RE: st: MI mvn with categorical data

[Cameron McIntosh <cnm100@hotmail.com>]

Andrea,You know, a response that suggests you use a different procedure should not be seen as a criticism that needs to be rebutted. People are genuinely trying to help you find the most appropriate method. And you have no way of knowing for sure that the MVN assumption actually works in your case (e.g., this isn't a simulation where you actually know the missing values), so it would be prudent to go with the method specifically designed to handle the mixed mode case. Cam

----------------------------------------
> Subject: Re: st: MI mvn with categorical data
> From: mac.stata@gmail.com
> Date: Fri, 30 Sep 2011 16:35:26 +0200
> To: statalist@hsphsun2.harvard.edu
>
> MVN is generally seen to be superior to ICE/chained which is why I try to go that rout. The procedure I cite is exactly for the case I mention. It is in no way inappropriate, neither for my data nor for MVN. I just cannot get it to work with Stata, properly.
>
> On Sep 30, 2011, at 16:08 , Stas Kolenikov wrote:
>
> > If you have Stata 12+, you'd want to explore -mi impute chained-. If
> > you have Stata 11.99-, you would want to take a look at -ice- by
> > Patrick Royston. There is little sense to use the tool that you
> > yourself know to be inappropriate. If Enders ignores Stata in his
> > book, it means you have to explore it on your own and find the best
> > performing methods that may not be even mentioned in that book.
>
>
> *
> * 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/