How to disentangle the total causal effect of a continuous treatment on an outcome into natural direct and indirect effects using the generalized propensity score?
Abstract
Hsu et al. (2018) propose in this working paper semi- and nonparametric methods for disentangling the total causal effect of a continuous treatment on an outcome variable into its natural direct effect and the indirect effect that operates through one or several intermediate variables or mediators.
Hsu et al. (2018) have an approach which is based on weighting observations by the inverse of two versions of the generalized propensity score (GPS), namely the conditional density of treatment either given observed covariates or given covariates and the mediator.
Hsu et al. (2018) show that the effect estimators are asymptotically normal when the GPS is estimated by either a parametric or a nonparametric kernel-based method. A simulation study and an application to the Job Corps program is also provided.
References
Citation
@online{hsu2018,
author = {Hsu, Yu-Chin and Huber, Martin and Lee, Ying-Ying and Pipoz,
Layal},
title = {Direct and Indirect Effects of Continuous Treatments Based on
Generalized Propensity Score Weighting},
date = {2018-06-01},
url = {https://folia.unifr.ch/unifr/documents/306658},
langid = {en}
}