Bootstrapping Proportions of Categorical Variables in R or Stata

Question

I need help in doing bootstrap in either R or Stata software. I want to calculate the proportion of those who say Yes and No to e.g. effectiveness of a policy

In Stata I have this code

bs "summarize y1" "r(mean)", reps(200) size(770)

what should be the value for r(mean) to estimate the proportions?

Also, I have this code in R:

test <- function (q13){
    test13 <- table(q13)
    rel_freq <- test13/sum(test13)
    return(rel_freq)
      }

results <- boot(data=q13, statistic=test,
                R=200)

How do I correct the code? I'm getting the error

Error in statistic(data, original, ...) : unused argument(s) (original)

Answer 1

In Stata you can use proportion if a variable has more than two categories:

//sample data

sysuse auto, clear
keep if (headroom==2.0 |headroom==2.5)
gen prop=.
replace prop=0 if headroom==2.0
replace prop=1 if headroom==2.5

//say 0 is yes and 1 is no

set seed 123
bootstrap _b, reps(100):proportion prop

Updated as per @Nick: For binary variable, the following is sufficient

bootstrap r(mean), reps(100): summarize prop, meanonly

..........................................................................................................................................................................

In R, you can do as follows using boot package and mtcars data:

library(boot)
set.seed(123)
x<-mtcars$vs
myprop<-function(x,i){
sum(x[i]==0)/length(x)
}

bootprop <- boot(x,myprop,100)