How can I aggregate the sum of a variable only for each subgroup?

[Xu Ji]

I have here a Steam Dataset which includes individual steam user and their playtimes(overall) and the games they played. I further divided the player in hardcore(=1) and casual player (=0). Overall I want to test how various factors have influence on the overall playtime of the players, but now I want to build 2 regressions, one for hardcore players and one for casual players(because I think that the effect of every factor can differ between those two). But in order to do that, I need the sum of the overall playtime from the 2 subgroups. I tried

egen playtime_type = sum(playtime_sum), by (hightype)

, but the outcome just doesn't make sense. How can I aggregate the sum of playtime only for each subgroup?

Here is a example from the dataset

* Example generated by -dataex-. For more info, type help dataex
clear
input str17 steamid float(playtime_sum hightype)
"76561197960265729"    0 0
"76561197960265730"   45 0
"76561197960265730"   45 0
"76561197960265730"   45 0
"76561197960265733" 1710 0
end

[Hemanshu Kumar]

I notice you have several duplicates in there, which will probably cause the sum to go wrong. You can either

duplicates drop steamid, force
egen playtime_type = sum(playtime_sum), by(hightype)

to get rid of the duplicates, or you can

egen tag = tag(steamid)
egen playtime_type = sum(playtime_sum*tag), by(hightype)
drop tag

which makes the duplications cease to count.

[Xu Ji]

I am sorry I forgot to mention that every row means one game. So ID 7656...30 appears 3 times means there are three games documented.

[Hemanshu Kumar]

I think it would be useful for you to show what values you expect the variable to take for this sample dataset

[Nick Cox]

Cross-posted at https://stackoverflow.com/questions/...-for-subgroups

Please note our policy on cross-posting, which is that you are asked to tell us about it.

[Xu Ji]

Thank you Nick, I will pay attention to that in the future.