WARNING: Obtaining minimum variance quadratic unbiased estimates as starting values for the covariance parameters failed.

[RubenPnx]

Hi all,

I am trying to get the seasonal estimates of the breeding population (% of the breeding population in a concrete period) from diverse Species in the different seasons of the year. To note that I have obtained this data from diferent longitudinal studies, having unbalanced data. So I am using:

Population to control the pseudoreplication

ID_study to control the different samplings,

Species (Genus*Family) to take into account the taxonomy

However, I would like to take into account the density using it as offset.  The problem is that when I include this term in orocethe SAS code I get the following answer: "WARNING: Obtaining minimum variance quadratic unbiased estimates as starting values for the covariance parameters failed".

proc glimmix data=breeding method=laplace plots=(ResidualPanel(conditional marginal));

class Season Species Genus Family Population ID_study Month ;

model Breed_population =  Season Month (Season)

  /dist=beta offset=Density

  ddfm=none covb s;

random intercept / subject = Population (ID_study);

random intercept / subject= Species (Genus*Family);

nloptions maxiter=150;

lsmeans Season/cl diff PLOT=diff (NOabs center) adjust=tukey;

lsmeans Season/cl ilink;

run;

Anyone know how I can overcome this problem?

Thanks,

Rubén

[RubenPnx · Accepted solution]

The warning was due to the amount of decimals of the offset. When I reduce it to 2-3 decimals I could to carry out the model.

I hope it may help someone

Sorry for the inconvences!!!