Dataset for “The Effect of Artificial Intelligence-Assisted Programming Learning on Students’ Computational Thinking: A Meta-Analysis”

This dataset supports the meta-analysis titled The Effect of Artificial Intelligence-Assisted Programming Learning on Students’ Computational Thinking. The underlying research hypothesis was that AI-assisted programming learning has a positive effect on students’ computational thinking (CT), and that this effect may vary across study characteristics and instructional conditions. The dataset contains study-level coded information extracted from 19 empirical studies, including publication information, educational level, teaching strategy, programming environment, sample size, intervention duration, CT measurement tool, instructional function of AI, type of AI, methodological quality scores, and the effect-size data used in the meta-analysis. Two researchers coded the studies independently, and inter-coder reliability was high (Cohen’s kappa = 0.902). Study quality was also independently assessed using the Kmet et al. checklist, with quality scores ranging from 0.818 to 0.917 and inter-rater reliability of 0.925.

The data were analyzed in Comprehensive Meta-Analysis 3.0 using a random-effects model because the included studies differed in participants, interventions, and implementation contexts. Hedges’ g was used as the effect-size indicator. Publication bias was examined using funnel plot inspection, Begg’s test, Egger’s test, fail-safe N, and trim-and-fill analysis; the results suggested that publication bias was unlikely to materially affect the findings. The overall pooled effect was large and positive (Hedges’ g = 0.949, 95% CI [0.650, 1.247], p < .001), indicating that AI-assisted programming learning was associated with significantly improved CT. However, heterogeneity was substantial (Q = 129.398, I² = 86.1%), so moderator analyses were conducted. Significant subgroup differences were found for CT measurement tool and instructional function of AI, while educational level, teaching strategy, programming environment, sample size, intervention duration, and AI type did not show significant between-group differences. In particular, larger effects were observed when CT was measured by scales rather than tests, and when AI served practice and feedback or motivation and exploration functions. A leave-one-out sensitivity analysis showed that the results were robust. These data should therefore be interpreted as study-level evidence showing an overall beneficial effect of AI-assisted programming learning on CT, while also indicating that the magnitude of this effect depends partly on how CT is measured and how AI is used instructionally.