Reaching the Last Mile: Digital Modules Transform Math Learning in Rural Philippines

The study finds that tablet-based computer-assisted instruction significantly improved math learning in remote Philippine schools, with especially large gains when teachers used the digital modules consistently. However, weak infrastructure, limited teacher capacity, and declining usage over time prevented the intervention from reaching its full potential.

The Asian Development Bank’s research division, together with the University of Minnesota and the Institute for Integrated Development Studies in Nepal, delivers a compelling investigation into whether digital technology can reverse the deep learning losses suffered by students in the Philippines’ most isolated public schools. After more than two years of COVID-19 closures, among the longest in the world, students in “last mile schools,” many without electricity or internet, were left relying on printed worksheets. The study tested whether tablet-based, computer-assisted instruction could fill this gap, revealing both impressive learning gains and the stubborn structural obstacles that hinder sustained use.

Bridging a Pandemic-Induced Learning Crisis

Even before COVID-19, Philippine students ranked near the bottom globally in math and reading. Once schools closed, learning poverty surged, especially in remote regions. To confront this, tablets containing digitized learning modules (DLMs) in math and English were distributed to students in 21 randomly selected schools. Solar-powered Moodle servers enabled teachers to assign lessons and track progress even without home connectivity. But delays in device procurement, LAN installation, and the return to face-to-face instruction meant the technology was used only briefly as the main instructional tool and mainly served as a supplement thereafter.

A Rigorous Trial, Uneven Implementation

The study employed a randomized controlled trial across 42 schools, surveying over 1,000 junior high school students at baseline in 2022 and retesting them in 2024. While treatment and control groups were well balanced, implementation varied widely. Many teachers, unfamiliar with Moodle and constrained by unstable power supplies, used static PDF files instead of the interactive modules. Server logs showed that schools used only about a third of available content at most, and usage fell sharply in the second year when active technical support ended. These logistical challenges undercut the intervention’s potential, yet learning gains still emerged.

Math Scores Jump, Especially for High-Grit Students

Despite modest implementation, students in treatment schools recorded 0.34 standard deviation gains in math, significantly larger than typical results found in global education interventions. English scores, however, showed little change, reflecting the difficulty of teaching language comprehension through short digital exercises. When focusing on schools that actually used the DLMs, math gains rose to 0.46 SD, confirming that meaningful usage drove improvements. Most striking, the few schools that continued using the system into the next academic year achieved enormous gains of 1.6 SD, a transformative effect rarely seen in education research. Students with higher baseline grit benefitted most, as did higher-achieving students, with quantile analyses showing the strongest impacts among those scoring in the top half of the distribution.

Promise, Potential, and Persistent Barriers

The findings reveal a technology with profound potential, capable of delivering learning gains equivalent to more than a full year of schooling, and even more when sustained over time. Yet the study also underscores the fragility of EdTech adoption in resource-poor environments. Technical failures, insufficient teacher training, and a lack of ongoing support prevented widespread adoption, and use collapsed once external support ended. Without complementary interventions that strengthen teachers’ digital competencies and support lower-performing students, such programs could inadvertently widen learning gaps. In conclusion, computer-assisted instruction can be a powerful tool for remote Philippine schools, but realizing its full potential requires building the infrastructure, skills, and institutional backing needed to sustain the technology beyond its pilot stage.