DOE (design of experiments) in Minitab 16

[Nermal]

Hi,

Im planning to carry out a screening DOE, with 2 factors and 7 levels. with this part im fine.

Where i get a little confused is when there is more than 1 response. In my case the will be 4. different dimensions on the product. Can a DOE have more than response?

any help would be great

 

[Miner]

Here's your avatar...

Yes, a DOE can have more than one response. You simply analyze each response separately.

Why are you running a screening modeling design with 7 levels? Are these categorical variables or continuous? If continuous, I would recommend a response surface design instead.

 

[Nermal]

Thank you

sorry that had a slight typing error, 7 factors and 2 levels.

hoping the screening DOE can rule out thr non factors.

The data is all continuous.

hopefully it makes more sense now

 

[Statistical Steven]

The problem is not analyzing the data, it is interpreting the analysis. As Miner commented, you will have 4 separate analysis, but I will go out on a limb and say that different factors will be significant for different responses. This is the hard part. What do you do when Factor A is significant for Response 1 and Factor B is significant for Response 2 but not Response 1? My advice, before you analyze the data, rank your responses so you can decide which response has more weight when you need to decide what is the best approach for your process.

 

[Nermal]

Thank you Steven.
what you say is what I though different responses will possibly have different significant factors.
So your solution is great. Give each response a weight before the doe is carried out.

I was thinking along the lines of giving the factors weights based on the Pareto for each response. But that could mean a response which isn't as important as another clouding the results.

Thanks for your responses

 

[Miner]

What software will you use to analyze the results? I work with Minitab 17, which has a response optimizer. This allows you to maximize a desirability index by both weighting the individual responses as well as deviation from each response to a target value. I am not sure what other software has available.

See What is response optimization?

 

[Nermal]

We will be using Minitab 16, which has a response optimizer too. not sure how different it is in Minitab 17.

Thanks for the link will have a little study today. We have used them in the past, with out understanding it fully. Plus this DOE is a little more complex than our usual 2 level 2 factor. After the screening DOE I will select the 3 most significant factors and carry out a full 2 level DOE on them.