2026-10-07T22:26:30Z https://api.figshare.com/v2/oai

oai:figshare.com:article/34018470 2026-09-29T14:52:13Z category_28864 category_30391 category_29938 category_26182 item_type_3 month_year_09_2026

Data and code for "Gender bias across LLMs is common and highly heterogeneous" Edoardo Bolzoni (25133328) Valerio Capraro (25133331) Artificial intelligence not elsewhere classified Cognitive and computational psychology not elsewhere classified Applied ethics not elsewhere classified Gender, sexuality and education large language models gender bias stereotypes moral judgment LLM evaluation AI bias auditing replication study <p dir="ltr">Raw response data and Stata analysis code for the study "Gender bias across LLMs is common and highly heterogeneous" (Bolzoni & Capraro). Includes per-phrase response data for Study 1 (gender attribution to stereotyped phrases) and per-condition response data for Study 2 (moral judgment of abuse/torture against a woman or a man), across ten LLMs tested between April 2025 and July 2026, plus the Stata do-files used to compute inclusivity indices, conduct all statistical tests, and generate the figures reported in the manuscript and SI Appendix.</p> 2026-09-29T14:52:13Z Dataset Dataset 10.6084/m9.figshare.34018470.v1 https://figshare.com/articles/dataset/Data_and_code_for_Gender_bias_across_LLMs_is_common_and_highly_heterogeneous_/34018470 CC BY 4.0