Average vs. distributional effects: Evidence from an experiment in Rwanda.

• Reading interventions can impact readers along the reading score distribution differently. • Distributional effects for the Literacy Boost program in a district in Rwanda are studied using Quantile Treatment Effects. • Literacy Boost aimed to make classroom pedagogy and parent and community involvement in reading more effective. • The impact of the program is concentrated among students (girls) at or above the 25th percentiles of higher-order reading outcomes. • Important in terms of policy to tailor programs to the needs to different types of readers. Programs and policies in education overwhelmingly focus on improving the average academic achievement for students. However, educational interventions focused on literacy improvement can impact readers at different levels of the reading score distribution differently, and the distributional differences may be more pronounced by sex. In this paper, I explore data from a field experiment in a district in Rwanda that was intended to make classroom reading pedagogy more effective and engage students, families, and communities in reading activities outside of school to improve students' reading outcomes. I study the distributional effects of the intervention on students on two higher-order reading outcomes – Kinyarwanda fluency and Kinyarwanda text comprehension, measured for students who met a basic literacy threshold. I find that a) the set of interventions had a positive and significant, albeit modest effect on students at or above the 25th percentile of the reading distributions b) the program had positive and significant effects on reading outcomes for girls only, and these effects differed along the reading achievement distribution. This work underscores the importance of tailoring programs and policies to the needs of different types of learners. [ABSTRACT FROM AUTHOR]