Statalist The Stata Listserver

[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Multiple imputation and BIC

From   "=?UTF-8?B?5a6L5pe25q2M?=" <[email protected]>
To   [email protected]
Subject   Re: st: Multiple imputation and BIC
Date   Thu, 12 Apr 2007 10:15:16 +0800

Dear Maarten,

Thank you very much, this is very helpful.


On 4/9/07, Maarten buis <[email protected]> wrote:
--- 宋时歌 <[email protected]> wrote:
> I am estimating a set of random effect model using XTMIXED. My data
> has quite some missing values on important covariates. I use ICE to
> do multiple imputation and use MIM to automatically combine the
> results.
> The estimation seems fine, but when I try to get postestimation
> results, I have some problems. Here is what I did:
> mim: xtmixed height age boy ...||id:age, mle
> mim: estat ic
> I got error message "command estat not recognised by mim; try
> specifying category option". Then I tried:
> mim, cat(fit) estat ic
> I got "cat(fit) estat ic not allowed"
> If I have log likelihood, I can compute BIC and AIC myself, but the
> MIM seems to suppress that part of model output.
> My question is: what should I do to get log likelihood value from
> models on multiply imputed data set using MIM?

There is good reason why -mim- suppresses the log likelihood: it does
not exist after multiple imputation. Multiple Imputation gives you an
approximation of the model for each parameter separately; simultaneous
tests of multiple parameters, or model fit statistics don't behave very
well after multiple imputations. Some corrections do exist, and some
are implemented in -mim-, but this is very much still an area of active
research. What is possible in terms of post-estimation can be seen in
the section "Examples - Post Estimation" in the help-file of -mim-.

Hope this helps,

Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

visiting address:
Buitenveldertselaan 3 (Metropolitan), room Z434

+31 20 5986715

New Yahoo! Mail is the ultimate force in competitive emailing. Find out more at the Yahoo! Mail Championships. Plus: play games and win prizes.*
*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index