Didactic methods and small-group instruction for low-performing adolescents in mathematics : results from a randomized controlled trial

Can high-dosage tutoring help low-performing adolescents? We implement a randomized experiment to test a twofold intervention: A teacher training program customized for instructing 8th graders who perform poorly in mathematics and two 4-6 week periods of targeted math instruction for lowperforming 8th graders, a majority in small homogeneous groups and the rest in larger and more heterogeneous groups. We randomized 24 schools to treatment and 24 schools to control. For students receiving small-group instruction, we find that test scores increase by .06 SD. Moreover, the share of low-performing students decreases by up to 25 percent. We find no impact on treated students in large groups. Classroom observations and surveys to teachers indicate higher fidelity to the didactic methods among teachers managing small groups.