DOE L18 Design matrix

[#1]

Hi,

I am running a Screening Design of experiment to identify the key factors affecting the output variable. The DOE includes 7 factors. Out of which 6 factors have 2 levels and 1 factor has 3 level. JMP is suggesting the L18 Hunter design matrix. But if I review the design matrix two of the runs (15 & 17, 16 & 18 in attached picture) have the same factor settings. Can you suggest why this is the case?

/Tahir

[#2]

Hi Tahir,

It's just the way that it happens to pick the first 3-level column. In this case, Fit Model would use those 2 replicates for lack of fit. 

Another option is to add an additional 3-level factor, create the design, and drop the first 3-level column, using the second instead (this design shouldn't have any replicates).

Yet another choice is to bring the same factors into Custom Design, which may give you a different design structure.

Hope that helps,

Ryan

[#3]

Hi Ryan,

Thanks for the explanation but is it correct to say that even if i run the DOE with original settings, my analysis will still be valid and the main factors identified will be the significant factors?

/Tahir