Experimental Design Details
Schools and site selection will be conducted attending to the compositional characteristics of centers, accessibility and study logistics (i.e. feasibility for the IRC, access to population of interest). To select target settings, we have requested enrollment data from the Ministry of Education (MoE) to identify clusters of schools that could be randomly assigned to either the control or treatment condition for the duration of the study. Randomization will be conducted at the school level, yet only one randomly selected KG2 classroom per school will participate in the intervention. According to 2020 enrollment data, provided by the Ministry of Education in Jordan, in the governorates of Irbid, Balqa, and Karak there are 338 schools with at least 1 KG2 classroom (our universe for sampling), yet on average there are 1.76 (sd = 0.25) KG2 classrooms per school. Attending to these distributional features, and to prevent diffusion of the intervention across classrooms, we have decided to include only one KG2 classroom per school. This classroom will be randomly selected after randomization to the treatment condition, if the selected school reports more than one KG2 classroom. When schools only report one KG2 classroom that will be selected to participate. We have on a preliminary basis identified the sampling universe of schools with KG 2 classrooms for this study. Special attention will be paid to create baseline equivalence between both groups using such information. Aggregate school-level information (completely de-identified) about children or family’s refugee status, household socio-economic status, and special educational needs previously collected by the government of Jordan will also be requested as these variables can serve as a proxy indicator of children’s development and will employed in balance checks and stratification sampling strategies prior to random assignment. All children enrolled in randomly selected schools and classrooms will be eligible to participate, after obtaining parental consent and child assent. Based on preliminary data (from the 2019 and 2020 school year), we have identified that schools have 1.74 KG2 classrooms per site (sd = 0.25), and that classrooms, on average enroll 24.19 (sd= 9.41) children. Using this estimation as our best current guess, we anticipate that around 4500 children will be participating in the evaluation. Considering the current challenges that COVID-19 pose to in-person child participation in schools we have taken a conservative approach and anticipate that about 30% of the eligible sample of children could attrit of the study at any time. Also, according to these data, schools enrolled in 2020 95.48% children of Jordanian descent, with only 4.52% of children reporting a different nationality (i.e., Syrian, Palestinian).