Research map / Learning theory and algorithms / Fairness
The Illusion of Fairness: Auditing Fairness Interventions in Algorithmic Hiring with Audit Studies
Disa Sariola, Patrick Button, Aron Culotta, Nicholas Mattei
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| evaluate EBR on a test set with ITE label-bias repair instead of the raw test set (ablation) | EBR-trained random forest · Hiring audit data (original) | FPRD: -0.031 -> 0.042 | Hurt |
| apply ITE label repair to both training and test data instead of EBR training | random forest classifier · Hiring audit data (original) | FPRD: 0.042 -> 0.017 | Improved |
| apply ITE label repair to training and test data | random forest classifier · Hiring audit data (original) | AUC: roughly 1% point decrease | Hurt |
| use a more flexible model (MLP instead of RF) under EBR Train - ITE Test (ablation) | EBR Train - ITE Test · Hiring audit data (original) | FPRD: 0.042 -> 0.095 | Hurt |
| double the human discrimination in the audit data (ablation) | EBR Train - ITE Test (RF) · Hiring audit data (doubled discrimination) | FPRD: 0.042 -> 0.133 | Hurt |
| double the human discrimination in the audit data (ablation) | ITE Train & Test (RF) · Hiring audit data (doubled discrimination) | FPRD: 0.017 -> 0.049 | Hurt |
| ITE repair under doubled discrimination | random forest classifier · Hiring audit data (doubled discrimination) | AUC: roughly 2.1% points decrease | Hurt |
| add selection bias (Spanish disparity 0.8) to the data (ablation) | EBR with unmodified vs ITE-repaired test (RF) · Hiring audit data (RF, Spanish disparity 0.8) | FPRD: -.08 vs 0.145 | Hurt |
| add selection bias (Spanish disparity 0.8) to the data (ablation) | EBR with unmodified vs ITE-repaired test (MLP) · Hiring audit data (MLP, Spanish disparity 0.8) | FPRD: -0.11 -> 0.27 | Hurt |
| ITE Train & Test under extreme selection bias (ablation) | classifier · Hiring audit data (Spanish disparity 0.8) | FPRD | Mixed |
Closest papers by meaning
- FairJob: A Real-World Dataset for Fairness in Online Systems (NeurIPS 2024 · similarity 0.87)
- Cost Efficient Fairness Audit Under Partial Feedback (NeurIPS 2026 · similarity 0.86)
- FairPFN: A Tabular Foundation Model for Causal Fairness (ICML 2025 · similarity 0.86)
- Understanding challenges to the interpretation of disaggregated evaluations of algorithmic fairness (NeurIPS 2025 · similarity 0.86)
- Fairness-Accuracy Trade-Offs: A Causal Perspective (AAAI 2025 · similarity 0.86)
- Identifying Reasons for Bias: An Argumentation-Based Approach (AAAI 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
