Bridging the Gap: Evaluating AI-Powered Math Tutoring to Support Low-Income Middle School Students

Despite longstanding policy efforts, low-income and marginalized students in the
UK continue to face persistent mathematics attainment gaps. This study evaluates the
impact of Eedi, an AI-powered mathematics tutoring tool that focuses on solving the
misconception cascade, on achievement outcomes in UK state-funded secondary schools.
We will conduct a clustered randomized controlled trial (RCT) during the 2026–2027 school
year, involving approximately 236 classes (around 5,900 students). This study’s primary
outcome is improvement in mathematics achievement, measured through standardized
assessments (Renaissance STAR Math), with a focus on differential effects for low-income
students. Secondary outcomes include student engagement and teacher experience. The
intervention integrates the tutoring platform into classroom instruction, compared with
business-as-usual practice. Power calculations are based on detecting a 0.2 standard
deviation effect size. Data sources will also include student surveys and platform analytics.
By building on evidence from a prior UK RCT conducted by Eedi, this study will provide
rigorous causal estimates of the effectiveness of education technology in reducing
attainment gaps and inform policy decisions on scalable, cost-effective approaches to
advancing educational equity.

This project is supported through the Learning Engineering Virtual Institute (LEVI), in collaboration with J-PAL North America, which builds off of a previous pilot project.