Guidelines for fitting custom screening DoE

[#1]

Hello JMP User Community,

I am seeking for some suggestions/guidelines to fit a Custom Screening DoE for main effects. Below is the screen shot of the model fitting. My thought process is to remove those factors with t ratio less than 2 one by one until all factors left with t ratio above 2 and then use those factors to conduct a Response Surface Design for more detailed investigation. I will also use my practical knowledge when eliminating factors. Here are some questions that I have:

1. is it safe to eliminate the factors with their t ratios less than 2? According to JMP, it is basically equivalent to p value of 0.05 although there is no p value available for this design. Assuming no practical knowledge about the impact of the factors on the response.

2. Any other methods should be used? 

Thank you!

[#2]

Hi,

Can you say something about the design? It seems like you have 10 factors. So 11 parameters (including the intercept) to estimate. And the Fit Model report says 20 observations (so presumably 20 runs). So I am struggling to understand why you are not getting p-values. Do you have random effects in this model as well?

More information would help the JMP Community to help you. Maybe you can attach an anonymised version of the data.

Regards,

Phil

[#3]

@statman is correct that I had another post for the same question with more details "Custom Screening DoE". Statman gave me a lot guidance on that topic. I also reviewed JMP Help for some examples myself and think I got a good idea about what I should do. Thank you for getting back to me and being willing to help!

[#4]

No need to add multiple topics.  You already have a post containing exactly the same information with the same questions.

[#5]

You are correct and thank you very much again for all the suggestions and guidance for my model. Really appreciated your time and patience!