Hi,
I would like to apply the vcov function from the survey package for the
variables api00 and api99 grouped by the stype variable which can assume H,
M and E categories.
From the code in the survey package manual:
data(api)
dclus1<-svydesign(id=~dnum, weights=~pw, data=apiclus1, fpc=~fpc)
rclus1<-as.svrepdesign(dclus1)
mns<-svyby(~api00, ~stype, rclus1, svymean,covmat=TRUE)
vcov(mns)
I have tried the following changes in order to get the variance matrix
estimation (as part of Taylor Linearization).
> data(api)
> dclus1<-svydesign(id=~dnum, weights=~pw, data=apiclus1, fpc=~fpc)
> mns<-svyby(~api00+api99, ~stype, dclus1, svytotal)
> mns
stype api00 api99 se.api00 se.api99
E E 3162561.8 2962356.8 842713.7 796474.0
H H 293115.0 282283.9 104059.8 101492.9
M M 534308.7 514982.0 108710.7 105036.6
> vcov(mns)
E:api00 H:api00 M:api00 E:api99 H:api99
M:api99
E:api00 710166313660 0 0 0 0
0
H:api00 0 10828434454 0 0 0
0
M:api00 0 0 11818006832 0 0
0
E:api99 0 0 0 634370797647 0
0
H:api99 0 0 0 0 10300818294
0
M:api99 0 0 0 0 0
11032691751
Warning message:
In vcov.svyby(mns) : Only diagonal elements of vcov() available
How to obtain the cross terms in the matrix above?
I have no clue on how to implement that.
Thanks a lot!
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
