GLIMMIX for repeated measures

aska_ujita — Sat, 04 May 2019 18:30:39 GMT

Hello there...

I have collected data of behavior of cows.

I collected reactivity of each cow (0=no reactivity and 1=yes) by day of lactation (day 1, 3, 7, 15, 30, 45 and 60).

Half of the animals are treated and other half are control.

I am using GLIMMIX procedure to analize that.

proc glimmix data=A;
class treatment po dl;
model reactivity=treatment dl po treatment*dl treatment*po/ dist=binomial link=logit solution;
random dl/ subject=cow;
run;

treatment is treated or control

po is the parturition order (primiparous and multiparous)

cow is the animal, I have 20 different cows per treatment

dl is the day of lactation (1, 3, 7, 15, 30, 45 and 60)

I am doing like this but didn't converge...

What is wrong??

Thank you very much!!

Aska.

Re: GLIMMIX for repeated measures

Rick_SAS — Tue, 07 May 2019 12:51:36 GMT

There can be many reasons why the model does not fit the data. I am not an expert on mixed models, but I think you should specify the RESIDUAL option so that GLIMMIX knows you are modeling an R side correlation within each cow's measurements:

random dl / subject=cow residual;

topic GLIMMIX for repeated measures in Statistical Procedures

GLIMMIX for repeated measures

Re: GLIMMIX for repeated measures