Encode so that variables take on same values across similar variables

[Emily Ellis]

Hi all,

I'm attempting to recode several string variables. Specifically, I have 5 variables representing industries, ie, Industry1, Industry2 Industry3, etc. Possible values for each variable are the same (eg, Construction, Manufacturing, Health Care, etc). Because the "encode" command assigns values to string observations alphabetically, in "Industry1", Construction may be coded as 1, but if Construction does not appear in the variable "Industry2", Health Care may be coded as 1.

Is there an efficient way to encode all 5 variables so that the same string value takes on the same numeric value across all 5?

Thank you in advance!

[Nick Cox]

ssc desc multencode 


---------------------------------------------------------------------------------
package multencode from http://fmwww.bc.edu/repec/bocode/m
---------------------------------------------------------------------------------

TITLE
      'MULTENCODE': module to encode multiple string variables into numeric

DESCRIPTION/AUTHOR(S)
      
      multencode creates new numeric variables newvarlist, with value
      labels defined and attached, based on the string variables
      strvarlist. The same set of value labels is used for all the new
      variables.
      
      KW: encode
      KW: string
      KW: numeric
      KW: value labels
      
      Requires: Stata version 9
      
      Distribution-Date: 20090121
      
      Author: Nicholas J. Cox, Durham University
      Support: email [email protected]
      

INSTALLATION FILES                           (type net install multencode)
      multencode.ado
      multencode.hlp
---------------------------------------------------------------------------------
(type -ssc install multencode- to install)

[Emily Ellis]

Thank you so much. I'll be sure to remember this one!

[Jeph Herrin]

No disrespect to Nick's solution, but you can do this directly using the -label()- option of -encode-

encode stringvar1, gen(numvar1) label(codes)
encode stringvar2, gen(numvar2) label(codes)

The first use, the label -codes- is created. The second time it is reused, and extended as necessary. This ensures that all the variables share the same labels.

hth,
Jeph

[Nick Cox]

Jeph's idea is neat and uses just official commands. The downside is likely to be an untidy set of labels. If stringvar1 is based on frog, newt and toad and dragon isn't seen before stringvar2 then the labels will be in that order.