What works and at what cost to improve foundational learning for school-aged children in conflict and protracted crisis? A meta-analysis and cost-effectiveness analysis of education interventions for school-aged children

This study synthesizes evidence on education interventions that improve foundational learning outcomes for school-aged children in conflict and crisis-affected settings. Drawing on a comprehensive evidence review and meta-analysis of 36 experimental and quasi-experimental studies conducted between 2000 and 2024, we examine the impacts and cost-effectiveness across multiple categories of education interventions. Guided by the Education Research in Conflict and Crisis (ERICC) framework, we organize findings across four drivers of learning: access, quality, continuity, and system coherence. Across studies, the average effect size for foundational literacy is 0.27 standard deviations and 0.21 SD for numeracy, both exceeding the median effect size of 0.10 SD reported across low- and middle-income countries more broadly. Software with academic games exhibits strong and consistent positive impacts across literacy and numeracy outcomes, while structured pedagogy shows similar effects for literacy and phone-based tutoring for numeracy. Other interventions such as targeted instruction show emerging evidence of high impact for literacy but have not demonstrated impact on numeracy, despite being well established in more stable settings. Results indicate that interventions associated with larger gains tend to operate through common mechanisms, including reducing access and participation constraints, aligning instruction with learners’ skills through personalization, strengthening instructional capacity, and sustaining learning during periods of disruption. However, many interventions with evidence of impact lack costing data in conflict settings. Analysis of cost-effectiveness reveals substantial heterogeneity, with phone-based tutoring yielding strong evidence of higher cost-effectiveness, while evidence for several other intervention categories remains limited or uneven. The findings highlight the potential for meaningful learning gains in crisis settings and the importance of cautious interpretation given data constraints, contextual variability, and gaps in cost reporting.