Using egen with a conditional

[Darren Anderson]

Hi guys,

Still a STATA rookie. I'm trying to figure out how to store a conditional median as a variable.

So I can get egen HL_cutoff = pctile( Day_max ), p(50) to work. But the trouble is I need it to be conditional - I need HL_cutoff to be the median of Day_max among people who have another variable, V1=0.

I've tried things like"
egen HL_cutoff = pctile( Day_max ), p(50) if V1==0
egen if V1==0 HL_cutoff = pctile( Day_max ), p(50)

EDIT: to be clear, I just what the computation of the median to be conditional. The variable should still apply to everybody.

What is the easiest way to do this?

Thanks!

[Andrea Discacciati]

clear
input y v1
1 0
2 0
3 0
4 0
5 0 
6 1
end

egen m1 = pctile(cond(v1 == 0, y, .))

See also Cox N., Speaking Stata: Compared with . . . , The Stata Journal (2011) 11, Number 2, pp. 305–314

[Nick Cox]

Thanks to Andrea for the clear example and the publicity. Let's be clear that Andrea's code puts the median in all observations while

 
 egen m1 = pctile(y) if v1 == 0

would put it only in those observations specified.