The Effect of Integrated STEM Discovery Learning Model on Computational Thinking Skills in Heat and Its Conversion Material for Grade VII Students at SMP Negeri 1 Kalasan

This study examined the effect of a Discovery Learning model integrated with STEM (Science, Technology, Engineering, and Mathematics) on seventh-grade students’ computational thinking (CT) in the topic of heat and heat transfer. A quasi-experimental nonequivalent control group design was employed at SMP Negeri 1 Kalasan, involving two intact classes: VII B (experimental) and VII C (control). The instructional treatment embedded the Discovery Learning phases—stimulation, problem identification, data collection, data processing, verification, and generalization—within STEM activities that required students to model heat phenomena, analyze temperature–time data from simple experiments, and design testable solutions related to heat transfer and energy conversion. CT outcomes were assessed using a test aligned to core dimensions (decomposition, pattern recognition, abstraction, and algorithmic thinking), complemented by observation sheets to monitor implementation fidelity and student engagement. Data analyses indicated that the experimental class achieved significantly higher CT performance than the control class, with notable gains in decomposing heat-transfer problems, identifying patterns in empirical data, and articulating stepwise solution procedures. Observation results further revealed more active inquiry, collaboration, and iterative design behavior in the experimental group. These findings demonstrate that integrating STEM tasks into Discovery Learning effectively strengthens CT within science learning on heat and its conversion. The approach bridges conceptual understanding and procedural reasoning by engaging learners in authentic problems and guided design cycles. Practically, teachers are encouraged to incorporate structured STEM design challenges and targeted scaffolds for data interpretation and algorithmic expression. Future research should widen the sample, examine long-term retention, and explore professional development supports that sustain high-fidelity implementation.