2
$\begingroup$

I am currently doing GLIMMIX models in SAS. I choosed GLIMMIX as my data is count data, generally well described by a Poisson distribution.

My question: What assumptions can you test to ensure that your model is appropriate with GLIMMIX ?

From SAS, I learn the assumptions of GLIMMIX (http://support.sas.com/documentation/cdl/en/statug/63347/HTML/default/viewer.htm#statug_glimmix_a0000001402.htm) but I'm not sure I understand them. And I don't see how I could test them.

I also know that when the Pearson chi-square/DF is approximately 1, Poisson is good and when it's > 1, you should use a negative binomial.

Is there anything else ?

Thanks.

$\endgroup$

1 Answer 1

-1
$\begingroup$

After verification with people at my lab, I think I can answer my own question:

No, there is nothing else.

I will be very happy if anyone else has a different answer, but I think mine is correct.

$\endgroup$
2
  • $\begingroup$ Can you tell me what did you find out about testing the glimmix assumptions? $\endgroup$
    – user110691
    Commented Apr 1, 2016 at 18:37
  • $\begingroup$ @Marcel chi-square/df seems to be the only thing. You could check for outliers, but that's about it! Does it answer your question? $\endgroup$
    – Emilie
    Commented Apr 4, 2016 at 12:35

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.