Proc Glimmix covtest warning

DIHS — Tue, 24 Nov 2015 17:50:52 GMT

Hi,

I am trying to compare covariance parameters using covtest with the general specification in Proc Glimmix. Without using covtest everything seems to work. With the covtest statement I get the warning ‘The initial estimates did not yield a valid objective function’ and I get no p value for the comparison of the covariance parameters. Does anyone have any idea why this is happening? My code essentially looks at the covariance between two traits (age at maturity and size) with family as the subject. All the covariance parameters are positive except the two for the covariance between the two traits (one for each of the two treatments), i.e the families which matured later were smaller in both treatments. I would like to know if the treatment changes the covariance between the traits with family as the subject. My code is:

proc glimmix data=mydata;

class family trait ID treatment;

model response=trait treatment;

random trait / subject = family group= treatment type=un;

random _residual_/subject=family(ID) group=treatment type=un;

covtest general . 1 . . -1 /estimates;

run;

Thanks for your time

Re: Proc Glimmix covtest warning

plf515 — Tue, 24 Nov 2015 21:32:01 GMT

In this paper by Kiernan, Tao and Gibbs (all with SAS), there is a section about this error. I tried to cut and paste that section, but the formatting got all messed up.

Re: Proc Glimmix covtest warning

SteveDenham — Wed, 30 Dec 2015 16:31:55 GMT

This may be something as simple as replacing the '.'s in the GENERAL option with zeroes:

covtest general 0 1 0 0 -1;

Steve Denham

topic Proc Glimmix covtest warning in Statistical Procedures

Proc Glimmix covtest warning

Re: Proc Glimmix covtest warning

Re: Proc Glimmix covtest warning