Sorting Panel Data to clean errors

[Sadie Belechak]

I am using NLSY1979 panel data.

* Example generated by -dataex-. For more info, type help dataex
clear
input int(caseid_1979 year weight) float weightchange
   30 1990 170 -826
  170 1990 178 -818
  637 2008 225 -315
 1207 2010 165 -426
 2290 2006 250 -200
 4120 2010 170 -295
 4263 2018 188 -214
 4429 1990 188 -808
 7673 2016 135 -300
 8151 1990 136 -860
 8220 1990 139 -857
 8331 2014 140 -200
10480 1990 170 -826
11984 1990 160 -836
end

I want to sort my variable weightchange in ascending order, but I want to still see every weight observation for that ID. This way I can determine if the extreme weight values are mistakes or if they are reasonable weights in comparison to the ID's weights in every year. Using the code

*sort weightchange caseid_1979

sorts my data in data editor browse as:



I want all the observations for each caseID to show in order from the lowest weight change to highest. So it will show the ID for every year with the weightchange of -860 rather than just the year with the dramatic weight change.

[Clyde Schechter]

Isn't it

sort caseid_1979 weightchange

[Jared Greathouse]

Duplicate.

[Sadie Belechak]

sort caseid_1979 weightchange

displays the data like this
.
Which is the format I want, but I was wanting the whole caseID observations for the ID with the lowest weight change( so the case ID that has a weight change of -860) listed first and so on...

[Clyde Schechter]

So do you mean this?

by caseid_1979, sort: egen lowest_weightchange = min(weightchange)
sort lowest_weightchange caseid_1979 year

Note: I could not really test this with your example data, because in your -dataex- example, each caseid_1979 had only a single observation.

[William Lisowski]

by caseid_1979, sort: egen wc_min = min(weightchange)
sort wc_min caseid_1979 year
list, sepby(wc_min caseid_1979)

[Sadie Belechak]

This code worked. Thank you so much.

by caseid_1979, sort: egen minweightchange=min(weightchange)
sort minweightchange caseid_1979 year