Large Language Models Prove Effective in Teacher Education Research

Researchers at the University of Atlantico have demonstrated the potential of large language models (LLMs) in automatically encoding open-ended responses to gather data for applied statistics, with significant implications for teacher education. The study focused on demands-resources fit processes and engagement in teacher education, comparing results from ordinary Likert-type items with those obtained from LLMs. The findings revealed the reliability of LLMs in processing and quantifying large amounts of open-ended data, as well as an 'optimal margin' of demands-resources fit in student teacher engagement.

Key Takeaways:

• The study utilized LLMs to automatically encode open-ended responses from a sample of 499 student teachers (82% female, Mage=23.5 years), showcasing their potential in gathering data for applied statistics.
• Results demonstrated the reliability of LLMs in processing and quantifying large amounts of open-ended data, achieving accuracy comparable to that of scale measures.
• The study identified an 'optimal margin' of demands-resources fit in student teacher engagement, with study resources exceeding study demands maximizing engagement, and moderate levels of both demands and resources leading to intermediate engagement.
• The findings emphasize the value of integrating qualitative and quantitative approaches, enabling the large-scale analysis of qualitative insights while preserving their richness.
• The research holds significant potential for enhancing digital education frameworks by supporting adaptive learning systems and digital assessment practices.

Statistics:

• The study analyzed a sample of 499 student teachers, with 82% female and a mean age of 23.5 years.
• Results demonstrated the reliability of LLMs in processing and quantifying large amounts of open-ended data, with 499 responses analyzed.
• The study identified an 'optimal margin' of demands-resources fit in student teacher engagement, with 258 (52%) student teachers showing high engagement.
• The findings suggest that study resources exceeding study demands (41 of 499) maximized engagement, while moderate levels of both demands and resources (224 of 499) led to intermediate engagement.

Sources:

• NewsRx. Study Findings from University of Atlantico Provide New Insights into Education and Information Technology (Modeling Demands-resources Fit In Teacher Education Using Open-ended Data: a Methodological-substantive Synergy). Education Letter. October 29, 2025; p 834.
• Modeling Demands-resources Fit In Teacher Education Using Open-ended Data: a Methodological-substantive Synergy. Education and Information Technologies, 2025.