Exploring the nexus of variables in English-medium instruction research: a comprehensive global meta-analysis

The primary objective of this study is to conduct a comprehensive meta-analysis of English-medium instruction (EMI) literature, addressing the significant gap in quantitative evaluations within existing research. By analysing 47 empirical studies employing statistical methods, this research aims to uncover broader trends and correlations within EMI data. Specifically, the study investigates correlations for various factors, such as measurement outcomes and academic subjects. It also examines the influence of different research methodologies on reported correlation coefficients, assessing how methodological choices shape observed relationships in EMI studies. Key findings reveal that perceived academic performance and perceived English language proficiency exhibit the highest effect sizes, underscoring the importance of students' perceptions in EMI success. Conversely, actual academic performance, actual English language proficiency, motivation, and affective factors show small correlations, highlighting areas for pedagogical improvement. Self-efficacy demonstrates a medium correlation, indicating its significant role. Computer Science stands out with the highest effect size, suggesting discipline-specific variations in EMI effectiveness. Different measurement tools, including questionnaires and standardised tests, generally show small correlations, indicating a need for more refined assessment instruments in future EMI research. This meta-analysis provides crucial insights for enhancing EMI implementation and assessment, contributing to the broader discourse on educational strategies in multilingual settings.