Conceptualizing Treatment Leakage in Text-based Causal Inference

TL;DR

This paper explores the problem of treatment leakage in text-based causal inference, demonstrating how it biases treatment effect estimates and proposing a text distillation method to mitigate this issue.

Contribution

It formally defines treatment leakage in text-based causal inference, analyzes its impact, and introduces a pre-processing technique called text distillation to address the bias.

Findings

Treatment leakage causes bias in ATE estimates.

Text distillation reduces bias caused by treatment leakage.

Simulation results confirm effectiveness of the proposed method.

Abstract

Causal inference methods that control for text-based confounders are becoming increasingly important in the social sciences and other disciplines where text is readily available. However, these methods rely on a critical assumption that there is no treatment leakage: that is, the text only contains information about the confounder and no information about treatment assignment. When this assumption does not hold, methods that control for text to adjust for confounders face the problem of post-treatment (collider) bias. However, the assumption that there is no treatment leakage may be unrealistic in real-world situations involving text, as human language is rich and flexible. Language appearing in a public policy document or health records may refer to the future and the past simultaneously, and thereby reveal information about the treatment assignment. In this article, we define the treatment-leakage problem, and discuss the identification as well as the estimation challenges it raises. Second, we delineate the conditions under which leakage can be addressed by removing the treatment-related signal from the text in a pre-processing step we define as text distillation. Lastly, using simulation, we show how treatment leakage introduces a bias in estimates of the average treatment effect (ATE) and how text distillation can mitigate this bias.