I have a survey data, with categorical variables. I would like to compute the proportion for each responses in Q1, as well as generating the confidence interval or uncertainty around the estimate. Is that doable? The dataset already has sample weight generated. I am using R survey package but can't seem to get my head around it. An example table below Thank you

Create a survey design object

survey_design <- svydesign(ids = ~1, data = xxx, weights = ~Personweight)

Calculate the multinomial proportions

proportions <- svyciprop(~Q1, survey_design)

#error seems to suggest that this function doesn't work for non-binary variable.

If you're happy with the ordinary +/- 1.96 standard error confidence intervals, you can just use svymean and confint.

Using one of the built-in examples:

> svymean(~stype, dclus2)
          mean     SE
stypeE 0.68118 0.0556
stypeH 0.13432 0.0567
stypeM 0.18450 0.0222
> confint(svymean(~stype, dclus2))
            2.5 %    97.5 %
stypeE 0.57212715 0.7902345
stypeH 0.02312234 0.2455123
stypeM 0.14094122 0.2280625

If you want one of the other methods, you need to do some data wrangling. You could use svyciprop on binary indicators

> svyciprop(~I(stype=="E"),dclus2,method="xlogit")
                       2.5% 97.5%
I(stype == "E") 0.681 0.560 0.782

or you could transform using svycontrast, use confint, and then transform back, which is what actually happens in svyciprop

> m<-svymean(~stype, dclus2)
> l<-svycontrast(m, quote(log(stypeE/(1-stypeE))))
> c<-confint(l,df=degf(dclus2))
> exp(c)/(1+exp(c))
             2.5 %   97.5 %
contrast 0.5599558 0.782011

The only issue now is how to write the loop that does this for all levels. Perhaps something like

> f<-function(level) eval(bquote(
       svyciprop(~I(stype==.(level)),dclus2,method="xlogit")
     ))
> lapply(c("E","M","H"),f)
[[1]]
                       2.5% 97.5%
I(stype == "E") 0.681 0.560 0.782

[[2]]
                       2.5% 97.5%
I(stype == "M") 0.185 0.144 0.234

[[3]]
                         2.5%  97.5%
I(stype == "H") 0.1343 0.0547 0.2939