Use of clubsandwich over sandwich

[MikeJSeo]

clubsandwich seems to be for clustered data (i.e. matched samples). I'm not sure using this vcovCR is any better for IPTW related.

If you look at phillippo's appendix page 8
he uses vcovHC from sandwich package

In the weightit package (package for IPTW), they mention using robust standard error using sandwich package, but not cluster robust standard error.

In the matching methods package created by same author matchit, they mention cluster robust standard error. From the description, it seems that when using matching methods, cluster robust standard errors is better than robust standard errors.