Machine and Human Observable Differences in Groups’ Collaborative Problem-Solving Behaviours

Abstract

This paper contributes to our understanding of how to design learning analytics to capture and analyse collaborative problem-solving (CPS) in practice-based learning activities. Most research in learning analytics focuses on student interaction in digital learning environments, yet still most learning and teaching in schools occurs in physical environments. Investigation of student interaction in physical environments can be used to generate observable differences among students, which can then be used in the design and implementation of Learning Analytics. Here, we present several original methods for identifying such differences in groups CPS behaviours. Our data set is based on human observation, hand position ( fiducial marker) and heads direction (face recognition) data from eighteen students working in six groups of three. The results show that the high competent CPS groups spend an equal distribution of time on their problem-solving and collaboration stages. Whereas, the low competent CPS groups spend most of their time in identifying knowledge and skill defi ciencies only. Moreover, as machine observable data shows, high competent CPS groups present symmetrical contributions to the physical tasks and present high synchrony and individual accountability values. The findings have signifi cant implications on the design and implementation of future learning analytics systems.