Dear JMP Community,
Sorry if this has already been discussed, I think it has been, but I don't seem to find the relevant thread.
I am attempting to run either nominal logistic or generalized linear model with binomial distribution (in this particular case if I am not mistaken these are pretty much the same thing) with a categorical response variable and a nested design, in which I have the outer factor I am most interested in and a nested inner factor which is basically part of random variation. Of course I want to test the outer factor against the inner one, so I specify the nested inner factor as random. But I cannot, because these models do not take random effects.
Is there a fundamental reason why random effects cannot be evaluated in these models? Perhaps this option exists in newer versions of JMP? (I'm still on JMP 10). Or in JMP Pro?
If not, what are my options? I do not want to pool / ignore the inner factor because there is for sure a lot of variance among its levels and I am inflating my power.
Maybe I should do some sort of resampling, randomly choosing 1 replicate from each nested factor category?
Many thanks for any suggestions!
