Hierarchical clustering of groups’ collaborative discourses during the computer-supported collaborative concept mapping process
Computer-supported collaborative concept mapping (CSCCM), as one of the computer-mediated instruction and learning strategies, has been used to foster collaborative knowledge construction (CKC). Previous research has characterized groups based on final knowledge artifacts, products, or performances, rather than the temporal, process-oriented characteristics generated during the collaborative learning process. To fill this gap, this research clustered groups into distinct clusters based on the collaborative discourse data by using agglomerative hierarchical clustering approach, and examined the process characteristics of different clusters and associated performances. Four clusters were identified and labeled. Cluster 1, the high-performing cluster, was characterized as the actively-engaged, idea-centered, consensus-achieved, and socioemotional-engaged cluster. Cluster 2, the low-performing cluster, was characterized as the inactively-engaged, information-shared, goal-oriented, and reflection-revolved cluster. Cluster 3, the medium-performing cluster, was characterized as the inactively-engaged, problems-unsolved, and reflection-revolved cluster. Cluster 4, the medium-performing cluster, was characterized as the actively-engaged, idea-centered, and goal-oriented cluster. Based on the results, this research proposed instructional strategies and assessment implications for improving CSCCM research and practice.