Save variable labels in R so that Stata can read them

Question

I have a dataset with over 170 variables that looks as follows:

df <- data.frame(var1 = 1:3, var2 = 2:4, var3 = 2:4, var4 = 2:4, var5 = 2:4)

I have manually added variable values using Hmisc:

library(Hmisc)
var.labels = c(var1 = "label 1",
 var3 = "label 2",
 var4="label 4" )
label(df) = as.list(var.labels[match(names(df), names(var.labels))]) 

Do note that not all variables have labels and that it is much easier for me to specify variable labels by name (var*), rather than position given my large number of colums.

The problem is that when I save my dataset with write.csv(df,"df.csv")or with write.dta(df,"df.dta") my variable labels get lost.

How can I save the data in a way that can be re-imported into R and Stata with the labels restored?

• MrFlick: Standard CSV files don't have any way to save labels. You can save an object in R in a binary format with saveRDS which would preserve labels but you can't read that into Stata. Perhaps you can just save the labels in a separate file and merge within each program.

Answer 1

It is a little tricky. You need to label all variables. If you do not want to label a variable, you may still need to put variable = "". Otherwise write.dta will ignore all variable labels.

In R

df <- data.frame(var1 = 1:3, var2 = 2:4, var3 = 2:4, var4 = 2:4, var5 = 2:4)
attr(df, "var.labels") <- c(var1 = "label 1", var2 = "label 2", 
 var3 = "", var4="label 4", var5 = "")
foreign::write.dta(df, "dat_stata.dta")

In Stata, you get:

. des
Contains data from C:...dat_stata.dta
 obs: 3 Written by R. 
 vars: 5 
 size: 60 
------------------------------------------------------------------------------------------------------
 storage display value
variable name type format label variable label
------------------------------------------------------------------------------------------------------
var1 long %9.0g label 1
var2 long %9.0g label 2
var3 long %9.0g 
var4 long %9.0g label 4
var5 long %9.0g 
------------------------------------------------------------------------------------------------------
Sorted by: 

Please note: I used Stata 14, and R package foreign

• Alex: I was wondering is there a way to automatically attribute empty labels to those variables to which I do not specifically assign a value? My dataset has over 700 variables and i only need to label those that are most relevant. something like attr(df, "var.labels") <- c(var1 = "label 1", var2 = "label 2" , var4="label 4") and then tell r to automatically assign "" to the rest of the variables
• Guibor Camargo Salamanca: It works, but you have to make sure to ADD A LABEL TO EACH VARIABLE, otherwise, it will not work!

Answer 2

You can set the colnames of the dataframe to those labels that are available:

colnames(df)[which(!is.na(label(df)))]<-as.character(label(df)[which(!is.na(label(df)))])

• Alex: Thanks for this, but this is not really what I want, I would like to have my variable names that are easily accessible and then have labels saved so that when I have doubts on the data I know what it is about.