[Andrew Musau]

You need to find the 40% of affected variables and then provide a data example based on them.

[daniel klein]

Well, show us one of those 40 % of unaffected variables and provide a data example that represents the problem then. More verbal descriptions and vague speculations won't help here.

[Sonnen Blume]

Well, show us one of those 40 % of unaffected variables and provide a data example that represents the problem then. More verbal descriptions and vague speculations won't help here.

Thank you. Here are two such variables:

* Example generated by -dataex-. To install: ssc install dataex
clear
input double(is12reasnprx1 is12reasnprx4)
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
1 2
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
1 2
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
2 2
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
2 2
. .
. .
. .
1 2
. .
. .
1 1
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
. .
end
label values is12reasnprx1 is12reasnprx1
label def is12reasnprx1 1 " 1 Yes", modify
label def is12reasnprx1 2 " 2 No", modify
label values is12reasnprx4 is12reasnprx4
label def is12reasnprx4 1 " 1 Yes", modify
label def is12reasnprx4 2 " 2 No", modify

[daniel klein]

elabel define * (= #) (= ustrregexra(@, "^(\s*\d+\s+)(\w+)", "$2")) , modify

or, a little more carefully but including extended missing values

elabel define * (= #) (= ustrregexra(@, "^(\s*"+string(#)+"\s+)(\w+)", "$2")) , modify

The second line of code will only remove numeric prefixes that match the respective numeric values. Here is an example:

. label list
foo:
          42 42 foo
          73 42 bar
          .z .z foobar

. elabel define * (= #) (= ustrregexra(@, "^(\s*"+string(#)+"\s+)(\w+)", "$2")) , modify

. label list
foo:
          42 foo
          73 42 bar
          .z foobar