Additional file 1 of Evaluating gender bias in large language models in long-term care

Supplementary Material 1: Three appendices are included: (1) Evaluation of sentiment metrics: establishing which sentiment metrics are appropriate for conducting this analysis. (2) Model diagnostics and robustness checks: verifying the robustness of conclusions using several other methods. (3) Evaluation of themes: full lists of words counted in the frequency of the words appearing in each theme. The code to reproduce this analysis is available with synthetic data in the GitHub repository [53]