How to perfrom random effect model with Poisson regression with robust estimator?

[Gayatriv]

I am using proc glimmix but my model is not converging for two random effect however running without error but one warning:

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

proc glimmix data=remi.final_rem_gv method=quad;
class phc_name village gender;
model remission = gender/ s dist=poisson link=log offset=duration ddfm=none;
RANDOM INTERCEPT / SUBJECT=phc_name S CL TYPE=VC;
RANDOM INTERCEPT / SUBJECT=village(PHC_NAME) S CL TYPE=VC;
COVTEST /wald;
ods output estimates=est1 (keep= label ExpEstimate ExpLower Expupper Probt
rename=(ExpEstimate=or ExpLower=lower Expupper=upper Probt=pval)) CovParms=CovParms ;
run;

[jiltao]

You might use PARMS statement to specify your own starting values for the two variance components. For example,

parms (0.1)(0.05);

This could take several trial and error.

Thanks,

Jill