Degrees-of-freedom adjustment for FEs in complex survey data

Hello,

I would highly appreciate if you could help me with the following problem:

Since the svy command does not support areg, I want to manually demean my variables (I have more than 35'000 groups, so I cannot use individual dummy variables). I found this previous thread in which the appropriate degrees-of-freedom adjustment for the right standard errors is discussed: http://www.stata.com/statalist/archi.../msg00652.html

While this adjustment also works if I include sample weights, I get wrong standard errors if try to adjust for clusters and stratification. Does thus anybody know a similar degrees-of-freedom adjustment for complex survey data?

Thanks,


Patrick