order() get random result

[Wen Qin]

group = (1 \ 1 \ 1 \ 2 \ 2 \ 2 \ 3 \ 3 \ 4 \ 5 \ 5 \ 6 \ 7 \ 8 \ 8 \ 8)
order(group,1)

The result seems to be ramdon.


: order(group,1)
1
+------+
1 | 2 |
2 | 1 |
3 | 3 |
4 | 5 |
5 | 4 |
6 | 6 |
7 | 8 |
8 | 7 |
9 | 9 |
10 | 11 |
11 | 10 |
12 | 12 |
13 | 13 |
14 | 14 |
15 | 16 |
16 | 15 |
+------+

: order(group,1)
1
+------+
1 | 2 |
2 | 3 |
3 | 1 |
4 | 6 |
5 | 4 |
6 | 5 |
7 | 7 |
8 | 8 |
9 | 9 |
10 | 10 |
11 | 11 |
12 | 12 |
13 | 13 |
14 | 14 |
15 | 16 |
16 | 15 |
+------+


I expect that the result is

1
+------+
1 | 1 |
2 | 2 |
3 | 3 |
4 | 4 |
5 | 5 |
6 | 6 |
7 | 7 |
8 | 8 |
9 | 9 |
10 | 10 |
11 | 11 |
12 | 12 |
13 | 13 |
14 | 14 |
15 | 15 |
16 | 16 |
+------+

Is there any way to do this.

Thanks!

[Maarten Buis]

Why would you expect the first value 1 to be less than the second value 1? There is no way to say so, so Stata does the only sensible thing and sorts them in a random order. You can make an additional vector storing the original sort order and use that as a secondary variable on which to sort your data. However, I strongly recommend you reconsider this. Such sorts depending on the original sort order are extremely fragile. Instead I would look into getting a unique sort by substantive means.

mata
group = (1 \ 1 \ 1 \ 2 \ 2 \ 2 \ 3 \ 3 \ 4 \ 5 \ 5 \ 6 \ 7 \ 8 \ 8 \ 8)
orig_order = 1::16
order((group, orig_order),(1,2))
end

[Wen Qin]

Thanks, but the random order really surprised me.