VISUAL LEARNING ANALYTICS IN PEER INSTRUCTION: MAPPING STUDENTS’ CONCEPTUAL TRANSITIONS IN KINEMATICS

Abstract

Peer Instruction (PI) is widely known and most popular methods in physics. PI is an effective method for promoting student interaction and improving conceptual understanding in physics. However, instructors often lack real-time insights into students' thinking processes during PI. Guiding discussions effectively and addressing misconceptions became more challenging. This study examines how PI affects students' conceptual understanding and response transitions in kinematics, focusing on the visualization of response patterns using the Interactive Stratified Attribute Tracking (iSAT). The study involved 40 preservice science teachers and implemented eight conceptually rich multiple-choice questions (ConcepTests) adapted from Mazur's PI framework. Pre and post discussion responses were analyzed using descriptive statistics, normalized gain (N-Gain), paired sample t-test, and iSAT-based visualization. Results showed a significant increase in students' conceptual understanding after peer discussion (), with a moderate mean N-Gain of 0.30 and a large effect size (Cohen's ). Visual analysis using iSAT revealed important transition patterns such as correctly aligned and starburst on high gain questions, especially on non-graphic kinematics concepts. In contrast, questions involving motion graphs showed sliding and misaligned patterns, indicating persistent misconceptions and peer-driven response shifts toward incorrect choices. These findings suggest that while PI supports conceptual growth, it may inadvertently reinforce misconceptions in complex representational tasks. The integration of iSAT adds a valuable dimension to peer learning research by mapping students' transitions across discussion phases. Future research should explore instructional supports to reduce negative peer influence and strengthen understanding of graphical representations in physics.