● Frequently Asked Questions about the Stata Command  "robustate":

Q1. How does the  "robustate"  command compare with the existing IPW estimator such as the  "teffects ipw"  command?

• A. "teffects ipw"  tends to produce larger standard errors than  "robustate". If the overlap is severely limited (i.e., if the tail index of the inverse propensity score is above 0.5), then the standard error for  "teffects ipw"  is not guaranteed to exist while that of  "robustate"  still exists. 

Q2. How does the  "robustate"  command compare with the IPW estimation with trimming/truncating small propensity scores?

• A. Trimmed and truncated estimators are biased for the average treatment effects (ATE), while the  "robustate"  estimator is de-biased and its standard error accounts for the effects of the de-biasing.

Q3. How does the  "robustate"  command compare with the matching estimators such as  "teffects pamatch"  and  "teffects nnmatch"  commands?

• A. The matching estimators tend to be biased for the average treatment effects (ATE) when the overlap is limited, while the  "robustate"  estimator being de-biased consistently estimates the ATE and its standard error accounts for the effects of the de-biasing.

Q4. How does the  "robustate"  command compare with the overlap weighting approaches?

• A. The  "robustate"  estimates the average treatment effects (ATE), while the overlap weighting approaches estimate only weighted averages of treatment effects and hence in general fail to estimate the ATE.

See further details about the Stata Command  "robustate".

● October 2021: Cluster robust double machine learning package now available in R and Python - thanks to Malte S. Kurz.