Covtest in Proc Glimmex

[Rupert]

Hey,

I am having issues with COVTEST not running in  Proc Glimmix. I am running the following code in SAS/Stat 9.2.

    proc glimmix data=schooltest ;

    class X1 X2  School Classroom;

    model Test(event='1') = X1 X2 X3 /solution corrb covb oddsratio link=logit dist=binary;

    random int/subject=Classroom type=chol solution g;

    random int/subject=School(Classroom) type=chol solution g ;

    covtest 'G reduced' zerog;

run;

There is no output statement for the COVTEST in the output screen nor is there an error showing in the log file.

I am wondering if I have programmed the code wrong or if there is" Hotfix" update that I have not found that would correct this.

Thanks

[SteveDenham]

It is possible that there are convergence problems of some sort--I have run into this with covtest homogeneity.

Try adding a restart option, e.g. covtest 'G reduced' zerog/restart;

I can't guarantee this will help, but it would be the first thing I would try in this case.  Number two would be to change the test to that for a diagonal G matrix.  With the full parameterization under the Cholesky structure, this may at least run.

Good luck.

Steve Denham

[Rupert]

Hello Steve,

Thanks for the quick response. I tried both of your suggested methods but no luck. I think my problem is that I can not even get the covtest statement to become active in the editor screen. The code remains 'red' for the covtest statement instead of turning the usual "green" as per the other statements in the program. I am running version SAS/STAT 9.2 (TS2M2) and was wondering if maybe there is a bug in this version of SAS that is preventing me from utilizing the covtest statement.

Thanks.

[SteveDenham]

Since I am running on 9.3, I'm not sure which options were viable in 9.2 and 9.22 (I am pretty sure that there was an update).  The code is 'red' in 9.3 as well. It looks like all of the options are available in 9.22, so it should be working.

/turn on grasping at straws mode

So, just how big is your G matrix, and how many obs do you have?  Maybe it is an estimation problem.

This one might need a visit to tech services.

Steve Denham

[lvm · Accepted solution]

COVTEST works properly starting in 9.2, including the release you have. The red color occurs, even though the statement is supported. You didn't say what you are getting for the estimated variance-covariance matrix. I am guessing that the program is not able to estimate the G-side parameters (or they are 0). If you don't have var-cov parameters, there is nothing for COVTEST to actually test. I think your model is over-parameterized in terms of random effects. You are asking for an unstructred matrix nested within another unstructured matrix (with the constraint imposed by the chol option). I doubt if this will work.