Evaluate a design build using Easy DOE

Question

We used Easy DOE from JMP 17 to generate a 5 factor/3 response design with 4 continuous (2 level) and 1 categorical (2 level) RSM with 36 runs.

Q1: Within the 36 runs there are 3 sets of duplicates. Is this a problem? We assume that since JMP designed it = no, but wanted to check for a better explanation.

Q2: Is there a way to perform a "Design Evaluation" of the Easy DOE generated design? In the Easy DOE process, there is not a "Design Evaluation" step (or button) unless we are missing something.

Thank y'all!

statman · Answer

Sorry, I'm a bit confused.  You have a 5 factor 2-level design.  With a Full factorial that is 32 treatments.  Not sure why you call this a RSM as RSM would require some ability to estimate departure from linear (e.g., 2nd order+, non-linear effects).

If you have identical treatments, those would be replicates that are randomized during the experiment.  Often those are used to estimate the MSE for statistical tests.

Back to evaluating the design...you need to start with; What is the objective of the experiment? What questions are you trying to answer?  What hypotheses are you trying to evaluate?  What model effects (and to what order) do you need to consider at this point in your investigation?  What inference space is necessary (don't forget about noise strategies)?  Where does this experiment fit in with your iterative investigation? (e.g., is this the first experiment and you are screening or is this some subsequent experiment and you are looking for an appropriate predictive model?)

Mark_Bailey · Answer

You can click the Export Data button on the Design tab. Then select DOE > Design Diagnostics > Evaluate Design.