Detecting and Mitigating Group Bias in Heterogeneous Treatment Effects

2026-02-23

AI Summary

This study addresses a systematic bias—termed “group bias”—that arises in heterogeneous treatment effect (HTE) modeling when aggregating individual conditional average treatment effects (CATE) to estimate group-level average treatment effects (GATE). The authors formally define this bias for the first time and develop a unified statistical framework that enables asymptotically normal bias measurement and hypothesis testing. They further propose a shrinkage-based bias correction method, yielding a computationally tractable closed-form solution for optimal adjustment under minimal assumptions. The approach efficiently detects and corrects group bias with high accuracy. Empirical evaluation on large-scale digital platform experiments demonstrates substantial bias reduction and improved decision-making performance in personalized intervention strategies, particularly in profit-maximization settings where it enhances both policy accuracy and real-world returns.

Technology Category

• Machine Learning: Ethics, Bias, and Fairness
• Computer Vision: Bias, Fairness & Privacy
• Philosophy and Ethics of AI: Bias, Fairness & Equity

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems
• Social Networks and Social Media: Fairness and bias in social network and social media analysis

Abstract

Heterogeneous treatment effects (HTEs) are increasingly estimated using machine learning models that produce highly personalized predictions of treatment effects. In practice, however, predicted treatment effects are rarely interpreted, reported, or audited at the individual level but, instead, are often aggregated to broader subgroups, such as demographic segments, risk strata, or markets. We show that such aggregation can induce systematic bias of the group-level causal effect: even when models for predicting the individual-level conditional average treatment effect (CATE) are correctly specified and trained on data from randomized experiments, aggregating the predicted CATEs up to the group level does not, in general, recover the corresponding group average treatment effect (GATE). We develop a unified statistical framework to detect and mitigate this form of group bias in randomized experiments. We first define group bias as the discrepancy between the model-implied and experimentally identified GATEs, derive an asymptotically normal estimator, and then provide a simple-to-implement statistical test. For mitigation, we propose a shrinkage-based bias-correction, and show that the theoretically optimal and empirically feasible solutions have closed-form expressions. The framework is fully general, imposes minimal assumptions, and only requires computing sample moments. We analyze the economic implications of mitigating detected group bias for profit-maximizing personalized targeting, thereby characterizing when bias correction alters targeting decisions and profits, and the trade-offs involved. Applications to large-scale experimental data at major digital platforms validate our theoretical results and demonstrate empirical performance.

Problem

• heterogeneous treatment effects
• group bias
• conditional average treatment effect
• group average treatment effect
• aggregation bias

Innovation

• group bias
• heterogeneous treatment effects
• causal inference
• bias correction
• shrinkage estimation