Conformal Sensitivity Analysis for Individual Treatment Effects

DOI10.1080/01621459.2022.2102503arXiv2112.03493MaRDI QIDQ6153976FDOQ6153976

Authors: Yixin Wang, David M. Blei 

Publication date: 19 March 2024

Published in: Journal of the American Statistical Association (Search for Journal in Brave)

Abstract: Estimating an individual treatment effect (ITE) is essential to personalized decision making. However, existing methods for estimating the ITE often rely on unconfoundedness, an assumption that is fundamentally untestable with observed data. To assess the robustness of individual-level causal conclusion with unconfoundedness, this paper proposes a method for sensitivity analysis of the ITE, a way to estimate a range of the ITE under unobserved confounding. The method we develop quantifies unmeasured confounding through a marginal sensitivity model [Ros2002, Tan2006], and adapts the framework of conformal inference to estimate an ITE interval at a given confounding strength. In particular, we formulate this sensitivity analysis problem as a conformal inference problem under distribution shift, and we extend existing methods of covariate-shifted conformal inference to this more general setting. The result is a predictive interval that has guaranteed nominal coverage of the ITE, a method that provides coverage with distribution-free and nonasymptotic guarantees. We evaluate the method on synthetic data and illustrate its application in an observational study.

Full work available at URL: https://arxiv.org/abs/2112.03493

zbMATH Keywords

uncertainty quantificationpredictive inferenceunconfoundednessdistribution shift

Mathematics Subject Classification ID

Statistics (62-XX)

Cites Work

• On-line predictive linear regression
• BART: Bayesian additive regression trees
• Causal inference by using invariant prediction: identification and confidence intervals. With discussion and authors' reply
• Random forests
• A Distributional Approach for Causal Inference Using Propensity Scores
• Adjusting for Nonignorable Drop-Out Using Semiparametric Nonresponse Models
• Estimation and Inference of Heterogeneous Treatment Effects using Random Forests
• Observational studies.
• The central role of the propensity score in observational studies for causal effects
• Statistics and Causal Inference
• Predictive inference with the jackknife+
• Distribution-free predictive inference for regression
• Title not available (Why is that?)
• A conformal prediction approach to explore functional data
• A tutorial on conformal prediction
• Distribution-free Prediction Bands for Non-parametric Regression
• Covariate Balancing Propensity Score
• Confounding and collapsibility in causal inference
• Title not available (Why is that?)
• On the distinction between the conditional probability and the joint probability approaches in the specification of nearest-neighbour systems
• Calibrating sensitivity analyses to observed covariates in observational studies
• Sensitivity analysis for inverse probability weighting estimators via the percentile bootstrap
• Flexible sensitivity analysis for observational studies without observable implications
• The limits of distribution-free conditional predictive inference
• Making sense of sensitivity: extending omitted variable bias
• Conformal Inference of Counterfactuals and Individual Treatment Effects

This page was built for publication: Conformal Sensitivity Analysis for Individual Treatment Effects

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6153976)