fill in values until next identical value (paneldata)

[Gustav Egede Hansen]

Hi everybody

I need to fill in some values. Here is an example dataset.

* Example generated by -dataex-. For more info, type help dataex
clear
input float(time id education wanted)
 1 9999  .  .
 2 9999  .  .
 3 9999  .  .
 4 9999 20 20
 5 9999  . 20
 6 9999  . 20
 7 9999  . 20
 8 9999  . 20
 9 9999  . 20
10 9999 20 20
11 9999  .  .
12 9999  .  .
 1 1000  .  .
 2 1000 25 25
 3 1000  . 25
 4 1000  . 25
 5 1000  . 25
 6 1000 25 25
 7 1000  .  .
 8 1000  .  .
 9 1000 30 30
10 1000  . 30
11 1000 30 30
12 1000  .  .
end

The first observation in education is when the individual starts an education and the next observation in education is when the individual ends the education. I need to fill in the rows between the first and the second observation, thus indicating the entire period of an educational program. So, what I need is the variabel called wanted.

I have looked into this but unfortunately I cannot make it work.

Best
Gutav

[Nick Cox]

https://journals.sagepub.com/doi/pdf...6867X231196519 discusses precisely this problem in its Section 5.1.

[Wei Hai Deng]

This might do it. I used the carryforward command to identify and fill last observation(s). It can be installed from ssc install carryforward.

gen byte mis = missing(education)
bysort id mis (time): gen last = 1 if mis == 0 & _n == _N
    bysort id (time): carryforward last, replace

by id (time): replace education = education[_n-1] if education == . & education[_n-1] !=. & missing(last)

[Gustav Egede Hansen]

Hi both

The code from Section 5.1. did the trick! Thank you!

I also ran the code you provided Wei, and it looks like, in row 7 and 8, it carried over the education from the previous row.

* Example generated by -dataex-. For more info, type help dataex
clear
input float(time id education wanted) byte mis float last
 1 1000  .  . 1 .
 2 1000 25 25 0 .
 3 1000 25 25 1 .
 4 1000 25 25 1 .
 5 1000 25 25 1 .
 6 1000 25 25 0 .
 7 1000 25  . 1 .
 8 1000 25  . 1 .
 9 1000 30 30 0 .
10 1000 30 30 1 .
11 1000 30 30 0 1
12 1000  .  . 1 1
 1 9999  .  . 1 .
 2 9999  .  . 1 .
 3 9999  .  . 1 .
 4 9999 20 20 0 .
 5 9999 20 20 1 .
 6 9999 20 20 1 .
 7 9999 20 20 1 .
 8 9999 20 20 1 .
 9 9999 20 20 1 .
10 9999 20 20 0 1
11 9999  .  . 1 1
12 9999  .  . 1 1
end