Research map / Learning theory and algorithms / Fairness
Social Bias Meets Data Bias: The Impacts of Labeling and Measurement Errors on Fairness Criteria
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Train a DP-constrained classifier using labels that underestimate qualification for the disadvantaged group | Fairness-constrained classifier · Disadvantaged-group label underestimation | DP fairness | No clear effect |
| Train a TPR-constrained classifier using labels that underestimate qualification for the disadvantaged group | Fairness-constrained classifier · Disadvantaged-group label underestimation | TPR fairness | No clear effect |
| Train an FPR-constrained classifier using labels that underestimate qualification for the disadvantaged group | Fairness-constrained classifier · Disadvantaged-group label underestimation | FPR fairness violation | Hurt |
| Use DP rather than an unconstrained maximum-utility policy under disadvantaged-group label underestimation | Decision policy · Biased training data | firm utility | Mixed |
| Use TPR rather than DP under disadvantaged-group label underestimation | Fairness-constrained classifier · Disadvantaged-group label underestimation | threshold sensitivity | Helped |
| Train a DP-constrained classifier using labels that overestimate qualification for the advantaged group | Fairness-constrained classifier · Advantaged-group label errors | DP fairness | No clear effect |
| Train an FPR-constrained classifier using labels that overestimate qualification for the advantaged group | Fairness-constrained classifier · Advantaged-group label errors | FPR fairness | No clear effect |
| Train a TPR-constrained classifier using labels that overestimate qualification for the advantaged group | Fairness-constrained classifier · Advantaged-group label errors | TPR fairness violation | Hurt |
| Measure only qualified-agent features incorrectly while leaving unqualified-agent features unchanged | Fairness-constrained classifier · Disadvantaged-group feature measurement errors | FPR fairness | No clear effect |
| Measure only qualified-agent features incorrectly while leaving unqualified-agent features unchanged | Fairness-constrained classifier · Disadvantaged-group feature measurement errors | DP and TPR fairness | Hurt |
| Impose a fairness constraint when training on biased data | Logistic regression classifier · Adult and German credit | test accuracy | Mixed |
| Use a fairness-constrained policy instead of maximum utility under severe training-data bias | Firm decision policy · FICO credit score | firm utility | Mixed |
Closest papers by meaning
- The Fragility of Fairness: Causal Sensitivity Analysis for Fair Machine Learning (NeurIPS 2024 · similarity 0.89)
- Understanding challenges to the interpretation of disaggregated evaluations of algorithmic fairness (NeurIPS 2025 · similarity 0.89)
- Reconciling Predictive and Statistical Parity: A Causal Approach (AAAI 2024 · similarity 0.89)
- Arbitrariness and Social Prediction: The Confounding Role of Variance in Fair Classification (AAAI 2024 · similarity 0.88)
- Fairness-Accuracy Trade-Offs: A Causal Perspective (AAAI 2025 · similarity 0.88)
- Mind the Gap: A Causal Perspective on Bias Amplification in Prediction & Decision-Making (NeurIPS 2024 · similarity 0.88)
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.