Problem with 'Destring'

[Paris Rira]

Dear all,
I am going to destring CAE_COD, but it says CAE_COD: contains nonnumeric characters; no replace. While all are numeric. I mean I am going to make data without labels. Any ideas appreciated.

* Example generated by -dataex-. For more info, type help dataex
clear
input str6 CAE_COD
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
"01100"
end

destring CAE_COD , replace
CAE_COD: contains nonnumeric characters; no replace

[Joseph Coveney]

With a default limit of the first hundred observations, -dataex- isn't reliably capable of showing the offending values, which appear to occur further on. In order to find them, try something like the following.

contract CAE_COD
list if missing(real(CAE_COD)), noobs

[Paris Rira]

Thank you for getting back to me.

. contract CAE_COD

. 
. list if missing(real(CAE_COD)), noobs

  +-----------------+
  | CAE_COD   _freq |
  |-----------------|
  | 19_205      254 |
  +-----------------+

[Nick Cox]

So what does 19_205 mean and why you do seek a numeric equivalent? What is perhaps more likely is that you need to use encode.

A non-destructive way to check what is problematic is to go -- with the original data

tab CAE_COD if missing(real(CAE_COD))

[Paris Rira]

It solved. Thank you Nick & Joseph.

[Paris Rira]

What about non-numeric variables? I am going to see their original formats (without labels).

* Example generated by -dataex-. For more info, type help dataex
clear
input str2 PAIS
"AE"
"AO"
"AT"
"AU"
"BE"
"BG"
"BM"
"BR"
"BZ"
"CA"
"CH"
"CN"
"CV"
"CW"
"CY"
"CZ"
"DE"
"DK"
"EG"
"ES"
"FI"
"FR"
"GB"
"GG"
"GI"
"GR"
"HK"
"HU"
"IE"
"IL"
"IM"
"IN"
"IS"
"IT"
"JE"
"JP"
"KR"
"KW"
"KY"
"LI"
"LU"
"MA"
"MO"
"MT"
"MU"
"MX"
"MZ"
"NG"
"NL"
"NO"
"NZ"
"PA"
"PL"
"PT"
"SA"
"SC"
"SE"
"SG"
"SI"
"SK"
"ST"
"TC"
"TN"
"TR"
"TW"
"US"
"VG"
"VI"
"WS"
"ZA"
end

[Nick Cox]

What about non-numeric variables do you want to know? If you need a numeric country identifier given a string variable, use encode.

[Paris Rira]

They are labeled. I need to show them they are. i.e. 1-AE, 2-AO, ...
I need to show them AE, AO,.. When I wanna

g foreign_aff= 0
replace foreign_aff= 1 if pais=='AE'

NOT this code
replace foreign_aff= 1 if pais==1

[Clyde Schechter]

g foreign_aff= 0
replace foreign_aff= 1 if pais=='AE'

Using the original string variables, not the -encode-d numeric version, the code you show will work correctly if you change the single quote characters (') to double quotes (").

Also, although it is common to see people coding

gen logical_variable = 0
replace logical_variable = 1 if logical_expression

in Stata it is quicker and more transparent to do this as

gen logical_variable = logical_expression

[George Ford]

Those look like country identifiers.

Another approach is:

egen cid = group(PAIS)

[Paris Rira]

Using the original string variables, not the -encode-d numeric version, the code you show will work correctly if you change the single quote characters (') to double quotes (").

it worked.
g foreign_aff= 0
replace foreign_aff=1 if PAIS=="PT"
replace foreign_aff=0 if PAIS~="PT"

in Stata it is quicker and more transparent to do this as


gen logical_variable = logical_expression

I don't quite get it. could you please adjust it according to my above codes? Thanks.

[Paris Rira]

Those look like country identifiers.

Actually they are abbreviations. This code reveals the labels not the abbreviations which i seek for.

. egen cid = group(pais)


. list cid in 1/10

     +-----+
     | cid |
     |-----|
  1. |  57 |
  2. |  50 |
  3. |  32 |
  4. |  48 |
  5. |  50 |
     |-----|
  6. |  48 |
  7. |   7 |
  8. |  26 |
  9. |  50 |
 10. | 169 |
     +-----+

[George Ford]

They aren't labels, they are numeric values. PAIS already gives you the identifier name, so it's not clear to me what you are trying to produce or need to do with the new variable.

list PAIS cid

[George Ford]

You can assign labels to cid equal to PAIS

https://stats.oarc.ucla.edu/stata/faq/how-do-i-assign-the-values-of-one-variable-as-the-value-labels-for-another-variable/

[Paris Rira]

They aren't labels, they are numeric values.

I don't like these numeric values. They always confound me. Once the variables themselves are shown in the data, so I can easily figure out which is which
All I need is to create a dummy variable=1 when pais is PT. (PAIS in Portuguese means country).