Estimating the Effect of ChatGPT-Assisted Instruction on Learning Outcomes in TVET, Vocational Education, and Engineering Education: A Meta-Analysis

Aim/Purpose: The aim of this study is to synthesize empirical evidence from Scopus-indexed studies and estimate the pooled effect of ChatGPT-assisted instruction on learning outcomes across the included TVET, vocational education, and engineering education contexts. Background: The rapid proliferation of ChatGPT has attracted growing interest in its potential to facilitate learning in TVET, vocational education, and engineering education. However, empirical findings remain inconsistent. Some studies report large learning gains, and others indicate small or weak effects. Methodology: We conducted a PRISMA-guided screening of Scopus-indexed, English-language, open-access publications from 2022 to 2026. Ten experimental or quasi-experimental studies with sufficient statistical information were used for effect-size calculation. We estimated the pooled effect size in JASP using a random-effects model. Complementary narrative and keyword analyses were also performed to contextualize the statistical findings by identifying the assessed learning domains, research keywords, opportunities, concerns, and future implementation strategies. Contribution: The specific contribution of this study lies in combining a quantitative estimate of the average effect with contextual evidence concerning implementation across the educational settings represented by the included studies. Findings: Across the 10 studies in this bounded corpus, ChatGPT-assisted instruction showed a statistically significant, moderate positive pooled effect across the cognitive, affective, and psychomotor outcomes included in the primary analysis (Hedges’ g = 0.739, 95% CI = 0.469–1.010, p < .001). Residual heterogeneity was substantial (I² = 83.12%), indicating considerable variation across studies. The positive direction and moderate magnitude remained stable across the assumed within-study correlations and in the analysis restricted to directly assessed outcomes. Recommendations for Practitioners: The findings support cautious use of ChatGPT as a supplementary learning tool. The meta-analysis did not establish any instructional condition or implementation strategy as universally effective. Recommendation for Researchers: Researchers should extend primary evidence to underrepresented educational contexts and report complete statistical and methodological information required for accurate effect-size calculation. Impact on Society: This study synthesizes evidence on ChatGPT-assisted learning outcomes across the included educational contexts, supporting more informed decisions about its responsible use without implying that the findings represent all TVET, vocational education, or engineering education settings. Future Research: Future research needs to broaden database coverage, include more types of documents when appropriate, and include non-English and closed-access publications when available. It should also explore the moderating effects of instructional design, prompt training, teacher guidance, assessment strategies, and students’ prior knowledge on ChatGPT-assisted learning.