[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: Zero-inflated Poisson/Negative Binomial for survey data
Yu, Binbing (IMS) <YuB@imsweb.com> wants to know if -svy- is supported by the
-zip- and -zinb- commands:
> I know that STATA can fit various regression to complex survey data as well
> as zero-inflated Poisson (ZIP) and negative binomial (ZINB) for simple
> random sample (SRS). I wonder if STATA can do ZIP/ZINP for complex survey
The following is true for Stata 8 and Stata 9, although Stata 8 requires an
extra -score()- option in the call to -zip- (and -zinb-).
While -zip- and -zinb- do not yet have a counterpart for survey data analysis,
you can use -suest- with the -svy- option to get linearized variance estimates
according to the design characteristics in your -svyset-.
For example say the sampling weight variable is -sampwgt-, then you can do
. svyset ... [pweight=sampwgt], ...
. zip ... [iweight=sampwgt], ...
. suest ., svy
The name of the 'pweight' variable in -svyset- must match the 'iweight'
variable used in -zip- (and -zinb-).
* For searches and help try: