So we're trying to analyze a screening design with some continuous and one nominal factor (12 levels). I am trying to figure out how to use the fit model/stepwise regression control and if feasible "rules" settings so that JMP does not matrix/assign levels and combinations to the nominal factor. I thought that the "whole" drop down in rules settings would take care of this, I thought that worked in the past, but alas, it keeps matrixing things. Any suggestions on how to overcome that? Thank you.