R to Stata: Exporting tibble with value labels

Question

My task is to write a Stata-readable dataframe with variable labels and value labels. It appears that the package haven is the most suitable one to do such an operation (better than readstata13 and foreign).

I begin with creating the tibble:

library(haven)

x1 <- labelled(
  sample(1:5), 
  c(Good = 1, Bad = 5)
)
x1

x2 <- labelled(
  c("M", "F", "F", "F", "M"), 
  c(Male = "M", Female = "F")
)
x2

df <- tibble::data_frame(x1, x2, z = 1:5)
df

Then, I export the tibble

write_dta(df)

When I read it in Stata, the labels replace the real values (in x1, 1 becomes equal to "Good" and 5 becomes equal to "Bad"). How can I read the tibble in Stata, in way such that the values remain from 1 to 5 and "Good" and "Bad" are stored as value labels?

Answer 1

It's possible but rather fiddly, you will have to attach labels manually using attr¹ function as in the example below.

# Attach variable and value labels to a new R data.frame
attr(df, "var.labels") <- c("my var label 1", "my var label 2")
# To create and attach value labels
val1 <- c(apple=10, beans=20)
val2 <- c(kg=1, bag=2)
attr(df, "label.table") <- list(val1, val2)
attr(df, "val.labels") <- c("val1", "val2")
# And finally describe the dataset
attr(df, "datalabel") <- "Collected from MWI 2013"

---

¹The original solution was provided by Melanie Bacou and is available in this gist.