Some general advice sought from the Stata community.
Say I'm doing a Difference in Differences exercise on N states or regions but only 1 of those states has been subject to the treatment. The other N-1 have not.
Inference is problematic presumably when -- even though there may be many cluster (states here) -- only 1 of them is treated.
Is there a standard way (or ways) to deal with such circumstances? Thanks.
0 Response to Clustering under few treated clusters
Post a Comment