Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: mi impute: ologit, mlogit, logit in one equation?


From   daniel klein <klein.daniel.81@googlemail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: mi impute: ologit, mlogit, logit in one equation?
Date   Tue, 17 May 2011 17:34:58 +0200

Hi Maria,

as the -mi impute- help file suggests, the only way to impute multiple
variables, if the pattern is not monotone, is to use multivariate
normal regression (or joint modeling). It is therefore not possible to
specify different models for different variables with -mi impute-.
That is because unklike -ice-, -mi impute- does not use chained
equations (or fully conditional specification).

As far as I know, fully conditional specification (FCS) performs well
in simulation studies, yet "[t]here appears to be [...] no really
satisfactory theory" (van Buuren et. al 2006) behind it, which migth
be the reason it is not implemented in offical Stata.
On the other hand -ice- works pretty well and may easily be combined
with offical -mi- (e.g. -mi ice-, -mi import-) so there may be little
need to impement an offical Stata command.

Would be nice to read some more sophisticated answers/explanations form others.

Best
Daniel

Van Buuren, S., Brand, J.P.L., Groothuis-Oudshoorn, C.G.M., Rubin, D.
B. 2006. Fully Conditional Specification in Multivariate Imputation.
Journal of Statistical Computation and Simulation, 76(12): 1049–1064

*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index