Counting the number of times a variable is 'present' in a household for the panel years

[Glory Sibale]

Hello everyone.

I have panel data collected in 4 years, at a 3-year interval. I have a dummy, for 1 if the desired characteristic (growing maize) is present, 0, otherwise. In my data, some households grew maize in some of the years and not all of the four years, some did not grow any maize at all, and some did in all the four years. I want to keep only those that grew at least once in the four years. Please help me with the command for this.

[Carlo Lazzaro]

Glory:
do you mean something along the following lines?

bysort panelid (timevar): egen wanted=count(growing_maize)
keep if wanted>=1

[Andrew Musau]

Glory:
do you mean something along the following lines?

bysort panelid (timevar): egen wanted=count(growing_maize)
keep if wanted>=1

The -count()- function of egen counts nonmissing values. Either -total()- or -max()- will do here, assuming the maize variable is a 0/1 indicator. See this FAQ for more: https://www.stata.com/support/faqs/d...ble-recording/

clear
set obs 5
gen id=_n
expand 5
bys id: gen year= 1990+_n
set seed 06012022
gen maize= runiformint(0,1) & inlist(id, 2,4)
l, sepby(id)
*START HERE
bys id: egen tokeep= max(maize)
keep if tokeep
l, sepby(id)

Res.:

. l, sepby(id)

     +-------------------+
     | id   year   maize |
     |-------------------|
  1. |  1   1991       0 |
  2. |  1   1992       0 |
  3. |  1   1993       0 |
  4. |  1   1994       0 |
  5. |  1   1995       0 |
     |-------------------|
  6. |  2   1991       1 |
  7. |  2   1992       1 |
  8. |  2   1993       1 |
  9. |  2   1994       0 |
 10. |  2   1995       1 |
     |-------------------|
 11. |  3   1991       0 |
 12. |  3   1992       0 |
 13. |  3   1993       0 |
 14. |  3   1994       0 |
 15. |  3   1995       0 |
     |-------------------|
 16. |  4   1991       1 |
 17. |  4   1992       1 |
 18. |  4   1993       1 |
 19. |  4   1994       1 |
 20. |  4   1995       0 |
     |-------------------|
 21. |  5   1991       0 |
 22. |  5   1992       0 |
 23. |  5   1993       0 |
 24. |  5   1994       0 |
 25. |  5   1995       0 |
     +-------------------+

.
. *START HERE

.
. bys id: egen tokeep= max(maize)

.
. keep if tokeep
(15 observations deleted)

.
. l, sepby(id)

     +----------------------------+
     | id   year   maize   tokeep |
     |----------------------------|
  1. |  2   1991       1        1 |
  2. |  2   1992       1        1 |
  3. |  2   1993       1        1 |
  4. |  2   1994       0        1 |
  5. |  2   1995       1        1 |
     |----------------------------|
  6. |  4   1991       1        1 |
  7. |  4   1992       1        1 |
  8. |  4   1993       1        1 |
  9. |  4   1994       1        1 |
 10. |  4   1995       0        1 |
     +----------------------------+

.

[Glory Sibale]

Glory:
do you mean something along the following lines?

bysort panelid (timevar): egen wanted=count(growing_maize)
keep if wanted>=1

Yes Carlo,

But I did not necessarily want it to

count

Since counting would then give me a total of the times they have grown maize. Instead, I have used

max

Thank you very much for your help.

[Glory Sibale]

The -count()- function of egen counts nonmissing values. Either -total()- or -max()- will do here, assuming the maize variable is a 0/1 indicator. See this FAQ for more: https://www.stata.com/support/faqs/d...ble-recording/

clear
set obs 5
gen id=_n
expand 5
bys id: gen year= 1990+_n
set seed 06012022
gen maize= runiformint(0,1) & inlist(id, 2,4)
l, sepby(id)
*START HERE
bys id: egen tokeep= max(maize)
keep if tokeep
l, sepby(id)

Res.:

. l, sepby(id)

+-------------------+
| id year maize |
|-------------------|
1. | 1 1991 0 |
2. | 1 1992 0 |
3. | 1 1993 0 |
4. | 1 1994 0 |
5. | 1 1995 0 |
|-------------------|
6. | 2 1991 1 |
7. | 2 1992 1 |
8. | 2 1993 1 |
9. | 2 1994 0 |
10. | 2 1995 1 |
|-------------------|
11. | 3 1991 0 |
12. | 3 1992 0 |
13. | 3 1993 0 |
14. | 3 1994 0 |
15. | 3 1995 0 |
|-------------------|
16. | 4 1991 1 |
17. | 4 1992 1 |
18. | 4 1993 1 |
19. | 4 1994 1 |
20. | 4 1995 0 |
|-------------------|
21. | 5 1991 0 |
22. | 5 1992 0 |
23. | 5 1993 0 |
24. | 5 1994 0 |
25. | 5 1995 0 |
+-------------------+

.
. *START HERE

.
. bys id: egen tokeep= max(maize)

.
. keep if tokeep
(15 observations deleted)

.
. l, sepby(id)

+----------------------------+
| id year maize tokeep |
|----------------------------|
1. | 2 1991 1 1 |
2. | 2 1992 1 1 |
3. | 2 1993 1 1 |
4. | 2 1994 0 1 |
5. | 2 1995 1 1 |
|----------------------------|
6. | 4 1991 1 1 |
7. | 4 1992 1 1 |
8. | 4 1993 1 1 |
9. | 4 1994 1 1 |
10. | 4 1995 0 1 |
+----------------------------+

.

Yes, Andrew. I went to the link you have attached to understand more, and it has worked. I have used the code below, and have finally kept the non-zeros. Thank you very much for your help.

egen farmer = max(maize), by(HHno)