st: estimation after multiple imputation: difference between regular estimation and post-imputation estimation

[Mandy fu <mandy.fu1@gmail.com>]

Hello everyone,

I was wondering if anyone could give some hint what causes the
differences of using regular estimation and post-multiple imputation
estimation.
Let me use an example. After multiple imputation, I know I should use
--mim: xtmixed----- instead of simply --xtmixed--- to conduct the
growth curve analysis. But I am not very clear about the reason. Could
anyone give some help here?

Additionally, based on previous discussion in statalist , using
----mim:xtmixed--- is very tricky to get the model level
characteristics, such as log likelihood. If I would like to compare
the fit-goodness of several models , does it make sense to use the
log-likelihood for the models from using ---xtmixed--on the imputed
data sets as an approximate?

Thanks for your time!

Best regards,
Mandy
*
*   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/