values between 0.05 and 0.1) for reduction of off-task behavior in the afternoon in both GBG Basic and GBG w Coach conditions compared to the standard condition.

Conclusions: The preliminary results based on simple HLM analysis are promising.  Our next step is to estimate the potential intervention effects by using a model that fully utilizes the minute-by-minute structure of the data (an IRT approach).  We will also analyze the data using growth modeling approach that utilizes all three data collection points (fall, winter, and spring).