creating one variable combining two different observations

[Denise Vella]

I currently have treatment 1/0.

I generated the following code:

///This creates yearly sum for each surgeon ID for treatment = 1 and control = 1
gen control=1 if procedure = 0
gen treatment = 1 if procedure=1 
bys surgeonid year:egen yearly_treatment=sum(treatment)
bys surgeonid year:egen yearly_control=sum(control)

q: Is there a way you can think of to create one variable where for each surgeon eg 12, will have the no of treatment = 1 cases per year and control = 1 cases per year
The reason is because I am using IPTW and trying to compare treatment to control
I need to weight to compare those surgeons who have treatment cases experience per year eg = 5 with those who have control cases experiences per year = 5

I tried to create a ratio of yearly_treatment/yearly_control , but this creates several options therefore taking my logistic regression using iPTW ages to run

let me know if you have any other thoughts



[/CODE]

* Example generated by -dataex-. For more info, type help dataex
clear
input float(treatment_cases control_cases procedure surgeonid) str4 YEAR
1 1 1 12 "2005"
1 1 0 12 "2005"
0 1 1  1 "2006"
0 1 1 12 "2007"
1 1 1  5 "2008"
1 0 0 12 "2009"
1 0 0  6 "2010"
1 1 0  5 "2008"
end

[George Ford]

one variable cannot measure two different things