Estimating individual treatment effects under unobserved confounding using binary instruments

Dennis Frauen, Stefan Feuerriegel

Keywords: Causal machine learning, treatment effect estimation, instrumental variables

TL;DR: We propose a multiple robust machine learning framework for estimating individual treatment effects using binary instrumental variables.

Abstract: Estimating individual treatment effects (ITEs) from observational data is relevant in many fields such as personalized medicine. However, in practice, the treatment assignment is usually confounded by unobserved variables and thus introduces bias. A remedy to remove the bias is the use of instrumental variables (IVs). Such settings are widespread in medicine (e.g., trials where compliance is used as binary IV). In this paper, we propose a novel, multiple robust machine learning framework, called MRIV, for estimating ITEs using binary IVs and thus yield an unbiased ITE estimator. Different from previous work for binary IVs, our framework estimates the ITE directly via a pseudo outcome regression. (1) We provide a theoretical analysis where we show that our framework yields multiple robust convergence rates: our ITE estimator achieves fast convergence even if several nuisance estimators converge slowly. (2) We further show that our framework asymptotically outperforms state-of-the-art plug-in IV methods for ITE estimation. (3) We build upon our theoretical results and propose a tailored neural network architecture called MRIV-Net for ITE estimation using binary IVs. Across various computational experiments, we demonstrate empirically that our \modelname achieves state-of-the-art performance. To the best of our knowledge, our MRIV is the first multiple robust machine learning framework tailored to estimating ITEs in the binary IV setting.

Community Implementations: [ 1 code implementation](https://www.catalyzex.com/paper/estimating-individual-treatment-effects-under/code)