multiple factors in svyby vs. svytotal

Question

I'm trying to use svytotal to calculate the population breakdown of each pair (PUMA, BDSP). When I use svyby, it works as I would expect:

> svyby(~NP, ~factor(PUMA)+factor(BDSP), ss12hsf, svymean)
        factor(PUMA) factor(BDSP)        NP         se
7501.0          7501            0 0.8328120 0.10738595
7502.0          7502            0 1.1790473 0.10128726
7503.0          7503            0 1.0707710 0.05241559
7504.0          7504            0 0.9982076 0.16726445
...

However, when I try to do the same with svytotal, it seems to just do a list concatenation instead of a cross product:

> svytotal(~factor(PUMA)+factor(BDSP), ss12hsf)
                  total      SE
factor(PUMA)7501  66732  514704
factor(PUMA)7502  66350  570194
factor(PUMA)7503  71129 1597630
factor(PUMA)7504  54108  595059
factor(PUMA)7505  44327  498771
factor(PUMA)7506  40461  424705
factor(PUMA)7507  33792  216696
factor(BDSP)0     50036  867466
factor(BDSP)1    111847 1049884
factor(BDSP)2    111976 1586937
factor(BDSP)3     69279 1112640
factor(BDSP)4     24386  496505
factor(BDSP)5      5841  384093
factor(BDSP)10     3534   88227

Any ideas or suggestions?

Answer 1

there is an example of how to do this in ?svytotal

svytotal( ~interaction( var1 , var2 ) , design )