Minitab DOE - 4-factor, 2-level, 8 trial, level IV fractional DOE help needed

[Dennis B]

Hi,

I am on the last day of my Minitab free trial and am struggling to peform a 4-factor, 2-level, 8 trial, level IV fractional DOE.

I have created my DOE design and set Factor high and low levels and stored the design. What is the next step please???

Dennis B

[Miner]

Not exactly sure what you are looking for. If you are satisfied with the design and have already determined how many repeats and replicates you will run and have accounted for this in the design, the next step is to run the experiment, and enter the results back into Minitab. Next, run the analysis (Stat>DOE>Factorial>Analyze Factorial Design) analyze the effect plots and residual plots, then run the graphs (Stat>DOE>Factorial>Factorial Plots) for main effects and interactions. If you have run repeats, you can also analyze the variability (Stat>DOE>Factorial>Analyze Varaiability).

[Dennis B]

Ooops, yes it does help if I enter some responses i am having a very foggy day. many thnaks for the quick reply

[Dennis B]

Hi Miner,

Many thanks for previous advice, but after reviewing the experiment result (after entering the response column...) the analysis output looks strange:

F and P are displayed as * (as are residual error, SE fit and St Resid.). There is also final message "Could not graph the specified residual type because MSE = 0; or the degrees of freedom for error = 0.

For the exercise I am trying to run (for a distance MBB course) I have an instruction to "Let MSE = 0.05". I cannot find any setting in Minitab to specify MSE setting - is this my problem; or am I missing something else??

many thanks in anticipation

Dennis B

[Miner]

With zero repeats and all terms (factors plus all interactions) included in the model, all degrees of freedom are used and MSE cannot be calculated. You must pool non-significant factors to create this.

In your case, I recommend taking the higher order (4-way and 3-way) interactions (AxBxCxD, AxBxC, AxBxD, etc,) out of the model. This will create an MSE.

You could remove the four-way interaction first, and run the analysis, then start removing the 3-way interactions with the largest p-value until the MSE is approximately 0.05.

[Dennis B]

Hey Miner,

Thank you very much for your help. Unfortunately I was unable to try your idea as my Minitab Trial has just expired (before my eyes )

I will have to try to find some other trial software solution to help with the last excerise on my MMB course. Any suggestions would be greatfully received.

cheers

Dennis B