[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: survey discrete-time survival
Philip Ender <firstname.lastname@example.org> has a question about -pweight-s and survival
> We have a client with cross sectional survey data with psu's, strata,
> and pweights who wants to do discrete-time survival analysis. She has
> expanded the data to person-period form with an unequal number of
> periods per individual.
> What should she do with the pweights; leave them as is; or divide
> them by the number of periods?
Let's assume the dataset has been -stset- using the -id()- option to identify
each person. If 'expanded the data to person-period form' is essentially the
same as using -stsplit-, then leave the -pweight-s as they are. Otherwise
we would need more information to comment any further, although we suspect
that the -pweight-s should still remain untouched.
* For searches and help try: