Skip to contents

drcmd

Doubly Robust Causal Inference with Missing Data

Authors: Keith Barnatchez and Griffin DesRoches

drcmd is an R package for implementing doubly-robust estimators of counterfactual means in the presence of general missing data patterns. drcmd leverages links between influence curves for counterfactual means under no missingness, and the influence curve corresponding to the missingness pattern in the user-supplied data. Detailed discussion of the theoretical details behind the methods used in drcmd can be found in Kennedy (2016), Tsiatis (2006), and van der Laan and Robins (2003).

Users can fit nuisance functions through Super Learner (a stacking algorithm).

Please see the package vignette and documentation for further details.

Installation 

devtools::install_github('keithbarnatchez/drcmd')
## Example
## Citation
Kennedy, E. H. (2016). Semiparametric theory and empirical processes in causal inference. Statistical causal inferences and their applications in public health research, 141-167.
Tsiatis, A. A. (2006). Semiparametric theory and missing data (Vol. 4). New York: Springer.
van der Laan, M. J., & Robins, J. M. (2003). Unified methods for censored longitudinal data and causality. Springer New York.

Contributing, reporting issues