Abstract:

This experience report presents an empirical case study investigating the impact of immersive STEM workshops on the computational thinking (CT) skills of primary schoolchildren. The intervention was delivered in Microsoft Dream Space, a dedicated STEM education hub designed to inspire students and teachers to explore digital technology and computational concepts. Our research question focused on whether a pair of one-off workshops could result in measurable improvements in CT skills among 9–12-year-old students. This question is particularly relevant for education systems where CT is not yet embedded in the curriculum, and where short interventions may be the only available approach.

The study involved 160 students from five schools during spring 2024. The intervention comprised two educator-led workshops at Dream Space, complemented by teacher-guided activities in school over three weeks. Workshop activities combined unplugged and plugged approaches, including problem decomposition games, algorithm sequencing tasks, and coding challenges using MakeCode for micro:bit and Minecraft Education. These activities were designed to foster CT skills such as logic, algorithms, pattern recognition, and decomposition, while aligning with national education policies and providing teachers with follow-on resources.

To evaluate the intervention, we implemented a pre-post assessment methodology using Bebras tasks, which are puzzles that require CT skills to solve, but do not require programming knowledge. The assessment design incorporated isomorphic task pairs and a counterbalanced test version approach to mitigate testing effects and difficulty bias. Ordinal outcome data were analysed using cumulative link mixed models with planned comparisons for each CT skill. Additionally, computer self-efficacy (CSE) was measured using a modified validated scale adapted for younger learners.

Results revealed a statistically significant overall improvement in CT skills (p = 0.005) following the intervention. Logic showed the strongest individual skill improvement, with correct responses increasing from 46% to 61% (p = 0.003). Decomposition and algorithms also demonstrated positive trends, while pattern recognition showed modest gains, highlighting areas for targeted pedagogical refinement. Interestingly, the proportion of “don’t know” responses increased for pattern recognition tasks, suggesting greater metacognitive awareness post-intervention. CSE results were mixed, with slight decreases in overall confidence, possibly reflecting students’ growing awareness of the complexity of computing concepts.

Although limited in scope with one intervention style and one measurement approach, the findings suggest that immersive workshops supported by structured follow-up activities can effectively enhance CT skills in primary education. These results inform future research directions, including refining interventions to address weaker skill areas, integrating qualitative data to explore learner experiences, and developing robust tools for CT assessment. The study also raises important considerations for scaling CT education in contexts where CT is not on the curriculum.

Keywords:

Computational thinking, primary education, K-6 education, Bebras, pre-post assessment, active learning, STEM.