st: RE: variable label length limit

[Phil Schumm <pschumm@uchicago.edu>]

On Feb 7, 2010, at 11:55 AM, Nick Cox wrote:
> The primary purpose of variable labels in Stata is surely better  
> labelling of variables within tables and graphs. The reluctance of  
> Stata to change the length limit does not, I think, arise from any  
> memory limit that would bite hard, but from the fact that variable  
> labels much over 80 characters would hardly be readable in most  
> cases. In most cases they would need to be truncated.
>
> That's not to deny that in circumstances like yours many users want  
> to store metadata with their data.
>
> The very longstanding Stata answer is to consider using  
> characteristics. See -char- or -notes-.


I would echo Nick's comment here -- you don't typically want the  
question text of an item in the variable label (especially if it  
exceeds 80 chars), since it often doesn't make for a good row or  
column heading.  For survey data, we store the full question text --  
along with any additional instructions, prompts, etc. -- in one or  
more characteristics.  This makes it easily accessible, both during  
interactive use and when generating reports.

WRT translating from SPSS, are you sure that Stat/Transfer doesn't  
provide an option to transfer long variable labels into Stata notes  
(which, BTW, are themselves implemented as characteristics)?  If not,  
then it should be straightforward to write an SPSS command which would  
generate a do-file to add the notes that you could execute against the  
translated dataset.


-- Phil

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/