Individuals in a group: exploring engagement patterns via within-group configurations of role profiles and their impact on performance in collaborative problem solving

Collaborative problem-solving (CPS) has been widely used in K-12, higher education, and informal learning to enhance the quality of student learning. Understanding the relationship between learning engagement and group performance is crucial for CPS pedagogy and analytics. However, few empirical studies investigated individual engagement role profiles in CPS and the impact of individual engagement, as reflected in within-group configurations of role profiles, on learning performance. In this study, deep learning was employed to evaluate learner engagement, cluster analysis was performed to identify engagement role profiles, and the qualitative cases were used to in-depth compare within-group configurations of role profiles in higher- and lower-achieving groups. The results showed that six distinct role profiles with varying levels of cognitive and socio-emotional engagement have been identified: free-rider, normal-constructer, immersed-engager, balanced-engager, questioning-refuter, and active-constructer. In addition, high-achieving groups were generally characterized by dominant role profiles of active-constructers and balanced-engagers, while lower-achieving groups were characterized by dominant role profiles of free-riders. Based on these findings, some implications for promoting high-quality collaborative learning through optimal design and structure of CPS activities are discussed.