Experimental Design
Of the approximately 300 students at this middle school, 150 will be assigned to receive in-school math tutoring. Our primary hypothesis is whether the average end-of-year MAP math score for the treatment students is higher than that of control students. We will also explore whether there is heterogeneity in the treatment effect based on the nature of tutoring. Specifically, 90 of the 150 treatment students will receive thrice weekly sessions in 3-student groups, whereas the remaining 60 will receive twice-weekly sessions in 2-student groups. The cost per student is the same ($40 per week per student) for both of these conditions, so a higher treatment effect for one type would imply a higher return on investment for that specification of tutoring frequency/group size.