Hello,
I am working on an analysis with a sample that was drawn using a stratified cluster design. Therefore, I have to use the survey commands (svy) to set the proper analytic weight, strata, and primary sampling unit. I also need to use the subpop command to ensure that the standard errors are estimated correctly. Specifically, I am estimating a series of logistic regression equations.
Ex:
svylogit y x1 x2 x3, subpop(samp) or
I just recently learned how to impute missing data using the ice command. Once I impute the missing values to create five imputed data sets (m = 5), I try to estimate my model using the micombine feature:
micombine svylogit y x1 x2 x3, subpop(samp) or
I receive the warning:
Warning: svylogit is not a certified regression command for micombine.
micombine will continue mechanically, but correct results are not guaranteed.
You must take responsibility that Rubin's rules are appropriate here.
I assume from this message that micombine cannot handle survey commands (yet) in Stata. I am using version 8.0.
Is there a way to estimate my model correctly with the survey commands and subpop statement simultaneously on all five imputed data sets? Or do I need to estimate the models separately on each data set and pool the estimates by hand?
Any advice or suggestions are appreciated.
Thanks!
Robert
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
