R help - Nov 2008 - ordered logistic regression of survey data with missing variables

Hello:
I am working with a stratified survey dataset with sampling weights  
and I want to use multiple imputation to help with missingness.

1. Is there a way to run an ordered logistic regression using both a  
multiply imputed dataset (i.e. from mice) and adjust for the survey  
characteristics using the weight variable? The Zelig package is able  
to do binary logistic regressions for survey data and handle the  
missing data (logit.survey) but I could not find a way to do both for  
an ordered logistic model.

2.  I assume I should use the weights in the process of creating the  
multiply imputed datasets as well. Is there a way to do so in any of  
the multiple imputation packages in R?


Thanks so much

Thomas Soehl
---
Department of Sociology - UCLA
Los Angeles, CA 90095
soehl@ucla.edu




	[[alternative HTML version deleted]]

Thomas,

You might want to read this first:
http://www.citeulike.org/user/ctacmo/article/1269394. And this, too:
http://www.citeulike.org/user/ctacmo/article/637812.

On 11/3/08, Thomas Soehl <soehl at ucla.edu>
wrote:
> Hello:
>  I am working with a stratified survey dataset with sampling weights
>  and I want to use multiple imputation to help with missingness.
>
>  1. Is there a way to run an ordered logistic regression using both a
>  multiply imputed dataset (i.e. from mice) and adjust for the survey
>  characteristics using the weight variable? The Zelig package is able
>  to do binary logistic regressions for survey data and handle the
>  missing data (logit.survey) but I could not find a way to do both for
>  an ordered logistic model.
>
>  2.  I assume I should use the weights in the process of creating the
>  multiply imputed datasets as well. Is there a way to do so in any of
>  the multiple imputation packages in R?
-- 
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.

You can analyse multiple imputations with the survey and mitools  package, 
and there is a toy example including ordinal logistic regression at 
http://faculty.washington.edu/tlumley/survey/svymi.html

If am I reading their documentation correctly, 'mice' creates what Rubin
calls 'proper imputations', for which the calculations are correct 
('improper' imputations are more efficient but the simple variance 
calculations are wrong).

The bootstrap approach that Stas Kolenkov pointed out looks attractive as 
long as it is computationally feasible.


 	-thomas


On Mon, 3 Nov 2008, Thomas Soehl wrote:
> Hello:
> I am working with a stratified survey dataset with sampling weights
> and I want to use multiple imputation to help with missingness.
>
> 1. Is there a way to run an ordered logistic regression using both a
> multiply imputed dataset (i.e. from mice) and adjust for the survey
> characteristics using the weight variable? The Zelig package is able
> to do binary logistic regressions for survey data and handle the
> missing data (logit.survey) but I could not find a way to do both for
> an ordered logistic model.
>
> 2.  I assume I should use the weights in the process of creating the
> multiply imputed datasets as well. Is there a way to do so in any of
> the multiple imputation packages in R?
>
>
> Thanks so much
>
> Thomas Soehl
> ---
> Department of Sociology - UCLA
> Los Angeles, CA 90095
> soehl at ucla.edu
>
>
>
>
> 	[[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>
Thomas Lumley			Assoc. Professor, Biostatistics
tlumley at u.washington.edu	University of Washington, Seattle