Poster
This study explores the efficacy of using large language models (LLM’s) to identify motivationally supportive instruction in STEM classrooms. Drawing on self-determination theory, we focused on three support strategies: providing rationales, linking content to real life, and demonstrating enthusiasm. Initial coding and analysis of agreement between models and human raters revealed challenges and possible enhancements for LLMs to record supportive instruction. Pursuing these improvements could greatly enhance the efficiency of classroom observation in educational research.
McGill University, Montreal
Commutatus
Commutatus
McGill University, Montreal