Global effects of robot-based education on academic achievements, computation, motivation, and performance

Haoran Tang, Wei Xu, Yu Feng, Wenwen CaoHumanities 2025 · DOI

TL;DR

This paper provides a rigorous meta-analysis of 36 studies to quantify the impact of robot-based education on student outcomes. The researchers found that integrating robots into the curriculum leads to moderate-to-large improvements in academic achievement, computational knowledge, motivation, and overall performance compared to traditional teaching methods.

What Problem Does This Paper Address?

While the "cool factor" of robots in schools is undeniable, educational researchers have long debated whether they provide a substantive pedagogical advantage or are merely a high-tech distraction. Since the COVID-19 pandemic accelerated the shift toward digital and remote learning, the use of robotic platforms (from LEGO Mindstorms to humanoid social robots) has surged.

However, most existing research consists of isolated, small-scale studies. There has been a lack of high-level synthesis to determine if these benefits are consistent across different age groups, subjects, and cultural contexts. This paper addresses that gap by pooling data from dozens of studies to provide a "global" statistical view of how robots actually affect learning outcomes.

Methods

The authors conducted a systematic review and meta-analysis following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines.

Key technical aspects include:

  • Study Selection: Out of 609 initial records from databases like Web of Science and ScienceDirect, 36 studies met the strict inclusion criteria (e.g., randomized controlled trials with sufficient data).
  • Effect Size Calculation: The researchers used Hedge’s $g$ to measure the magnitude of the effect. Hedge’s $g$ is preferred over the more common Cohen’s $d$ in meta-analyses because it provides a more accurate estimate for studies with smaller sample sizes.
  • Statistical Modeling: Due to high heterogeneity ($I^2 > 50\%$) across the included studies, the authors employed a random-effects model, which assumes that the true effect size varies between studies.
  • Robustness Checks: To ensure the results weren't skewed by a single outlier or "file drawer" bias (the tendency to only publish positive results), they performed sensitivity analyses (leave-one-out), Begg’s and Egger’s tests, and funnel plot visualizations.

Datasets

The "dataset" for this meta-analysis consists of 36 peer-reviewed studies published between 2010 and 2025. These studies covered a wide demographic range, from preschoolers learning basic reasoning to university students engaged in complex engineering simulations. The robotic platforms involved varied from programmable kits (like LEGO) to social humanoid robots.

Key Insights and Ideas

  • The "Computational" Edge: The strongest positive effect was seen in computational knowledge ($g = 0.85$). This suggests that robots are uniquely suited for teaching logic, programming, and systems thinking because they provide immediate, physical feedback for abstract code.
  • Constructivism in Practice: The authors attribute these gains to the theory of constructivism—the idea that students learn best by actively building mental models through physical interaction. Robots serve as "tangible objects to think with."
  • Motivation as a Mediator: While the effect on motivation was "medium" ($g = 0.47$), it remains a critical factor. The interactive nature of robots stimulates curiosity and reduces the "fear of failure," which in turn drives better academic performance.
  • Consistency Across the Board: Perhaps the most surprising insight was the robustness of the findings. Even when the researchers systematically removed individual studies from the analysis, the overall positive effect remained stable, suggesting that the benefits of robot-based education are not just a fluke of a few high-performing programs.

Key Results

The meta-analysis revealed statistically significant positive effects across all five hypothesized dimensions:

Outcome Effect Size (Hedge’s $g$) Interpretation
Computational Knowledge 0.85 Large effect
Learner Performance 0.81 Large effect
Academic Achievement 0.72 Medium-to-large effect
Overall Educational Outcomes 0.71 Medium-to-large effect
Learning Motivation 0.47 Medium effect

All results were significant at the $p < 0.01$ or $p < 0.05$ levels, and no significant publication bias was detected.

Limitations and Future Work

Despite the positive results, the authors acknowledge several hurdles:

  • High Heterogeneity: The wide variety of robots, teaching styles, and student ages makes it difficult to prescribe a "one-size-fits-all" robotic curriculum.
  • The "Teacher Factor": The success of these programs often depends on the teacher's technical literacy and pedagogical approach, which is difficult to quantify in a meta-analysis.
  • Accessibility and Cost: Robots remain expensive. Future research needs to explore how to make these benefits accessible to schools with fewer resources.
  • Long-term Impact: Most included studies were short-term. The field needs longitudinal data to see if the "robot advantage" persists over years or if the effect fades once the novelty wears off.

The authors conclude that while robots are a proven "effective pedagogical approach," the next frontier is optimizing program design to ensure these tools are used for deep learning rather than just classroom entertainment.

Summary updated

Abstract

Given the demonstrated efficacy of robots in educational contexts, a wide range of robotic platforms have been specifically designed for educational purposes. While numerous individual studies have explored the effect of robot-based education, few systematic reviews or meta-analyses have synthesized robots’ effect on educational outcomes such as academic achievement, computational knowledge, learning motivation, and performance. This study aims to bridge this research gap by studying the effect of robots on educational outcomes through a rigorous meta-analysis and systematic review following established guidelines (PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses). Literature was retrieved from numerous online databases such as Web of Science, ScienceDirect, and EI Compendex. Based on inclusion and exclusion criteria, 36 studies were finally included for this study. Hedge’s g was adopted to calculate the effect sizes, I2 statistics was used to measure heterogeneity, and sensitivity analysis and publication bias tests such as Begg’s test, Egger’s test, and funnel plots were used to verify the robustness. The study concludes that robot-based education is associated with moderate-to-large improvements in academic achievement (g = 0.72; 95%CI = 0.30~1.13, medium-to-large effect), computational knowledge (g = 0.85; 95%CI = 0.55~1.15, large effect), motivation (g = 0.47; 95%CI = 0.19~0.75, medium effect), performance (g = 0.81; 95%CI = 0.22~1.40, large effect), and overall educational outcomes (g = 0.71; 95%CI = 0.50~0.92, medium-to-large effect) compared to traditional methods. No significant publication bias was detected, and the results were demonstrated to be robust. The study has confirmed that robot-based education is an effective pedagogical approach. Thus, there is a pressing need for future research to focus on designing robot-based education programs with a specific focus on optimizing educational outcomes.
BibTeX
@inproceedings{tang2025global,
  author = {Haoran Tang and Wei Xu and Yu Feng and Wenwen Cao},
  title = {{Global effects of robot-based education on academic achievements, computation, motivation, and performance}},
  booktitle = {Humanities},
  year = {2025},
  doi = {10.1057/s41599-025-05546-9},
}
Figure