While estimating from the survey data involving stratification & clustering survey design and using survey package of r, is it possible to estimate at the cluster level? For eg; for following survey design:
data(api) ## syntax for stratified cluster sample dclus=svydesign(id=~dnum, strata=~stype, weights=~pw, data=apistrat, nest=TRUE)
This is an example which is reproduced from the survey package. Here, dnum is district, stype is Elementary/Middle/High School and pw is weight (inverse of probalitlity of the students selected from each of the strata). In this case, can I estimate population parameter at district level? For example, to estimate total enrollment for each of the district:
svyby(~enroll,~dnum, design=dclus, svytotal)
I got the following output:
dnum enroll se 19 21751.32 21751.32 20 10494.50 10494.50 25 9416.73 9416.73 27 26923.30 26923.30 40 14843.30 14843.30 41 25774.43 25774.43
I believe this estimation is not correct, as no where I have given the break up of total students at district level (cluster level). Any help would be greatly appreciated.