This dataset supports the llm-psychometric-fidelity-audit repository, which evaluates the psychometric fidelity of large language model responses on the IPIP Big Five personality questionnaire.
The dataset includes cleaned human survey responses and merged LLM-generated responses to the same 50 IPIP Big Five items. It is intended for reproducing the paper analyses, including comparisons of central tendency, distributional shape, reliability, latent… See the full description on the dataset page:
https://huggingface.co/datasets/Anonymous0624/llm-psychometric-fidelity-audit.
