Does anybody know how to perform a delete-a-group jackknife variance estimate, on a stratified sample, in SAS?
I was looking into PROC SURVEYMEANS.
If the sample design involves stratification, each stratum must have at least two PSUs to use the jackknife method.
Let stratum h_n be the stratum from which a PSU is deleted for the n-th replicate. Stratum h_n is called the donor stratum."
Sounds like you could define each of your groups to be a PSU, but from the documentation it sounds like they are assuming that a given PSU will be contained in only one strata, whereas each group for delete-a-group jackknife preferably cuts across all strata....