RUME 28
2026
Alexandria, Virginia
Using Large Language Models for Qualitative Coding of Structured Data on Post COVID Research Based Instructional Strategies
Page: 1079
In the spring of 2025, introductory Physics, Calculus, and Chemistry instructors were surveyed about their usage of active learning techniques and use of Research Based Instructional Strategies. Using open-source large language models running on consumer hardware, responses to open ended survey items data were analyzed to better understand the usage of specific strategies during and after the Pandemic Response Teaching period resulting from the COVID-19 pandemic, as well as for reasons how and why instructors adopted, maintained, or stopped RBIS. While the large language models were able to apply qualitative codes to the data, their effectiveness was significantly varied based on the dataset, initial prompt, large language model used, code specificity, and code prevalence.