The package implements the methods described in Ghosh et al. (2020): Sufficient Dimension Reduction for Feasible and Robust Estimation of Average Causal Effect. It uses a semiparametric locally efficient dimension reduction approach to assess both the treatment assignment mechanism and the average responses in both treated and nontreated groups. It then integrates all results through imputation, inverse probability weighting and doubly robust augmentation estimators.
Keyworkds: Average Treatment Effect, Doubly Robust Estimator, Efficiency, Inverse Probability Weighting, sufficient dimension reduction
The SDRcausal package can be found here: github.com/stat4reg/SDRcausal
To install and load this package in R from GitHub, run the following commands:
(all feedback welcome: mohammad.ghasempour AT umu.se)
Note to Mac user: Parallelization does not work because OpenMP is not available on Mac OS, and thus only one thread can be used. Mac users may need to run the following command in a terminal window before installing the package:
Ghosh, T., Ma, Y. and de Luna, X. (2020). Sufficient Dimension Reduction for Feasible and Robust Estimation of Average Causal Effect. Statistica Sinica. On-line ahead of print. DOI: 10.5705/ss.202018.0416. ArXiv version with Supplementary material: arXiv:1811.01992.
Ghasempour, M. and de Luna, X. (2021). SDRcausal: an R package for causal inference based on sufficient dimension reduction. arXiv:2105.02499.