Code to replicate the simulation study in the paper "Calibrating doubly-robust estimators with unbalanced treatment assignment".
The code implements two different simulations strategies:
- Simulation 1: Synthetic DGP
- Simulation 2: Empirical Monte Carlo Study
When the argument --data_path
is None
, the code will run the synthetic DGP simulation. Otherwise, it will run the empirical Monte Carlo study. The data used in the empirical Monte Carlo study is obtained from https://doi.org/10.23662/FORS-DS-1203-1.