max over a category but storing results over all categories

[Angela Jiang]

Hi all,

I have code of the following form. h1paid and h2paid are variables I am trying to generate. h1paid stand for whether any of the helpers in h_category 1 is paid, h2paid stand for whether any of the helpers in h_category 2 is paid,

* Example generated by -dataex-. For more info, type help dataex
clear
input float id byte(helper_category helper_paid) float(h1paid h2paid)
1 1 1 1 2
1 1 1 1 2
1 2 0 1 2
1 2 0 1 2
end

The way I current achieve this is:

bys id: egen byte h1paid_int = max(helper_paid) if helper_categ ==1
bys id: egen byte h1paid = max(h1paid_int)

The problem is when I just do the first line I am maxing over the category I want but the result is also only stored in that category, what I want is to max over the category but store the results over all categories. In the end I want to keep only one observation per id with those summary variables h1paid h2paid. Is there a better way to do this? Thanks!

Best,
Angela

[Clyde Schechter]

by id, sort: egen h1_paid = ///
    max(cond(helper_categ == 1, helper_paid, .))
by id, sort: egen h2_paid = ///
    max(cond(helper_categ == 2, helper_paid, .))
collapse (first) h?_paid, by(id)

[Nick Cox]

For a write-up of the technique of Clyde Schechter
see Section 9 of https://journals.sagepub.com/doi/pdf...867X1101100210

The entire paper -- for most users -- will mix what they already know, really, with perhaps some new trickery.

[Angela Jiang]

Thanks Clyde and Nick. I will read the write up, thanks for sharing this, I am sure myself (and many others) will find it very useful.