Re: st: svyset simple question

[Steven Samuels <sjsamuels@gmail.com>]

You do have to specify a PSU.  The study documentation will tell you  
what variable to use. Otherwise the within-imputation estimated  
variance will be way off (much too small).  You should also specify a  
stratum variable if it is available, or variances will be bigger than  
needed.  You can ignore the fpc.


In -svyset-, unlike regression commands, the "default" settings can be  
completely wrong.

Steve

On Jan 20, 2011, at 1:10 AM, Hassan Chawdhury wrote:

Dear All,
I am freaking out here. Could anyone please help me out?

I have done 5 imputations  and then did---mi svyset [pw=weightvar]---  
for the CCHS 3.1 survey data set by Statistics Canada. Then I ran all  
my regressions with the following commands ---mi estimate: post: svy:  
regress/ologit/nbreg---.

I have also done all my regressions without any imputation and with  
svyset [pw=weightvar] for the CCHS 3.1. However, this is a separate  
data file without any imputaion for the same survey.

For none of the svyset, I specified "psu", "strata", or "fpc" since I  
did not think that it would be important. Do I have to specify them?  
Using the Stata 11 IC and Stata 11 SE, I am still getting the results  
with the defaults. Are they different or wrong?

Any help is much appreciated. Thanks a lot.
Hassan Chawdhury



*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/