topic Re: Modification of values of factor in DOE in Discussions

Modification of values of factor in DOE

bbenny7 — Tue, 14 May 2024 06:57:34 GMT

I have run a DOE in which one factor is continuous and it has quadratic effect in the model. The range in -0.15 to +0.15, so JMP creates the DOE with -0.15, 0.00 and +0.15.

The problem is that it may be difficult during the experiment to have exactly these number. Therefore, I was wondering whether it is possible to modify in the DOE table after the experiment is performed.

Is this going to affect the results from the model?

I have attached the original DOE table looks like, and the modified DOE table showing how it could look like after the experiment.

Re: Modification of values of factor in DOE

Victor_G — Tue, 14 May 2024 10:46:35 GMT

Hi @bbenny7,

Yes, it is always possible to edit values after having generated the design, the analysis will take ito consideration the edited values :

Little changes in factors' values (like 0,145 instead of 0,15) won't change a lot the performances of the design and results of the analysis, but I'm more concerned about the row 3, where there is a sign difference between the original and modified designs. Is it a typo, an experimental issue, ... ?

The sign change creates a big lack of optimality, significant reduction in chances of detecting the effects in the model (power analysis) and increase the prediction variance over the experimental space and correlations between factors :

If the sign change at row 3 is only a mistake/typo (and is kept negative like in original design), the designs are very similar :

You can compare the designs by using the platform Compare Designs (jmp.com)

Hope this will help you,

Re: Modification of values of factor in DOE

bbenny7 — Tue, 14 May 2024 14:26:51 GMT

Hi @Victor_G , thank you for you explanation.

In Row 3 there was a typo, so I should be fine with adjusting the values.