Deep Learning-Based RADEC With a Mini Book for Critical Thinking in Grade 5 Science

Authors:

Abstract
The study examined the effectiveness of a Deep Learning-based RADEC (Read–Answer–Discuss–Explain–Create) intervention assisted by a Mini Book for enhancing fifth-grade students’ critical thinking in science learning on sound and its properties. A quantitative quasi-experimental nonequivalent control-group design involved 54 students from two intact classes at an anonymized elementary school in Bandung City, Indonesia. The experimental group received three sessions organized through RADEC and oriented toward mindful, meaningful, and joyful learning, with the Mini Book used continuously as concise content, guiding questions, and experimental tasks. The comparison group received contextual learning supported by instructional videos. Critical thinking was measured using a 10-item essay test representing four operational indicators. The experimental mean increased from 64.52 to 81.04, whereas the comparison mean increased from 63.33 to 75.19. The experimental N-gain was 0.4666, indicating a moderate improvement, and the posttest difference was significant, t(52)=3.099, p=.003. The findings provide context-specific evidence that the integrated intervention was associated with stronger short-term critical-thinking performance than the comparison condition. Because the design did not isolate the separate effects of Deep Learning, RADEC, and the Mini Book, the results should be interpreted as evidence for the intervention package rather than for any single component.