I'm working on some RSM designs for a split-split-plot experiment for two different design spaces. One design has a linear constraint between the very-hard-to-change (VHTC) and hard-to-change (HTC) variables, while the other does not. I created the designs using the Custom Design tool in JMP 12.2 (on a Mac), with varying numbers of whole plots and subplots to examine the effect on the power for the model terms involving the VHTC variable. I was also examining the effect of the split-plot variance ratio on the power for those same model terms under the Evaluate Design tool (under Advanced Options). I found that JMP would not allow me to change the split-plot variance ratio for the designs with the linear constraint, while I could do so for the unconstrained design. The default variance ratio settings are 1.0 for the whole plot and subplot variance ratios. With the constrained designed, I would attempt to change them, but the variance ratio values would always default back to a value of 1.0. Is this a "bug" in the program, or is there a technical reason why the variance ratio cannot be changed for a constraint between the VHTC and HTC variables? I do have the factors set properly as VHTC, HTC, and ETC, and I do have the Whole Plots and Subplots included in the model. Thanks.
