From pilot to policy: experimental evidence from scaling online tutoring

Can effective tutoring programs survive the transition from pilot to scale, and from researcher to government implementers? We address this question using a randomized controlled trial of an online mathematics tutoring program scaled from a successful pilot and implemented entirely by regional education authorities in Spain. Critically, tutoring was delivered by interim public-school teachers-qualified educators already integrated into the government's teacher registry-rather than specially recruited tutors, offering a test of whether effects persist under a lower-cost, readily scalable staffing model. Assignment to tutoring increased end-of-year grades by 0.15σ and standardized math test scores by 0.11σ - approximately one-third of the original pilot effects. Treated students also reported improved confidence in their math abilities and reduced math anxiety. An experimental socio-emotional training module for tutors improved affective outcomes but did not enhance academic gains. The findings demonstrate that online tutoring can generate meaningful benefits at scale under government implementation, while providing realistic benchmarks for the "voltage drop" policymakers should anticipate when adopting evidence-based interventions.