How to analyze whether my DOE model is significant or not?

rashi1

I ran CCD with 2 center points. I have only two factors and one response. The summary results and analysis of variance show a very low r-square value, and the p-value is also higher than 0.05. Is my model correct or incorrect? If incorrect, then what can I do to make it work?

image 1.png

image 2.png

statman

There is not enough information to provide specific advice, but I have these thoughts:

1. Always assess practical significance before statistical.  Did the response variable change enough to be meaningful?

2. I notice a potential outlier in your residuals plot (-30, 60). 

3. None of your model terms are significant.  

4. Your RMSE is huge compared to the mean of the data.

5. You removed terms from the model that may be significant (Lack of fit table).  Your only estimate of pure error was the replicated center point. 

 

If you are running central composite designs, these are usually considered optimization designs.  This means you should already know significant factor effects and you are trying to develop a more specific, detailed, or complex model.  Often a surface profiler is the best way to look at this type of experiment, but I suspect there is more work to be done to get the important factors in your process before optimizing.