DEEP LEARNING FOR INTEGRATING THE PANCASILA STUDENT PROFILE IN ELEMENTARY EDUCATION

Abstract

This study examined whether a Deep Learning pedagogical approach could improve Pancasila Education achievement and support development of the Pancasila Student Profile among Grade IV pupils. A quantitative quasi-experimental nonequivalent control group design was used with 40 pupils from two intact classes at SD Inpres 2 Piru: 20 in the experimental group and 20 in the control group. Achievement was assessed through multiple-choice pre-tests and post-tests, while profile-related change was summarized from pre-test/post-test questionnaire scores. The experimental group increased from M = 69.50 (SD = 8.414) to M = 85.50 (SD = 6.262), whereas the control group increased from M = 67.25 (SD = 8.807) to M = 76.00 (SD = 7.363). The post-test group difference was significant, t(38) = 4.396, p < .001, Cohen's d = 1.39. The experimental group also obtained an overall Pancasila Student Profile N-Gain of 0.6134, a medium gain under the conventional normalized-gain criterion. The findings indicate a strong achievement advantage in this study context and moderate within-group improvement in profile scores. Because the profile outcome was not reported as a between-group comparison, causal interpretation of profile development remains limited.