Abstract:Abstract This article proposes a class of M-estimators that double weight for the joint problems of nonrandom treatment assignment and missing outcomes. Identification of the main parameter of interest is achieved under unconfoundedness and missing at random assumptions with respect to the treatment and sample selection problems, respectively. Given the parametric framework, the asymptotic theory of the proposed estimator is outlined in two parts: first, when the parameter solves an unconditional problem, and second, when it solves a stronger conditional problem. The two parts help to summarize the misspecification scenarios permissible under the given framework and the role played by double weighting in each. As illustrative examples, the article also discusses the estimation of causal parameters like average and quantile treatment effects. With respect to the average treatment effect, this article shows that the proposed estimator is doubly robust . Finally, a detailed application to Calónico and Smith’s (The women of the national supported work demonstration. J Labor Econom. 2017;35(S1):S65–S97.) reconstructed sample from the National Supported Work training program is used to demonstrate the estimator’s performance in empirical settings.

What problem does this paper attempt to address?

This paper attempts to address the identification challenges in causal inference due to nonrandom treatment assignment and missing outcomes. Specifically: 1. **Nonrandom treatment assignment**: In observational studies, the selection of individuals to enter a program or intervention is often not random, which can lead to bias in the estimation of causal effects. 2. **Missing outcomes**: If the observed outcomes of interest are missing due to loss to follow - up or non - response, this will further increase the difficulty of estimating causal effects. To solve these two problems, the author proposes a class of doubly weighted M - estimators. These estimators improve the accuracy and consistency of the estimation by simultaneously considering nonrandom treatment assignment and missing outcomes. The main contributions include: - **Doubly weighted**: Use the propensity score and the probability of missing outcomes for weighting simultaneously to identify the true parameters. - **Asymptotic theory**: Propose the asymptotic theory of doubly weighted estimators, which is divided into two parts: - Part 1: When the parameter solves an unconditional problem, describe the potential model misspecification situation. - Part 2: When the parameter solves a stronger conditional problem, allow the weights to be possibly misspecified. - **Doubly robustness**: For the Average Treatment Effect (ATE), the proposed estimator is doubly robust. That is, if either the conditional mean model or the weights is correctly specified, the estimator is still consistent. - **Application example**: Through the application to the sample of the National Supported Work training program reconstructed by Calónico and Smith, show the performance of the estimator in actual data. In conclusion, this paper aims to provide a comprehensive method to deal with the problems of nonrandom treatment assignment and missing outcomes, especially in causal inference. By introducing doubly weighted M - estimators, the author not only improves the consistency and accuracy of the estimation but also provides theoretical guarantees and practical application examples.

Doubly weighted M-estimation for nonrandom assignment and missing outcomes

Doubly robust estimation of multivariate fractional outcome means with multivalued treatments

Double robust semiparametric weighted M-estimators of a zero-inflated Poisson regression with missing data in covariates

Double robust estimation of functional outcomes with data missing at random

Doubly robust treatment effect estimation with missing attributes

Robust propensity score weighting estimation under missing at random

Multiply robust estimation of the average treatment effect with missing outcomes

A weighting method for simultaneous adjustment for confounding and joint exposure-outcome misclassifications

Double robust estimation of partially adaptive treatment strategies

Marginal mean weighting through stratification: A generalized method for evaluating multivalued and multiple treatments with nonexperimental data.

Doubly Robust Estimation in Causal Inference with Missing Outcomes: with an Application to the Aerobics Center Longitudinal Study

A Unified Framework for Efficient Estimation of General Treatment Models

On Varieties of Doubly Robust Estimators under Missingness Not at Random with a Shadow Variable

Identification, Doubly Robust Estimation, and Semiparametric Efficiency Theory of Nonignorable Missing Data With a Shadow Variable

High-dimensional model-assisted inference for treatment effects with multi-valued treatments

Joint calibrated estimation of inverse probability of treatment and censoring weights for marginal structural models

A nonparametric super-efficient estimator of the average treatment effect

Doubly Robust Inference in Causal Latent Factor Models

Investigating an Alternative for Estimation from a Nonprobability Sample: Matching plus Calibration

Doubly Robust Semiparametric Difference-in-Differences Estimators with High-Dimensional Data