Exclude certain observations only in one variable?

[Oscar Weinzettl]

Hello

gen gift_group_test = 0 if gift_total < . & gift_received == 1
replace gift_group_test = 1 if gift_total > 0 & gift_total < 10000
replace gift_group_test = 2 if gift_total >= 10000 & gift_total <= 100000
replace gift_group_test = 3 if gift_total > 100000
label define gift_group_test  0 "<10K" 1 "10K-100K" 2 "100K+" 
label val gift_group_test gift_group_test

I am trying to group observations given the amount of gifts they have received. Yet for some reason in the category 1 ("<10K") I have 2 observations who have not received any gifts, their gift_total = 0.

No matter what I do I cannot solve the issue, so I was wondering, can I suppress these two observations in a way that I keep them within my data set, but just throw them out of the variable gift_group_test? Is that possible?

[Nick Cox]

Show the two observations in question please.

[Jorrit Gosens]

I'll just add that your code could use some improvements, apart from the issue you raise.

This:

gen gift_group_test = 0 if gift_total < . & gift_received == 1

Should probably be:

gen gift_group_test = 0 if gift_total == 0

Note also that any value is smaller than missing, so 'if gift_total < . & gift_received == 1' is redundant: any value exactly equal to 1 (or 0) would be smaller than missing.

Your code:

gen gift_group_test = 0 if gift_total < . & gift_received == 1
replace gift_group_test = 1 if gift_total > 0 & gift_total < 10000
replace gift_group_test = 2 if gift_total >= 10000 & gift_total <= 100000
replace gift_group_test = 3 if gift_total > 100000

Could be simplified to:

gen gift_group_test = 0 if gift_total == 0 
replace gift_group_test = 1 if gift_total > 0 
replace gift_group_test = 2 if gift_total >= 10000
replace gift_group_test = 3 if gift_total > 100000 & gift_total !=.

(again presuming you want the first group o be zero gifts received)

Also note hat your labels are probably not what you want:

label define gift_group_test  0 "<10K" 1 "10K-100K" 2 "100K+"

Should probably be:

label define gift_group_test  0 "Zero" 1 "<10K" 2 "10K-100K" 3 "100K+"

Based on how you define the group numbers