Treatment Heterogeneity As a Function of School Socio-Emotional Climate in a Cluster-Randomized Trial in the Democratic Republic of Congo

Introduction: Variation in the impact of school-based interventions is rarely uniform across all treatment recipients.  Whereas a multitude of factors at different levels of the social ecology (e.g., individual, school) might underlie heterogeneous treatment effects, the majority of research focuses on individual-level moderators, while factors and processes at other levels remain poorly understood.  Examining whether and how setting-level factors moderate the effectiveness of educational interventions is necessary to inform initiatives for the improvement of children’s learning opportunities.  This study examines treatment heterogeneity as a function of schools’ socio-emotional climate.

Methods: Data come from the second year of a Cluster-Randomized Trial of “Opportunities for Equitable Access to Quality Basic Education” (OPEQ), an intervention developed by the International Rescue Committee that consists of training Congolese teachers in new curricula integrating math and reading lessons with socio-emotional learning principles and practices. OPEQ expects to enhance teachers’ motivation and performance and children’s numeracy, literacy and socio-emotional wellbeing. 

Fifty-four clusters (groups of 2-6 schools) were randomly assigned to three treatment cohorts (Pilot, Scale Up 1, Scale Up 2) with different start-up years (2011, 2012, or 2013).  The evaluation sample includes 84 schools from all clusters, where approximately 80 2^nd to 4^th grade children per school were assessed in 2011 (baseline) and 2012.  

Using children’s scores in the Early Grade Reading Assessment (EGRA) and Early Grade Math Assessment (EGMA), as well as school-level aggregates of students’ perceptions of school socio-emotional climate (supportiveness, predictability and cooperativeness), we will: 1) estimate the difference in children’s achievement scores between clusters assigned to the Pilot (1 year of treatment), and Scale Up 1 cohorts (no treatment), and 2) explore whether school socio-emotional climates moderate treatment effects.

Results: Preliminary analyses with baseline data indicated very low levels of performance in reading and math. At least 50% of children obtained zero scores on seven of nine EGRA sub-tests, and at least 20% obtained zero on six out of nine EGMA sub-tests.  Regarding school climate, children perceived schools and teachers as supportive (M = 3.28, SD = .46, scale: 1-4) and reported moderate levels of predictability and cooperativeness (M = 2.63, SD = .73, scale: 1-4). Especially relevant to our second aim, there was significant variation between schools and clusters.

Conclusions: Results from this study are expected to shed light on setting-level processes that are likely to impact treatment effects and play a central role in children’s school experiences and learning outcomes.