How to generate a variable that takes value 1 if another variable is absent across all available rounds?

[Titir Bhattacharya]

Hi,

Please consider the following example data

* Example generated by -dataex-. To install: ssc install dataex
clear
input str8 id float(round hi)
"IN010001" 2 .
"IN010001" 3 1
"IN010001" 4 1
"IN010001" 5 1
"IN010002" 2 .
"IN010002" 3 1
"IN010002" 4 1
"IN010002" 5 0
"IN010003" 2 .
"IN010003" 3 1
"IN010003" 4 1
"IN010003" 5 1
"IN010004" 2 .
"IN010004" 3 1
"IN010004" 4 1
"IN010004" 5 1
"IN010005" 2 .
"IN010005" 3 0
"IN010005" 4 1
"IN010005" 5 1
"IN010006" 2 .
"IN010006" 3 0
"IN010007" 2 .
"IN010007" 3 0
"IN010007" 4 0
"IN010007" 5 0
"IN010008" 2 .
"IN010008" 3 1
"IN010008" 4 0
"IN010008" 5 0
"IN010009" 2 .
"IN010009" 3 1
"IN010009" 4 1
"IN010009" 5 1
"IN010010" 2 .
"IN010010" 3 1
"IN010010" 4 1
"IN010010" 5 1
"IN010011" 2 .
"IN010011" 3 1
"IN010011" 4 0
"IN010011" 5 1
"IN010012" 2 .
"IN010012" 3 1
"IN010012" 4 1
"IN010012" 5 1
"IN010013" 2 .
"IN010013" 3 1
"IN010013" 4 1
"IN010013" 5 1
"IN010014" 2 .
"IN010014" 3 1
"IN010014" 4 1
"IN010014" 5 1
"IN010015" 2 .
"IN010015" 3 0
"IN010015" 4 1
"IN010015" 5 1
"IN010016" 2 .
"IN010016" 3 0
"IN010016" 4 1
"IN010016" 5 1
"IN010017" 2 .
"IN010017" 3 1
"IN010017" 4 1
"IN010017" 5 1
"IN010018" 2 .
"IN010018" 3 1
"IN010018" 4 1
"IN010018" 5 1
"IN010019" 2 .
"IN010019" 3 1
"IN010019" 4 1
"IN010019" 5 1
"IN010020" 2 .
"IN010020" 3 1
"IN010020" 4 1
"IN010020" 5 0
"IN010021" 2 .
"IN010021" 3 1
"IN010021" 4 1
"IN010021" 5 1
"IN010023" 2 .
"IN010023" 3 1
"IN010023" 4 1
"IN010023" 5 1
"IN010024" 2 .
"IN010024" 3 1
"IN010024" 4 1
"IN010024" 5 1
"IN010025" 2 .
"IN010025" 3 0
"IN010025" 4 0
"IN010025" 5 0
"IN010026" 2 .
"IN010026" 3 1
"IN010026" 4 1
"IN010026" 5 1
"IN010027" 2 .
"IN010027" 3 0
end

I want to generate a variable never_taker that takes the value 1 if hi is 0 across all three rounds, and takes the value 0 if hi is 1 in any of the three rounds.

It would be great if someone could offer some suggestions.
Thanks!

[Rich Goldstein]

assuming that by "all 3 rounds" you mean rounds 3-5, the following should do it:

. bys id: egen byte rtot=total(hi) if inrange(round,3,5)

. bys id: egen num=count(round) if inrange(round,3,5)

. gen byte never_taker=rtot==0 & num==3

[Nick Cox]

As a small twist on @Rich Goldstein's idea consider say

bys id : egen count1 =total(hi & inrange(round(3, 5)))

gen never_taker = count1 == 0

[Titir Bhattacharya]

assuming that by "all 3 rounds" you mean rounds 3-5, the following should do it:

. bys id: egen byte rtot=total(hi) if inrange(round,3,5)

. bys id: egen num=count(round) if inrange(round,3,5)

. gen byte never_taker=rtot==0 & num==3

Thanks Rich. this was really helpful. Just one problem: when the value 1 appears for never_taker, it appears against rounds 3-5, but not against round 2. Say for eg, an ID "IN010026" is available for rounds 2,3,4,5. The variable "hi" takes 0 values across rounds 3-5. So never_taker would be 1. But this value 1 appears against rounds 3-5 but not against round 2. Value of never_taker for round 2 is still 0. Is there any way to modify the code such that if for any ID,

hi==0

for rounds 3-5, against all observations of that ID, we have

never_taker==1

?

[Titir Bhattacharya]

As a small twist on @Rich Goldstein's idea consider say

bys id : egen count1 =total(hi & inrange(round(3, 5)))

gen never_taker = count1 == 0

Thanks Nick for your answer. However, the first line of the code is producing an error:

bys id : egen count1 =total(hi & inrange(round(3, 5)))
invalid syntax
r(198);

[Nick Cox]

Sorry; should be

 
 inrange(round, 3, 5)

as in Rich Goldstein 's posts.

[Rich Goldstein]

Titir Bhattacharya I did not understand that you wanted the "1" on all occasions so use egen with the max option to get it for any missing founds; see

h egen

the above will give you a new variable; here is a way without a new variable:

bys id (round): replace never_taker=never_taker[_n+1] if never_taker==.

[Titir Bhattacharya]

Thanks Nick and Rich..Your solutions were really helpful