Creating a new variable

[Varsha Vaishnav]

Hello. I want to create a variable that shows average attainment at the household level. The weights for averaging across individuals within a household are defined as follows: Attainment score of individual i in the dimension under consideration divided by the total attainment score of that dimension. How can I carry it out?
I have attached an example dataset below.

* Example generated by -dataex-. To install: ssc install dataex
clear
input byte(hh_id a total_a)
11 4 9
12 4 9
13 6 9
21 6 9
22 6 9
23 6 9
31 7 9
32 6 9
33 5 9
end

[George Ford]

so you just want 4/9 for id 11?

[Varsha Vaishnav]

That will be the weight for averaging the variable 'a'. Basically, I want 4/9(4)+4/9(4)+6/9(6) for household 1, and likewise for other households.

[George Ford]

how big does hh_id get? and are there more than 9 members in a household? you're going to need a hh_id that does not include the person number.

[George Ford]

how big does hh_id get? and are there more than 9 members in a household? you're going to need a hh_id that does not include the person number.

here's a sketch

tostring hh_id, g(hh_ids)   
g hid = substr(hh_ids,1,1) //identify household, won't work when bigger than 1 digit 
g shr = a/total_a    // create share to sum
bys hid: egen wgt = sum(shr) //sum to weight
drop shr

[Varsha Vaishnav]

Hello, George. hh_id is a concatenation of the cluster and household numbers. At max, it is a 9-digit number. And yes, there are households in the dataset with more than 9 members.

[George Ford]

you got to get a unique household id. then you can egen within that household id.

[Varsha Vaishnav]

Thank you, George.