I’m currently using weights in my survival analysis after using IPTW in my observational dataset.
Where treatment weight = 1 and control weights vary.
Question: do I use weights with my predict command, or if it’s incorporated in stset I don’t need to reuse the attweight variable ?
Code:
stset survivalt [pw=attweight], failure(revised==1) scale(12) exit (time 10) //proceed to use predict gen(agercs) center(60) stpm2 hormon agercs* pr_1, scale(hazard) df(4) eform range tt 0 5 100 predict s0, survival timevar(tt) zeros ci
The reason I ask is because my predict graphs without incorporating attweight at stset is v different to my predict graph following incorporation attweight at stset
I wonder if the difference is due to the fact I need to include -attweight- at the predict level.
