Dear Statalisters
I need to calculate means and covariance matrices for use in structural
equation modelling. This means a lot of variables and missing data is a
serious problem. I need an efficient handling of missing data.
-hotdeck- imputation is not efficient when there are many variables with
missing data.
To the best of my knowledge (!), the most valid method to handle missing
data (MAR & MCAR) is to use Full Information Maximum Likelihood (FIML) or
Multiple Imputation (MI) techniques. I know that there is a set of tools for
analyzing MI-datasets available (SJ3-3 st0042) but there seem to be no tools
available for generating them.
I have also noticed a few .ado that handle missing data and uncertainties
with an EM algorithm (STB-55 sg139, STB-57 sbe38, SJ2-1 st0008). I know that
SPSS uses the same kind of algorithm to estimate covariance matrices and
means on data with missing values. LISREL also has an EM implementation. To
me (at least) it seems likely that a similar procedure would be possible to
.ado in Stata. But even if it was, the task is beyond my capabilities.
Is there anybody out there who has an efficient solution for missing data?
Is this a feature that Stata might consider to add to the official release?
Is it only me?
Thanks
Michael
-----------------
PhD-student
Department of Psychology
Stockholm University &
National Institute for
Psychosocial Medicine
PS
I know it can be done in SAS, SPSS, AMOS, LISREL (PRELIS), Mx and others ...
as well as with a few user written freeware NORM, amelial, EMCOV ... BUT! I
would rather ado it in Stata. And I have my reasons.
*
* 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/
