Hi, sorry I didn't see this until this morning. Tell me one thing about your data: Is each treatment considered an independent event, or are you measuring the cumulative effect of the treatment over time? The reason I ask, is that if I were using this data, I would treat this essentially as a probability model, like Logit or Probit, which asks how does the probability of Reactivity=1 change with being in the treatment group. But that requires independence across the time dimension.
Using PROC GLIMMIX would look something like this (I haven't tested this code, so you might have to play with it a bit):
PROC GLIMMIX DATA=yourdata;
CLASS cow day;
MODEL reactivity(event="1") = group / SOLUTION;
RANDOM intercept / subject=cow;
RANDOM intercept / subject=day;
RUN;
I also found this explanation of a binomial version of this at this link: https://support.sas.com/documentation/cdl/en/statug/63033/HTML/default/viewer.htm#statug_glimmix_a00...
proc glimmix data=yourdata;
CLASS cow; model reactivity/n = group / solution; random intercept / subject=cow; run;
I'm not as familiar with this type of regression, so I don't know that I can help you much with the binomial version. In these types of models, the interpretation of the coefficients is somewhat tricky. SAS doesn't offer a built in post-processing procedure to derive average marginal effects from Logit models. That may not be important for your discipline, but it's what social scientists typically use to measure the effect of a variable on the probability of "success." Hope that helps!
