# The Unbearable Weight of Massive Privilege: Revisiting Bias-Variance Trade-Offs in the Context of Fair Prediction

arXiv: https://arxiv.org/abs/2302.08704 · HTML: https://mlsanity.com/paper/2302.08704/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 5 of its 9 references that are in the map.
- First posted on arXiv 17 Feb 2023.

## Builds on (most cited first, 5 of 5)

- [Equality of Opportunity in Supervised Learning](https://mlsanity.com/paper/1610.02413/) (2016)
- [Fair prediction with disparate impact: A study of bias in recidivism prediction instruments](https://mlsanity.com/paper/1610.07524/) (2016)
- [Inherent Trade-Offs in the Fair Determination of Risk Scores](https://mlsanity.com/paper/1609.05807/) (2016)
- [Retiring Adult: New Datasets for Fair Machine Learning](https://mlsanity.com/paper/2108.04884/) (2021)
- [FairPrep: Promoting Data to a First-Class Citizen in Studies on Fairness-Enhancing Interventions](https://mlsanity.com/paper/1911.12587/) (2019)

## Built on by (most cited first, 1 of 1)

- [Decision-centric fairness: Evaluation and optimization for resource allocation problems](https://mlsanity.com/paper/2504.20642/) (2025)

## Abstract (arXiv)

In this paper we revisit the bias-variance decomposition of model error from the perspective of designing a fair classifier: we are motivated by the widely held socio-technical belief that noise variance in large datasets in social domains tracks demographic characteristics such as gender, race, disability, etc. We propose a conditional-iid (ciid) model built from group-specific classifiers that seeks to improve on the trade-offs made by a single model (iid setting). We theoretically analyze the bias-variance decomposition of different models in the Gaussian Mixture Model, and then empirically test our setup on the COMPAS and folktables datasets. We instantiate the ciid model with two procedures that improve "fairness" by conditioning out undesirable effects: first, by conditioning directly on sensitive attributes, and second, by clustering samples into groups and conditioning on cluster membership (blind to protected group membership). Our analysis suggests that there might be principled procedures and concrete real-world use cases under which conditional models are preferred, and our striking empirical results strongly indicate that non-iid settings, such as the ciid setting proposed here, might be more suitable for big data applications in social contexts.