Helping the Many by Tutoring the Few: Evidence from a Large-Scale Randomized Field Experiment

Being able to read well is a fundamental skill. However, on average around 20 percent of 15-year-olds in OECD countries are not proficient readers. Thus, finding cost-effective methods to help students at risk of reading difficulties is a high priority for educational policymakers. Tutoring shows promise in this regard but in-person, small-group instruction is considered costly and the scaling-up potential uncertain. Using data from a large cluster-randomized field experiment including 81 Danish schools and more than 2,000 students, we estimate the effects of an early literacy tutoring intervention in the first year of primary school. The treatment schools received a program in which the students most at risk of reading difficulties received tutoring in 15-minute sessions, 3-4 times a week, for around 10 weeks. Importantly, tutors were school personnel and the main extra resource supplied to the treatment schools was the program material. The total school-level effect is positive and significant on standardized tests of decoding (0.27 standard deviations, SD) and letter knowledge (0.16 SD). We show that there are substantial effects also on non-tutored peers, which indicate that scaled-up versions of tutoring programs may be more cost-effective than previously thought.