Not sorted when conducting bysort egen command even after xtset and sort

[Minjae Yun]

I found this posting useful:
https://www.statalist.org/forums/for...rage-by-groups
My bad that I didn't check the previous questions

(may discard below)
/*------------------------------------------------------------------------------------------------------------------
Hello,

I got into a weird and simple trouble when I tried to generate a three-year moving average.
My commands are followed ("id" is the unique identifier and "year" is the time variable):

xtset id year
[output]
       panel variable:  id (unbalanced)
        time variable:  year, 1979 to 2015, but with gaps
                delta:  1 unit

sort id year
bysort id year: egen average= mean(L.k1 + L2.k1 + L3.k1)

And I get this error: "not sorted"
I don't have duplicates when I do "duplicates list id year".

What could be potential issues here??
I have 514 id's.

By the way, I am using State 14.0 in Windows 10.

[Nick Cox]

bysort id year is redundant here as id year pairs define observations uniquely. If you want the mean of the previous three values then that is, after xtset,

gen wanted = (L1.k1 + L2.k1 + L3.k1)/3

In contrast, the mean of a constant sum is just that sum. The mean of (1 + 2 + 3) is 6, not 2.

All that said, tssmooth ma returns moving averages. Writing your own code can be instructive, but an official command already exists.