I'm trying to set up an analysis where I want to do linear regression on a fixed effect, say tablet weight. I am running an assay where I take several tablets of weight 20 and 40. The assays are random, say 1-5. To make things a bit more complicated, within each assay there are several aliquots that serve as subsamples or observations. I think the model is
Y(i,j,k) = b_0 + B_1 * W(i) + A(j) + E(k[i,j])
where A (assay) is random and nested in W
E is random and nested in W and A.
I understand that I can set up W as nominal and get the correct model and then, off-line, figure out the slope and intercept. But why is this model considered invalid in JMP. JMP says I can't have nested continuous terms.