Effects of robotics-assisted programming instruction on students’ computational thinking: a meta-analysis

Abstract

Evidence on whether educational robotics (ER)–assisted programming instruction improves students’ computational thinking (CT) remains mixed. To synthesize existing findings, we conducted a meta-analysis of 33 empirical studies (39 effect sizes) published between 2015 and 2025. Eligible studies were identified through searches of Web of Science, ScienceDirect, Taylor & Francis Online, Wiley Online Library, SAGE Journals, IEEE Xplore, and Google Scholar. Effect sizes were calculated using Hedges’ g, and a random-effects model was adopted because substantial heterogeneity was observed (I²=79.61%, Q=186.369, p<0.001). The results showed a statistically significant moderate positive effect of ER-assisted programming instruction on CT (g=0.621, p<0.001). Subgroup analyses further indicated that intervention duration and programming mode explained part of the heterogeneity: longer interventions produced larger effects, and plugged programming yielded stronger gains than unplugged programming. These findings support ER as an effective approach for promoting CT in programming instruction and suggest that instructional duration and programming mode should be carefully considered in
ER-supported learning design.