Understanding Students' Collaborative Problem Solving during STEM+C Learning using Multimodal Analysis

Abstract

This dissertation research explores the intersection of STEM+C learning, collaborative problem solving (CPS), and multimodal learning analytics (MMLA) in K-12 education. As technology-enhanced learning environments become more prevalent, integrating computational modeling into STEM curricula has emerged to enhance the synergistic learning of scientific and computational knowledge. However, the complexity of learning STEM and computing simultaneously poses significant challenges for students. Collaboration offers a potential solution by enabling students to utilize their collective knowledge and problem-solving skills. This work aims to provide a deeper understanding of how students engage in CPS behaviors during computational modeling tasks in a kinematics curriculum. To capture the complexity of CPS behaviors, the dissertation advances the use of MMLA techniques that allow for richer, context-specific details, allowing for a nuanced analysis of students' learning processes over time. Our approach employs multimodal analysis to investigate the metacognitive, social, and cognitive domain-specific aspects of students’ CPS behaviors and their relationship to STEM+C learning. By analyzing social metrics such as equity, turn-taking, and integrating science and computing concepts, this study identified behaviors linked to task success. This includes synergistic collaborative discussions during important learning processes like planning, enacting, monitoring, and reflecting. An in-depth case study analysis reveals how differing distributions of prior knowledge in physics and computing relate to the development of shared understanding and STEM+C learning. Additionally, this research explores the interplay between collaborative and individual problem-solving strategies, and interprets strategy use in terms of students’ STEM+C knowledge. Overall, this work contributes to three key areas: (1) enhancing the understanding of CPS behaviors in STEM+C learning, (2) developing context-specific MMLA techniques to analyze the evolution of students' learning behaviors and support in-depth qualitative analysis, and (3) exploring the relationship between CPS behaviors, STEM+C learning, and problem-solving skills. Through this multi-dimensional approach, the dissertation aims to inform future work on designing more effective learning supports and instructional strategies to enhance students’ simultaneous development of scientific and computational knowledge in K-12 education.