Using Learning Analytics to Understand Students' Discourse and Behaviors in STEM Education

This chapter discussed how learning analytics can be used to analyze students’ discourse and behaviors in technology-enhanced STEM learning environments (e.g., collaborative learning, simulations, and modeling). Various machine learning methods such as text classification, transition rate analysis and sequential pattern mining, network analysis, and multilevel modeling were adopted to understand the relationships between learning outcomes and processes in terms of students’ transformative and non-transformative discourse, multifaceted engagement, self-regulation, as well as evaluation and reformulation behaviors during collaborative inquiry learning or individual design process. Findings from multiple studies are synthesized. Finally, research gaps, challenges, and future directions are discussed from the theoretical, methodical, and practical perspectives.