Examining deep learning implementation in pancasila education: classroom practices, student outcomes, and challenges

Abstract

This study aimed to examine the implementation of deep learning-based Pancasila Education, explore students’ learning experiences and academic outcomes, and identify implementation challenges at a model senior high school in Indonesia. A qualitative descriptive design was employed involving 24 purposively selected informants at SMAN 2 Padang. Data were collected through classroom observations, semi-structured interviews, and document analysis. The data were analyzed using an interactive qualitative framework and validated through source and methodological triangulation. Deep learning principles were implemented inconsistently across classrooms. Observations identified 85% of instructional indicators in class X E3 compared with 40% in class X E10. Meaningful learning, structured reflection, and assessment follow-up remained insufficiently integrated. Student achievement varied across assessment components, while implementation challenges included differences in student readiness, instructional practices, and learning resources. Deep learning-based Pancasila Education demonstrates potential to support student engagement and contextual understanding. However, consistent instructional planning, differentiated support, and reflective assessment are necessary to strengthen implementation. The descriptive findings do not establish causal effects on academic achievement