Transferring Causal Effects using Proxies

2025-10-29

AI Summary

This paper addresses causal effect estimation under unobserved confounding in a multi-domain setting, where causal effects are heterogeneous across domains, treatment and outcome variables are continuous, and observed variables—including proxies—are discrete or categorical. Leveraging the assumption that latent confounder structure is identifiable via proxy variables, we establish the first nonparametric identifiability of causal effects in a target domain where only proxies—not the confounders themselves—are observed. We propose two consistent estimators, derive their asymptotic normality, and construct valid confidence intervals. Our theoretical results are validated through simulation studies and applied to an empirical analysis of how website rankings affect consumer choice, enabling cross-domain causal transfer and precise inference.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Application Domains: Humanities & Computational Social Science

Application Category

• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs

Abstract

We consider the problem of estimating a causal effect in a multi-domain setting. The causal effect of interest is confounded by an unobserved confounder and can change between the different domains. We assume that we have access to a proxy of the hidden confounder and that all variables are discrete or categorical. We propose methodology to estimate the causal effect in the target domain, where we assume to observe only the proxy variable. Under these conditions, we prove identifiability (even when treatment and response variables are continuous). We introduce two estimation techniques, prove consistency, and derive confidence intervals. The theoretical results are supported by simulation studies and a real-world example studying the causal effect of website rankings on consumer choices.

Problem

• Estimating causal effects with unobserved confounders across domains
• Using proxy variables to identify hidden confounding factors
• Transferring causal effects when treatment and outcome are continuous

Innovation

• Using proxy variables for hidden confounders
• Developing multi-domain causal effect estimation
• Proposing identifiability under continuous variables