The Efficacy of Generative AI on Students’ Performance in K-12 and Higher Education: A Meta-Analysis (Stage 1, 12:35 PM)

• In Event: e-Lightning Ed-Talk Session 3 - Stage 1

Abstract

This meta-analytic study synthesized the effect of generative AI (GenAI) interventions on students’ academic performance in K-12 and higher education settings to address mixed findings across studies. Eighty primary studies were analyzed. The mean effect size estimate showed significantly positive effects of GenAI’s interventions on student outcomes. However, a substantial degree of heterogeneity was observed, suggesting that moderating variables might explain differences in effects. Moderator analyses demonstrated that learning contexts (K-12 vs. higher education), language output (L1 vs. L2), fields of study (Humanities/Education vs. Other), and intervention durations had differential effects, while sample size and study quality did not show significance. Scientific significance and pedagogical importance are discussed in terms of the cognitive load theory and the technology acceptance model.

Authors

• Hye K. Pae, University of Cincinnati

• Hanzhong Sun, University of Cincinnati

• Detong Xia, Xi’an Jiaotong-Liverpool University

• Gulbahar H. Beckett, Iowa State University