Cluster-robust estimation

[gcasamat]

Hi, the causal_forest function of the R grf package has a very useful option 'clusters', that allows for cluster-robust estimation. I wonder if there is any plan to implement this feature in CausalForestDML ? Thanks

[vsyrgkanis]

There is. But it might take a couple months unfortunately

[vsyrgkanis]

For now you should definitely use the keyword argument “groups” at fit time where a group is the same as a cluster

[vsyrgkanis]

This will at least perform crossfitting in the right manner and in the future this same argument will be used in the causal forest splitting criterion to perform cluster based sub-sampling and sample-splitting, and also to perform cluster robust doubly robust ATE intervals

[gcasamat]

Thank you. I will thus wait for the implementation of cluster-robust estimation. And in the meantime, I will try the 'groups' argument, as you suggest.

[liaochris]

Hi! Wondering if there are any updates on the implementation of cluster-robust estimation for causal forests