Strange behavior of collapse command

[Upeksha Diss]

I have a problem with the collapse command in a dataset with 497 observations. When I collapse a numeric variable by a string variable, the number of observations is different on different computers. The type of string variable is str14. The command we use is as follows;

clear all

use testdata.dta
collapse (mean) unity, by (InputID)

After the above collapse command, the number of observations on windows computer is 227 while the number of observations on MAC is 235.

We have been working for hours for this problem, We appreciate any suggestions.
We have attached the data. Thank you very much in advance.

Best

[Nick Cox]

You need to be clear that you are using the same .dta file. What you presented has 227 distinct values of InputID which are all str14 (and all have 0000 as a redundant middle part).

[Upeksha Diss]

Thank you for the answer. I will get back after checking the files again.

[William Lisowski]

On my Mac using the dataset attached in post #1, we see 497 observations, none blank, with 227 distinct values.

. codebook InputID

------------------------------------------------------------------------------------------------
InputID                                                                                  InputID
------------------------------------------------------------------------------------------------

                  Type: String (str14)

         Unique values: 227                       Missing "": 0/497

              Examples: "NG199000000407"
                        "NG199000001002"
                        "NG199000001406"
                        "NG199000001904"

Added in edit: And the collapse command produces 227 observations.

. collapse (mean) unity, by (InputID)

. codebook InputID

------------------------------------------------------------------------------------------------
InputID                                                                                  InputID
------------------------------------------------------------------------------------------------

                  Type: String (str14)

         Unique values: 227                       Missing "": 0/227

              Examples: "NG199000000510"
                        "NG199000001014"
                        "NG199000001509"
                        "NG199000001903"

.

[Nick Cox]

My machine in #2 is under Windows. I strongly suspect that platform is immaterial here. The story lies elsewhere.