Consistent Labeling Across Group Assignments: Variance Reduction in Conditional Average Treatment Effect Estimation
Abstract
Numerous algorithms have been developed for Conditional Average Treatment Effect (CATE) estimation. In this paper, we first highlight a common issue where many algorithms exhibit inconsistent learning behavior for the same instance across different group assignments. We introduce a metric to quantify and visualize this inconsistency. Next, we present a theoretical analysis showing that this inconsistency indeed contributes to higher test errors and cannot be resolved through conventional machine learning techniques. To address this problem, we propose a general method called \textbf{Consistent Labeling Across Group Assignments} (CLAGA), which eliminates the inconsistency and is applicable to any existing CATE estimation algorithm. Experiments on both synthetic and real-world datasets demonstrate significant performance improvements with CLAGA.
Cite
@article{arxiv.2507.04332,
title = {Consistent Labeling Across Group Assignments: Variance Reduction in Conditional Average Treatment Effect Estimation},
author = {Yi-Fu Fu and Keng-Te Liao and Shou-De Lin},
journal= {arXiv preprint arXiv:2507.04332},
year = {2025}
}