Abstract
INTRODUCTION: STEM education's impact on student learning outcomes is nuanced, with differentiated effects across outcome types and academic levels. METHODS: This study uses meta-analysis to systematically analyse 66 experimental and quasi-experimental studies on STEM education published in international English journals from 2000-2024. RESULTS: The study finds that: (1) Subgroup analysis showed that STEM education had the most significant impact on cognitive outcomes in high school (d = 0.58) and reduced heterogeneity (I (2) = 62.1%), while the overall effect size was exploratory due to construct diversity. (2) Overall, STEM education has a moderate effect on students' learning outcomes, but the overall moderate effect size masks these key differences. (3) The effect of STEM education on students' learning outcomes is influenced by the moderating variables of sample size, academic level, subjects, experimental period and teaching method. DISCUSSION: These findings highlight the need to tailor STEM interventions to outcome type and academic level, strengthening the integration of theory and practice in STEM education.