[R] correct standard errors (heteroskedasticity) using survey design

[jour4life]

Hello all,

I'm hoping someone can help  clarify how the survey design method works in
R. I currently have a data set that utilized a complex survey design. The
only thing is that only the weight is provided. Thus, I constructed my
survey design as:

svdes<-svydesign(id=~1, weights=~weightvar, data=dataset)

Then, I want to run an OLS model, so:

fitsurv<-svyglm(y~x1+x2+x3...xk, design=svdes, data=dataset)

But, I want to check if there is evidence of heteroskedasticity. If so, how
would I correct the standard errors? Can the "sandwich" library do this? Are
the standard errors already adjusted. How else can I verify if
heteroskedasticity is still present? Can I still use the bptest()?

I read  an earlier post where someone used a dataset example entitled
"banco." But, her dataset included strata and cluster variables. Someone
responded that the "sandwich" library already adjusted for clustering. In my
situation, however, I only have a weight variable. 

I hope someone can clarify this problem for me.

Thanks,

Carlos

--
View this message in context: http://r.789695.n4.nabble.com/correct-standard-errors-heteroskedasticity-using-survey-design-tp4560122p4560122.html
Sent from the R help mailing list archive at Nabble.com.