Exploring Students’ Conceptual Understanding of Statistical Physics Through Multi-Source Learning: A Mixed-Methods Study

Abstract

Statistical Physics is an intellectually demanding subject that requires students to navigate abstract representations, probabilistic reasoning, and mathematical formalism simultaneously. In today’s learning environment, students naturally rely on multiple learning sources, textbooks, YouTube videos, Google searches, and increasingly, explanations generated by artificial intelligence (AI). This mixed-methods study explores how such multi-source engagement shapes conceptual understanding among 28 undergraduate physics education students enrolled in a Statistical Physics course. Quantitative data were collected through a Conceptual Understanding Test (CUT), a multi-source learning questionnaire, a misconception inventory, and rubric-based artifact analysis. Qualitative data were gathered from interviews and reflective journals. The findings reveal that students’ conceptual understanding falls within a moderate-to-good range (mean CUT score 69.6), although misconceptions persist, particularly regarding entropy, ensembles, and the partition function. Students with strong conceptual understanding tended to integrate multiple learning sources reflectively constructing coherence across representations while weaker students demonstrated fragmented and shallow engagement. The study concludes that multi-source learning has the potential to reinforce conceptual understanding, provided students develop strategies for synthesizing diverse explanations rather than consuming them indiscriminately.