[R] reading stata files: preserving values of variables converted to factors

Ben Saylor bensaylor at fastmail.fm
Fri Jul 27 21:03:51 CEST 2007

Thanks for the clarification.  I found the labels & values in


which looks the same whether convert.factors is TRUE or FALSE.


Thomas Lumley wrote:
> On Thu, 26 Jul 2007, Ben Saylor wrote:
>> Hi,
>> I am a Stata user new to R.  I am using read.dta to read a Stata file
>> that has variables with value labels.  read.dta converts them to
>> factors, but seems to recode them with values from 1 to <number of
>> factor levels> (looking at the output of unclass(<varname>)), so the
>> original numerical values are lost.
> Yes. The R factor type should not be used if you want the original 
> levels. It is not a 'labelled numeric' type and the numbers are an 
> implementation detail.
>>  Using convert.factors=FALSE
>> preserves the values, but seems to discard the labels.
> It doesn't discard the labels. They are kept in the attributes of the 
> data frame.
>     -thomas

More information about the R-help mailing list