Inverse probability weights for continuous variable before survival analysis

[spencer keene]

Hello,

I would like to reweight the observations in my analysis using inverse probability weighting based on a continuous variable called time_interval. This should then help weigh the covariates in the subsequent Cox regression analysis.


The below code works but I would like to be sure that I did it properly.

regress time_interval ages sex bmi
predict p_time_interval
gen w=.
replace w=1/p_time_interval
summarize w

stset duration2[pweight=w], failure(infection==1)

[spencer keene]

Just to add that I have a similar question to this: Inverse Probability Weights - Statalist but just wondering if I can use a continuous variable (time_interval) instead of a binary one (insampm)