Reading Between the Lines: Deconfounding Causal Estimates using Text Embeddings and Deep Learning

TL;DR

This paper introduces a deep learning-based framework that uses text embeddings to improve causal effect estimation in observational studies by effectively capturing unobserved confounders from unstructured text data.

Contribution

It presents a novel neural network-enhanced double machine learning approach that leverages text embeddings for causal inference, outperforming traditional methods in synthetic benchmarks.

Findings

Deep learning reduces bias to near zero in causal estimates.

Text embeddings capture unobserved confounders absent from structured data.

Standard tree-based estimators retain significant bias with unstructured text.

Abstract

Estimating causal treatment effects in observational settings is frequently compromised by selection bias arising from unobserved confounders. While traditional econometric methods struggle when these confounders are orthogonal to structured covariates, high-dimensional unstructured text often contains rich proxies for these latent variables. This study proposes a Neural Network-Enhanced Double Machine Learning (DML) framework designed to leverage text embeddings for causal identification. Using a rigorous synthetic benchmark, we demonstrate that unstructured text embeddings capture critical confounding information that is absent from structured tabular data. However, we show that standard tree-based DML estimators retain substantial bias (+24%) due to their inability to model the continuous topology of embedding manifolds. In contrast, our deep learning approach reduces bias to -0.86% with optimized architectures, effectively recovering the ground-truth causal parameter. These findings suggest that deep learning architectures are essential for satisfying the unconfoundedness assumption when conditioning on high-dimensional natural language data