When the Best Studies Show the Least: A Systematic Review and Typological Synthesis of Differentiated Instruction for Diverse English Learners in Inclusive Education

Abstract

Differentiated instruction is widely promoted as an inclusive response for diverse English learners, including those with special educational needs, yet syntheses rarely classify the evidence by how differentiation is operationalized, for whom, and under what design conditions. This study reports a systematic review with a configurative, typological synthesis of 56 controlled effect sizes drawn from 45 studies across 18 countries (2011–2024; 16,773 learners), following PRISMA 2020. A six-facet classification: learner/exceptionality, differentiation dimension, differentiation format, outcome domain, design rigor, and context, was developed and applied to every effect size. The map reveals a field that is broad but uneven: process and content adjustments dominate (55 and 41 of 56), while product and the learning environment are rarely addressed (20 and 13); reading accounts for 32 of 56 outcomes, while productive skills are severely underrepresented. Crossing the typology with design rigor exposes an inverse-rigor gradient: all 13 very-large effects (g ≥ 1.20) come from quasi-experimental studies (mean g = 1.03), whereas the 13 randomized effects average near zero (mean g = 0.10), with the null and negative results concentrated among young, struggling learners in well-resourced settings. Classifying, rather than averaging, reframes differentiation as a conditional, not a universal, good, and yields an evidence gap map for inclusive practice and future rigorous trials involving learners with identified support needs.