Prev: proc append to combine multiple tables on Unix
Next: SAS and ODBC dates
From: Ching on 19 Jan 2010 22:12
Hi everyone.

I'm currently doing a 3-level multilevel model and wanting to know if
it is still correct to report the result despite the below notes
produced. If not, do I have to respecify my model?

NOTE: Convergence criteria met.
NOTE: Estimated G matrix is not positive definite.
NOTE: Asymptotic variance matrix of covariance parameter estimates has
been found to
be singular and a generalized inverse was used. Covariance
parameters with zero
variance do not contribute to degrees of freedom computed
by DDFM=SATTERTH.

Thanks for your advice. Have a great day.
From: Dale McLerran on 20 Jan 2010 12:46
--- On Tue, 1/19/10, Ching <kcwong5(a)GMAIL.COM> wrote:

> From: Ching <kcwong5(a)GMAIL.COM>
> Subject: Notes produced from proc mixed
> To: SAS-L(a)LISTSERV.UGA.EDU
> Date: Tuesday, January 19, 2010, 7:12 PM
> Hi everyone.
>
> I'm currently doing a 3-level multilevel model and wanting to know if
> it is still correct to report the result despite the below notes
> produced. If not, do I have to respecify my model?
>
> NOTE: Convergence criteria met.
> NOTE: Estimated G matrix is not positive definite.
> NOTE: Asymptotic variance matrix of covariance parameter estimates has
> been found to be singular and a generalized inverse was used.
> Covariance parameters with zero variance do not contribute to
> degrees of freedom computed by DDFM=SATTERTH.
>
> Thanks for your advice. Have a great day.
>

Ching,

It is likely that the random effect variance for one of the
levels in your hierarchical model was estimated to be zero.
If that is the case, then yes, you can report the results
for this mixed model. Your model might be overparameterized
in terms of random effect specifications. But in most
circumstances, it does not hurt you much to be overparameterized
in the random effects specification.

Given the note about degrees of freedom obtained employing
the Satterthwaite method, it would appear that you would not
be penalized at all for using the full hierarchical model.
In general, it is better to report results for the full
hierarchical model than to reduce the three level model to
a two level model.

Dale

---------------------------------------
Dale McLerran
Fred Hutchinson Cancer Research Center
mailto: dmclerra(a)NO_SPAMfhcrc.org
Ph: (206) 667-2926
Fax: (206) 667-5977
---------------------------------------
 | 
Pages: 1
Prev: proc append to combine multiple tables on Unix
Next: SAS and ODBC dates