Hello,
I am using IPTW for Cox (survival model - proc phreg) model. However, it's said that the confident interval generated by this procedure will create bias even using a robust estimate. One stimulation suggested using bootstrap to obtain 95%CI.
I have never done bootstrap before and I wonder how to integrate the inverse weight into the bootstrap. Could anyone please help me with this? Could I use inverse weight as survey weight to calculate bootstrap? My guess is these 2 weights are different, one by design and another is pseudo weight to create population balance.
Many thanks!
