Bysort command to drop observations

[Jenna Kerry]

Hello,

I am using bysort command in Stata 18.0. My data set looks like this:

* Example generated by -dataex-. For more info, type help dataex
clear
input float(Family position_in_family age)
1 1 30
1 2 30
1 3  5
1 4 24
2 1 37
2 2 35
2 4 44
2 5 55
3 1 45
4 1 44
4 2 44
4 3 18
5 1 34
5 2 45
5 3  4
5 4 34
6 1 54
6 2 45
6 3 21
end

Here, I have family data. The position in family implies, 1 = husband, 2= wife, 3 = their kid, 4,5 = other family members present in the household.

I have the ages of the family members.

Now I want to drop those families where the age of the kid is above 6 years old. I am applying the bysort command.

bysort family(position): drop if age[3] > 6 & age[3] < 71

Please note, here 70 is the highest age of the child in my data set.

I am not sure why this code isn't working. I still see some family where the kid is much older than 6.

I would really appreciate if anyone can help!

[Andrew Musau]

bysort family: egen tag= max(position_in_family==3 & age>6)
drop if tag

Note that the above says that the individual is a kid (position_in_family==3) and age>6. This will also drop families that have at least one kid with a missing observation on age. If you do not want this:

bysort family: egen tag= max(position_in_family==3 & age>6 & age<.)
drop if tag

The logic of using the -max()- function of egen is explained in https://www.stata.com/support/faqs/d...ble-recording/.

[Jenna Kerry]

Dear Andrew Musau,

Thank you so much for the code. It worked perfectly.
Also thank you for sharing the reading!