Hi
all,
Is it worthwhile to
use the formulas of a stratified two stage sample when computing a
confidence interval for a proportion? Or can one be satisfied by stating that
"stratification compensates multistage sampling, since the former yields a
smaller standard error and the latter yields a bigger standard error than in
the case of a simple random sample" and therefore just use the formulas of
a simple random sample?
If it is worthwhile
to use the extended formulas, can anyone give me some references of books or
papers with these formulas (for proportions)?
Thanks,
Ward