CausalSpline: Nonlinear Causal Dose-Response Estimation via Splines

Read full text Summarise with VIRAT
Authors: Subir Hait
Cited by: 0
References: 4

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Estimates nonlinear causal dose-response functions for continuous treatments using spline-based methods under standard causal assumptions (unconfoundedness / ignorability). Implements three identification strategies: Inverse Probability Weighting (IPW) via the generalised propensity score (GPS), G-computation (outcome regression), and a doubly-robust combination. Natural cubic splines and B-splines are supported for both the exposure-response curve f(T) and the propensity nuisance model. Pointwise confidence bands are obtained via the sandwich estimator or nonparametric bootstrap. Also provides fragility diagnostics including pointwise curvature-based fragility, uncertainty-normalised fragility, and regional integration over user-defined treatment intervals. Builds on the framework of Hirano and Imbens (2004) for continuous treatments and extends it to fully nonparametric spline estimation.

Fields of Study & Subjects

Advanced Causal Inference Techniques Statistical Methods and Inference Statistical Methods and Bayesian Inference

Publication details

Publication date
25 March 2026
Language
EN
Document type
Dataset