Effectiveness of EMI in higher education: a multi-level meta-analysis

Journal article

This study is one of the first comprehensive multi-level meta-analyses investigating the effects of English Medium Instruction in higher education. It synthesizes a total of 41 samples, comprising a total of 8,747 participants, from 28 quantitative studies, including 23 samples on content learning (7,659 participants) and 18 on English performance (1,088 participants). A comparison of EMI students’ post-test and pre-test scores (within-group comparison) revealed that students significantly improved on measures of content knowledge (d = 1.57, p < .001) and English proficiency (d = 0.81, p < .001), although we acknowledge that this finding does not account for external factors or establish causality. A comparison of post-test performance between EMI and non-EMI groups (between-group comparison), which aimed to assess EMI's effectiveness relative to non-EMI, showed that EMI students achieved comparable outcomes in content learning (d = 0.13, p = 0.14) and significantly outperformed non-EMI students in English development (d = 0.33, p = 0.009). Moderator analyses further revealed that studies with methodological limitations, such as selection biases for within-group comparisons or unaddressed group homogeneity for between-group comparisons, often overestimate the effectiveness of EMI for content learning. Based on our findings, we offer practical implications for the implementation of EMI in higher education.

Authors: Lee, H; Rose, H; Macaro, E; Lee, JH

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Elsevier
• Journal: System
• Volume: 133
• Article number: 103755
• Publication date: 2025-06-24
• Acceptance date: 2025-06-19
• DOI: 10.1016/j.system.2025.103755
• EISSN: 1879-3282
• ISSN: 0346-251X
• Language: English
• Keywords: multi-level meta-analysis; content learning; English medium instruction; English learning