RE: st: filter by value labels

[Nick Cox <n.j.cox@durham.ac.uk>]

Thanks; your situation is now clearer to me. I think you need to -decode- but I understand that you are reluctant to do that. 

With string values, your conditions could be rewritten

... if strmatch(firstname, "Joe") & strmatch(firstname, "Doe") & strmatch(town, "Frankfurt")

... if strpos(firstname, "Joe") & strpos(firstname, "Doe") & strpos(town, "Frankfurt") 

as 

if strmatch() == 1 

is equivalent to 

if strmatch()

and 

if strpos() > 0  

is equivalent to 

if strpos() 

But those are minor details. 

Nick 
n.j.cox@durham.ac.uk 

On Behalf Of Stefan.Gawrich@hlpug.hessen.de

Thanks Nick,

this is the idea I was looking for but unfortunately it doesn’t fit to my example.


The use of  

"value_label_content":value_label_name 

is restricted to single values and their corresponding value labels where the latter must be known.

In my case value labels are unknown and I want to check and filter those containing defined substrings. 

In correction to my first post the command with non-encoded vars would be like that:
tab unitid if strmatch(firstname ,"*Joe*") == 1 & strmatch(firstname ,"Doe*") == 1 & strmatch(town ,"*Frankfurt*") == 1

(So "Joe" could have other names too, Frankfurt could be named "Stadt Frankfurt" or "Frankfurt am Main"...)
(The usual hassle with address data...)

There is none but if there was e.g. some function "vallabel" that would substitute variable values with their corresponding value labels (e.g. for comparison in -strmatch - ) I would encode more often and some of my datasets could be much smaller. 


*
*   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/