method to check if two variables are identical

[Michael Duarte Goncalves]

Hi everyone,

I am looking for a method to check if two (or more) variables are identical to each other (i.e. if the content is identical), independently of the data type (float, byte, str, etc.). I could do the check visually, but on a database with more than 7 million observations, I can't.

Does anyone have a little trick for this? It will save me a lot of time (and less gray hair by the way).

* Example generated by -dataex-. For more info, type help dataex
clear
input long i float t byte c str2 c_cod
1415 724 10 "FR"
1415 725 10 "FR"
1415 726 10 "FR"
1415 727 10 "FR"
1415 728 10 "FR"
1415 729 10 "FR"
1415 730 10 "FR"
1415 731 10 "FR"
1415 732 10 "FR"
1415 733 10 "FR"
1415 734 10 "FR"
1415 735 10 "FR"
1415 736 10 "FR"
1415 737 10 "FR"
1477 712  2 "BE"
1477 713  2 "BE"
1477 714  2 "BE"
1477 727  2 "BE"
1477 728  2 "BE"
1477 729  2 "BE"
1477 730  2 "BE"
1477 731  2 "BE"
1477 732  2 "BE"
1477 733  2 "BE"
1477 734  2 "BE"
end
format %tm t
label values c clab
label def clab 2 "BE", modify
label def clab 10 "FR", modify
label def clab 22 "VD", modify

For my variable c, the code is defined as follows. 10 is equivalent to FR, and 2 to BE. In theory, the content in `c` is supposed to be "the same" as the one in `c_cod`.
Thank you in advance!

Best regards,

[Ken Chui]

I don't have a formal code to suggest, but if I had to check I'd collapse and check for duplicate:

collapse (count) i, by(c c_cod)
duplicates report c
duplicates report c_cod

If they are unique pair, then after aggregation both of them should be unique.

[Michael Duarte Goncalves]

Ken Chui : Thank you very much! I will try the code and keep you posted.

Thank you again.

Michael

[Michael Duarte Goncalves]

Hi Ken Chui:

I confirm, your methods works well.
Thanks again.

Best,

Michael

[Nick Cox]

See also an FAQ on this topic at https://www.stata.com/support/faqs/d...ions-in-group/

Here is some code:

* Example generated by -dataex-. For more info, type help dataex
clear
input long i float t byte c str2 c_cod
1415 724 10 "FR"
1415 725 10 "FR"
1415 726 10 "FR"
1415 727 10 "FR"
1415 728 10 "FR"
1415 729 10 "FR"
1415 730 10 "FR"
1415 731 10 "FR"
1415 732 10 "FR"
1415 733 10 "FR"
1415 734 10 "FR"
1415 735 10 "FR"
1415 736 10 "FR"
1415 737 10 "FR"
1477 712  2 "BE"
1477 713  2 "BE"
1477 714  2 "BE"
1477 727  2 "BE"
1477 728  2 "BE"
1477 729  2 "BE"
1477 730  2 "BE"
1477 731  2 "BE"
1477 732  2 "BE"
1477 733  2 "BE"
1477 734  2 "BE"
end
format %tm t
label values c clab
label def clab 2 "BE", modify
label def clab 10 "FR", modify
label def clab 22 "VD", modify

bysort c (c_cod) : gen diff = c_cod[1] != c_cod[_N]

list if diff

and for the data example there is no output, which is fine. No news is good news.

[daniel klein]

In #1, changing

label def clab 10 "FR", modify
label def clab 22 "VD", modify

to

label def clab 22 "FR", modify
label def clab 10 "VD", modify

will not make any difference to the code suggested in #2 (Edit: or #5). The code only checks whether both variables define the same groups; it does nothing to check the underlying values. Whether that is a problem is not clear because the initial post does not define what is meant by "identical". By definition, string variables and numeric variables cannot be identical. Does identical mean that values of c_cond must not vary within values of c (and the other way round) or do the value labels have to match the string values, too?

[Michael Duarte Goncalves]

Hi Nick Cox: Thanks for the nice suggestion and for the stata FAQ resource.

Hi daniel klein: I apologize for the confusion. By identical I mean option 1 that you suggested: that the values of c_cond should not vary within the values of c (and vice versa).

Best,

Michael

[Nick Cox]

daniel klein has a great point, but I think the close of #1 is a better guide to the real question than the title of the thread. I read it as whether the values of two variables are in one-to-one correspondence. There is certainly no sense to saying that "BE" is identical to (*) 2.

(*) or "identical with", a phrasing often commended that sounds very strange to me, although English is my first language.

EDIT Crossed with #7, which confirms the guess here.