Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: Beta's and xtmixed
From
Richard Goldstein <[email protected]>
To
[email protected]
Subject
Re: st: Beta's and xtmixed
Date
Mon, 22 Oct 2012 10:44:12 -0400
I think that Jay is confused here; -xtmixed- is not a logit model;
-xtmelogit- is Stata's name for a mixed effects logit model
Rich
On 10/22/12 10:31 AM, JVerkuilen (Gmail) wrote:
> On Mon, Oct 22, 2012 at 5:56 AM, Filip Van Droogenbroeck
> <[email protected]> wrote:
>> Hello, everybody,
>>
>> I'm trying to get beta's for the -xtmixed- multilevel regression
>> command. There is no beta option for the xtmixed command. Anybody knows if there is a good reason for this? By using esttab I can get beta's after xtmixed. Can these beta's be trusted after the xtmixed command? Or are they only correct when doing normal linear regression?
>
> Standardized regression coefficients, which is what I assume you want,
> don't really mean the same thing in a mixed logit model as in linear
> regression, because while the variance of the x variables is
> meaningful, the variance of the linear predictor is set by choice of
> the logit link to pi^2/3. You really can't compare them across models
> readily. Usually you want to use standardized regression coefficients
> across models, so this is a problem.
>
> See Bauer, D.J. (2009). A note on comparing the estimates of models
> for cluster-correlated or longitudinal data with binary or ordinal
> outcomes. Psychometrika, 74, 97-105.
> http://www.unc.edu/~dbauer/manuscripts/Bauer_2009.pdf
>
> Chances are good you can get what you want with predicted
> probabilities computed on relevant scenarios, though that's not always
> true.
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/faqs/resources/statalist-faq/
* http://www.ats.ucla.edu/stat/stata/