Selective recoding of categorical variables

[Sonnen Blume]

Hi,

I have a lot of categorical variables that are coded differently, some as 1 to 3, some as 1 to 5, and so on.

There is no way to know from the names which ones are coded which way, so I am wondering if there is any method to detect the variables by the number of categories they have, and then recode them accordingly. e.g. I want variables with three categories to be labelled as "agree", "neutral, "disagree", and those with 5 as 'always', 'almost always', sometimes', 'often', 'never', and so on.

Please let me know if there is any way to achieve this.

Thank you.

[Nick Cox]

distinct from the Stata Journal lets you identify variables with 3 or 5 distinct values. Missings are ignored by default.

. webuse nlswork
(National Longitudinal Survey of Young Women, 14-24 years old in 1968)

. distinct, min(3) max(3)

---------------------------------
          |     total   distinct
----------+----------------------
     race |     28534          3
---------------------------------

[Sonnen Blume]

Thank you Nick! This is what I was looking for.