A Latent Transition Analysis to Explore the Effect of Two Training and Support Models for the Good Behavior Game on Teacher Reported Student Behavior

Introduction: Schools are a normative setting where children spend a considerable amount of time, and thus are an important environment in which to deliver preventive interventions.  A number of efficacious school-based prevention programs, such as the Good Behavior Game (GBG) have been identified, yet implementing these programs with fidelity remains a significant challenge. This presentation reports results from a randomized field trial that tested two different training and support models for GBG. The training models tested in the trial were ‘GBG Basic’ and ‘GBG w Coach’. The GBG Basic training model is included two days of group-based pre-implementation training supplemented by a one-day group-based booster session midyear while the GBG w Coach model in addition included a coach who worked directly with the teacher in the classroom: observing, modeling, mentoring, planning, and providing feedback.

Methods: Two professional development (PD) models are being compared with each other and with a control condition, in a randomized field trial in which classrooms were assigned to different conditions. We conducted latent transition analysis (LTA) to estimate whether the GBG Basic or GBG w Coach models would affect the student risk profiles on teacher reported data on student behavior (e.g., students moving from low to high or high to low risk classes). LTA uses longitudinal data to identify transitions between the risk profiles (e.g., dependent variable) based over time and further, the impact of independent variables such as the GBG training model on the transitions can be assessed. Participants in the first implementation year of the study included 18 schools, 71 first grade classrooms and 1,328 students.   

 

Results: Our a priori hypothesis based on prior research was that the impact of GBG would be greatest for students who entered first grade classrooms with heightened baseline risk.  We also hypothesized that teachers’ level of implementation and the effect of GBG would be greater in classrooms in which teachers received ongoing support (e.g. GBG w Coach). The implementation data based on teacher self-reports and classroom observations revealed an overall lower level of implementation than intended and no significant difference in level or quality of implementation were detected between the tested PD models.  The LTA analyses showed positive impact for GBG Basic compared to Standard in bilingual classrooms where the teacher reported student baseline risk factors were higher.  In the bilingual classrooms larger proportion of high or elevated risk students moved to low risk class. However, there was no difference in the proportion of students moving from the low risk class to elevated or high risk class. These results existed both for male and female students. Similar results were not found for the GBG w Coach model.

Conclusions: These results suggest that in moving programs into practice, focusing on the delivery system (e.g., training and support to teachers) of interventions (e.g. GBG) and understanding the interplay between the epidemiology and the context becomes of central importance. Both the intervention and the training model must match students’ and teachers’ needs to be implemented and to have the potential to provide positive effects.