A Stable and Efficient Covariate-Balancing Estimator for Causal Survival Effects

TL;DR

This paper introduces a new covariate-balancing estimator for causal survival analysis that is both stable and efficient, overcoming issues with small probability inverses in existing methods.

Contribution

It presents a novel covariate-balancing approach that improves stability and efficiency in estimating survival causal effects under conditionally-independent censoring.

Findings

The method is empirically stable in synthetic and semi-synthetic experiments.

It achieves asymptotic efficiency in estimating survival causal effects.

The approach addresses the problem of error amplification in existing nonparametric methods.

Abstract

We propose an empirically stable and asymptotically efficient covariate-balancing approach to the problem of estimating survival causal effects in data with conditionally-independent censoring. This addresses a challenge often encountered in state-of-the-art nonparametric methods: the use of inverses of small estimated probabilities and the resulting amplification of estimation error. We validate our theoretical results in experiments on synthetic and semi-synthetic data.