Challenges of the inconsistency regime

Novel debiasing methods for missing data models

Michael Celentano, Martin J. Wainwright

This repository contains R code to reproduce the figures in "Challenges of the inconsistency regime: novel debiasing methods for missing data models." Standard semi-parametric approaches for estimating population means when data is missing at random, or relatedly, average treatment effects, include outcome imputation, augmented inverse probability weighting (AIPW), and inverse probability weighting (IPW)/Horvitz-Thompson. These methods are inconsistent when both the outcome model and missingness/propensity model cannot be estimated consistently. We develop methods which achieve consistency for the population mean in a setting where neither the outcome nor missingness/propensity model can be estimated consistently.

Guide to paper's figures

• Figures 1-3: These plot data simulated by simulations/standard_estimators.R. The figures are produced by plots_paper/make_plots_standard_estimators.R.
• Figure 4: This plots data simulated by simulations/novel_estimators.R. The figure is produced by plots_paper/make_plots_novel_estimators.R.
• Figure 5 & 6: these plot data simulated by simulations/lambda_dependence.R. The figures are produced by plots_paper/make_plots_lambda_depdences.R.

The files in the utils directory define functions used by the simulations.

Figures are output to a directory fig_paper and simulation data is written to and read from a directory data. Simulation data used in the paper can be found here.