AI Ed Wiki logoAI Ed Wiki
Use with AI

📄 Research Article

Coding, robots, computational concepts, and machine learning using the microbit card and the Maqueen and Nezha kits. A study in initial teacher training

● high · 2026-08-16 · teacher-education, programming-education, educational-robotics, machine-learning, cs-education, computational-thinking

Synthesis: Sáez-López, García-Jiménez, and de Lara García-Cervigón (2026) report a quasi-experimental study of an integrated Educational Robotics intervention in initial Teacher Education, run with 144 first-year Primary Education undergraduates from three Spanish public universities (experimental n = 105 at the University of La Laguna and the University of Oviedo; control n = 39 at the Complutense University of Madrid; 90.5% female, mean age 19.14, none with a computing/engineering background). Nine one-hour sessions (April–May 2025) combined Micro:bit with MakeCode visual block programming, the Maqueen robot, and the Nezha kit (Lego Technic) — with projects such as a dice, a light-sensitive "Light Car," and an automatic traffic-light barrier — plus an introductory machine learning task (the "Shy Panda" supervised image-classification project in Scratch 3.0 and Machine Learning for Kids). Using the 10-item Coding, Robotics, and Machine Learning Test (CRMT) and three four-point Likert attitude scales, the experimental group significantly outperformed the control on computational-concepts knowledge (M = 6.05 vs 4.51; t = 3.401, p = .001, Cohen's d = 0.638) and reported significant gains on machine learning, Creativity, and perceived benefits in Math Education and art. The authors conclude that coding and robotics are essential in initial teacher training, where programming exposure remains limited.

Key Findings

Quasi-experimental design and sample. The study used a quasi-experimental pre-test/post-test design for knowledge outcomes (Student's t-test) and a post-test-only comparison for attitudes (Mann–Whitney U). Group assignment followed existing course groupings (purposive, non-random); pre-test CRMT equivalence was confirmed, and an a priori G*Power analysis (Cohen's d = 0.5, α = .05, power = .80) required 128 participants, exceeded by the achieved 144 (post hoc power = 0.96).

Significant CRMT knowledge gains. On the ten-item CRMT (two items each across sequencing, loops/iteration, conditional logic, sensors/input-output, and introductory machine learning; Cronbach's α = 0.81), the experimental group scored M = 6.05 vs the control's 4.51 (t = 3.401, p = .001, 95% CI [0.64, 2.42], Cohen's d = 0.638; Hedges/Glass correction 0.635).

Scale-based attitudinal gains. On three scales (Cronbach's α average 0.911), significant experimental-group improvements appeared for knowing what machine learning is (61.9% agreement, p = .001), robotics enhancing creativity (90.4%, the scale's highest rating), understanding sequence/loop/conditional concepts (~80%, all p = .001), and perceived benefits for mathematics (93.3%) and art/music. Natural and social sciences and educational-innovation items were highly rated (>80%) but not significantly different from control. Large effect sizes (r = Z/√N) included loops (−0.847), active engagement (−0.673), conditionals (−0.659), and mathematics (−0.631).

Intervention content and feasibility. Activities were deliberately low-cost and accessible — Micro:bit (~20–25 EUR), Maqueen (~50–70 EUR), Nezha (~80–200 EUR) — with a shared teaching guide and instructor training to ensure fidelity across institutions, supporting Active Learning and collaboration in small stable teams of three to four students.

Limitations. The authors note purposive non-random sampling, a gender imbalance (90.5% female), institutional differences between experimental and control universities, self-reported attitudinal measures (Dimension 2 without baseline equivalence), and a short intervention period, which together limit causal inference and generalizability.

Control condition and training gap. The control group received the same theoretical content (Computational Thinking, visual programming, and robotics/AI applications in primary education) through traditional lecture-based sessions rather than hands-on activities, isolating the effect of the practical intervention. The authors frame the study around explicit training gaps in initial teacher education — limited prior exposure to visual CS Education, weak conceptual grasp of sequencing/loops/conditionals, unfamiliarity with robotics hardware, and minimal experience integrating programming across curricular areas.

Implication. Educational robotics and machine-learning activities should be embedded in Teacher Education so future teachers can teach Computational Thinking and CS Education in Higher Ed, strengthening Educational Robotics across the CS Education curriculum.

Connected Concepts

Connected Articles

Citation

Sáez-López, J.-M., García-Jiménez, A.-S., & de Lara García-Cervigón, S. (2026). Coding, robots, computational concepts, and machine learning using the microbit card and the Maqueen and Nezha kits. A study in initial teacher training. Computers and Education Open, 100366. https://doi.org/10.1016/j.caeo.2026.100366