Re: st: pseudo R2 with multiple imputed data

[gresch@wzb.eu]

thanks for your advice,

I guess I will just use the mean value of the pseudo R2s and include a
footnote with the proceeding so it is transparent.
(and as reply to a former suggestion of Leandro Brufman: fitstat does not
work with mim)

best
Cornelia


> --- On Fri, 1/10/10, gresch@wzb.eu  wrote:
>> I am using -mim- to run some logistic regression models
>> with multiple imputed datafiles (using the cluster-option
>> since I have multilevel-data). I would like to compute
>> any of the pseudo R2s to illustrate differences between
>> some models. Does anybody of you know, how to get that?
>>
>> Up to now there is no output of the pseudo R2  by using
>> mim. As I read, for the regular R2 of a linear regression
>> I cannot just compute the mean value (Rubins Rule) due to
>> the distribution of R2. Here I need to fishers-z-transform
>> the different R2s first. So I guess there are similar
>> problems and maybe also solutions for the pseudo R2?
>
> My impressison is that pseudo-R2s are just rather approximate
> rule of thumb values, so I doubt whether anyone had a good
> look at the sampling distribution of these statistics. As long
> as you do not take those statistics too seriously I guess you'll
> be fine by averaging them or some transformation.
>
> -- Maarten
>
> --------------------------
> Maarten L. Buis
> Institut fuer Soziologie
> Universitaet Tuebingen
> Wilhelmstrasse 36
> 72074 Tuebingen
> Germany
>
> http://www.maartenbuis.nl
> --------------------------
>
>
>
>
> *
> *   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/