I have a main question about whether to specify ddf
in confint.svyglm()
. Using the two specifications below generate slightly different confidence intervals. Specifically, my questions are:
- If it would be preferable to use
ddf = degf(m$survey.design)
? - A secondary question is: which method is preferable,
Wald
orlikelihood
?
Any thoughts would be greatly appreciated!
library(survey)
data(api)
dclus2<-svydesign(id=~dnum+snum, fpc=~fpc1+fpc2, data=apiclus2)
m<-svyglm(I(comp.imp=="Yes")~stype*emer+ell, design=dclus2, family=quasibinomial)
confint(m, method="Wald", parm=c("ell","emer"), ddf = NULL )
confint(m, method="Wald", parm=c("ell","emer"), ddf = degf(m$survey.design ) )
confint(m, method="like", parm=c("ell","emer"), ddf = NULL )
confint(m, method="like", parm=c("ell","emer"), ddf = degf(m$survey.design ) )