Re: st: Use varlabel in expressions

[Joerg Luedicke <joerg.luedicke@gmail.com>]

Perhaps consider the following example. It is assumed that "na-"
is always "na-" exactly (e.g., not "na -" or "n a-") and that it will always
affect the exact first 3 columns of the variable label. If these
assumptions do not hold, something else may be needed,
depending on the exact problem.

*---------------------------
sysuse auto, clear
gen x=1
label var x "na-whatever"

foreach v of varlist _all {
	loc lab : variable label `v'
	if substr("`lab'", 1, 3)=="na-" {
		drop `v'
	}
}
*---------------------------

J.



On Fri, Mar 23, 2012 at 11:53 AM, Marcelo Lafleur <lafleurm@un.org> wrote:
> I've had not luck with this and all references in this list point to use
> of value labels.
>
> My question is simple: How, if possible, to drop variables based on a
> prefix in their varlabels?
>
> Examples:
>
> . des  mv218 mv225 mv245 mv246
>
> variable name       value label      variable label
> ----------------------------------------------------------------------------
> mv218                                           number of living children
> mv225                  mv225            na-current pregnancy wanted
> mv245                                           women fathered children
> with
> mv246                   mv246           na-married to mother when first
> child was born
>
> (I edited the output a bit)
>
>
> I want to drop mv225 and mv246 by testing that the variable label starts
> with "na-"
>
> Having hundreds of variables in each dataset, it would be impractical to
> do by hand.
>
> Any ideas?
>
> Marcelo
> *
> *   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/

*
*   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/