f



PROC MIXED - Estimated G matrix is not positive definite.

Hi,

I'm getting the following message: "Estimated G matrix is not positive
definite" when I run the following:

proc mixed data=mydata;
     class ID;
     model Y= / s ;
	 random intercept / subject=ID;
   run;

----------
Does this occur when the within variability is much larger than the
between variability? The whole point of this analysis is to measure
the covariance parameter estimates (residual and intercept). Is it
reasonable to just interpret the random intercept as being 0?

Thoughts?

Ryan
0
9/4/2008 9:04:33 PM
comp.soft-sys.sas 142827 articles. 4 followers. Post Follow

3 Replies
981 Views

Similar Articles

[PageSpeed] 44

On Sep 4, 5:04 pm, Ryan <Ryan.Andrew.Bl...@gmail.com> wrote:
> Hi,
>
> I'm getting the following message: "Estimated G matrix is not positive
> definite" when I run the following:
>
> proc mixed data=mydata;
>      class ID;
>      model Y= / s ;
>          random intercept / subject=ID;
>    run;
>
> ----------
> Does this occur when the within variability is much larger than the
> between variability? The whole point of this analysis is to measure
> the covariance parameter estimates (residual and intercept). Is it
> reasonable to just interpret the random intercept as being 0?
>
> Thoughts?
>
> Ryan

What does your data look like? Do you have more than one subject per
group as indicated by the ID variable?
Mark
0
mschult (17)
9/4/2008 9:12:37 PM
On Sep 4, 5:12=A0pm, msch...@bu.edu wrote:
> On Sep 4, 5:04 pm, Ryan <Ryan.Andrew.Bl...@gmail.com> wrote:
>
>
>
>
>
> > Hi,
>
> > I'm getting the following message: "Estimated G matrix is not positive
> > definite" when I run the following:
>
> > proc mixed data=3Dmydata;
> > =A0 =A0 =A0class ID;
> > =A0 =A0 =A0model Y=3D / s ;
> > =A0 =A0 =A0 =A0 =A0random intercept / subject=3DID;
> > =A0 =A0run;
>
> > ----------
> > Does this occur when the within variability is much larger than the
> > between variability? The whole point of this analysis is to measure
> > the covariance parameter estimates (residual and intercept). Is it
> > reasonable to just interpret the random intercept as being 0?
>
> > Thoughts?
>
> > Ryan
>
> What does your data look like? Do you have more than one subject per
> group as indicated by the ID variable?
> Mark- Hide quoted text -
>
> - Show quoted text -

Hi Marc, Thank you for responding! Yes. I have several observations
per subject as indicated by the ID variable--there is some missing
data. -Ryan
0
9/4/2008 9:39:52 PM
On Sep 4, 5:12=A0pm, msch...@bu.edu wrote:
> On Sep 4, 5:04 pm, Ryan <Ryan.Andrew.Bl...@gmail.com> wrote:
>
>
>
>
>
> > Hi,
>
> > I'm getting the following message: "Estimated G matrix is not positive
> > definite" when I run the following:
>
> > proc mixed data=3Dmydata;
> > =A0 =A0 =A0class ID;
> > =A0 =A0 =A0model Y=3D / s ;
> > =A0 =A0 =A0 =A0 =A0random intercept / subject=3DID;
> > =A0 =A0run;
>
> > ----------
> > Does this occur when the within variability is much larger than the
> > between variability? The whole point of this analysis is to measure
> > the covariance parameter estimates (residual and intercept). Is it
> > reasonable to just interpret the random intercept as being 0?
>
> > Thoughts?
>
> > Ryan
>
> What does your data look like? Do you have more than one subject per
> group as indicated by the ID variable?
> Mark- Hide quoted text -
>
> - Show quoted text -

Here's an example of the type of data I have:

ID      Product      Y
1             1          45
1             2          23
1             3          48
2             2          55
2             3          12
..
..
..

Note that ID 2 is missing info for product 1.

0
9/4/2008 9:48:32 PM
Reply: