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: gllamm or xtmixed models?
From
Alfonso Sánchez-Peñalver <[email protected]>
To
Stata List <[email protected]>
Subject
Re: st: gllamm or xtmixed models?
Date
Tue, 4 Feb 2014 10:20:22 -0500
True Maarten, which is why Statas likelihood ratio test in the -mixed- estimation when there are more than one variance to be estimated says that the LR test is conservative. If you click on the link there, which is the equivalent to typing - help j_mixedlr - you will find the explanation of what this means. In reading that explanation you will see that you indeed can perform an LR test for just one variance where the statistic is a 50:50 mixture of a chi2(0) and a chi2(1) distributions. This is what is discussed in the bulletin you referenced.
Best,
Alfonso Sánchez-Peñalver, PhD
Visiting Assistant Professor
Suffolk University
Senior Instructor
UMass Boston
On Feb 4, 2014, at 9:51 AM, Maarten Buis <[email protected]> wrote:
> On Tue, Feb 4, 2014 at 3:37 PM, Alfonso Sánchez-Peñalver wrote:
>> the first and their specifications are restricted models of the second specification. In the first one you are restricting the variance of the age slope to be zero, and in the third one you are restricting the mean of the age slope to be zero. You can easily test both restrictions with the individual significance test of each estimated parameter (mean and variance) from the second specification estimation.
>
> Testing the hypothesis that the variance is 0 is problematic becaus
> the null hypothesis is on the boundary of the parameter space (a
> variance cannot be negative, so 0 is the lowest possible value). See
> for example:
> Gutierrez, R., S. Carter, and D. M. Drukker. 2001. On boundary-value
> likelihood-ratio tests. Stata Technical Bulletin
> 60: 15-18. <http://www.stata-press.com/journals/stbcontents/stb60.pdf>
>
> -- Maarten
>
> ---------------------------------
> Maarten L. Buis
> WZB
> Reichpietschufer 50
> 10785 Berlin
> Germany
>
> http://www.maartenbuis.nl
> ---------------------------------
>
> *
> * 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/
*
* 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/