Implementing Deep Learning STEAM to Improve Students’ 4C Skills

Developing students' 21st-century competencies—critical thinking, communication, collaboration, and creativity (4C)—remains a significant challenge, especially in remote and resource-limited schools. This study investigates the effectiveness of a Deep Learning STEAM model in improving 4C skills among elementary students. A mixed-methods sequential explanatory design was employed with 21 fourth-grade students at SDN Masalima IV, a remote island school in Sumenep Regency, Indonesia. The intervention integrated Deep Learning principles (meaningful, mindful, joyful learning) into a five-phase STEAM model: Observation, New Idea, Innovation, Creativity, and Society. Data were collected through pre- and post-tests of 4C skills, classroom observations, interviews, and student project documentation. Quantitative data were analyzed using paired-sample t-tests, while qualitative data underwent thematic analysis. Statistical analysis showed a significant increase in students’ 4C scores from a mean of 47.19 to 81.67 (p 0.001), with a large effect size (Cohen’s d = 4.72) and normalized gain (n-gain = 0.75). Thematic findings revealed enhanced student engagement, reasoning, collaboration, and creative output during project-based learning, supported by increased self-confidence and reflective thinking. The Deep Learning STEAM model proved effective in enhancing 4C skills within a low-resource context by aligning cognitive depth with interdisciplinary and experiential learning. It offers a scalable, context-sensitive strategy for implementing 21st-century skills education in remote areas.