No Simple Answer to Data Complexity Ledger

This dataset contains instance-level output predictions and calculated complexity metrics from the models trained in Cook et al. (2025). Models are trained in text classification for the 4 tasks / dependent variables in the FairPsych data from Abbasi et al. (2021), which are noted in the 'score_var' column. Code for data loading / ID creation and downstream analysis can be found the paper's GitHub repository under utils/data_loading_no_demo.py and artifact_code/, respectively. Data for the Depression task is not included in this export per the IRB for its collection; please reach out to the authors to discuss additional data and code availability.