Minitab factorial design regression equation center point term

[bmnichowicz]

I have analyzed a factorial design which includes a center point. I want to copy the regression equation to use in Microsoft Excel. The regression equation ends with

+ 0.002380 Ct Pt

My question is: what is the value I should use for the Ct Pt variable? Is it zero? One? Is it the mean response value at the center point?

Thank you.

 

[Miner]

I have analyzed a factorial design which includes a center point. I want to copy the regression equation to use in Microsoft Excel. The regression equation ends with

+ 0.002380 Ct Pt

My question is: what is the value I should use for the Ct Pt variable? Is it zero? One? Is it the mean response value at the center point?

Thank you.

If the center point was not statistically significant, remove it from the regression model. If it was significant, augment your design with axial points and analyze it as a response surface design.

NEVER use a regression model with a center point term for predictions. Center points are used to detect lack of fit, curvature or block effects.

 

[JoeHaef]

The Center Point is a factor and possible predictor. If it is significant then include it in your model.

 

[Miner]

The Center Point is a factor and possible predictor. If it is significant then include it in your model.

Absolutely not. It cannot be used as a predictor. If it is significant, it only signals that there is curvature in one or more of the factors. You do not know which factor, so you must augment the design with axial points and analyze it as a response surface design to get a predictive model.

 

[JoeHaef]

What will reflect curvature in the model if you remove the centerpoints?

 

[Miner]

What will reflect curvature in the model if you remove the centerpoints?

A significant center point tells you that this model is inadequate and that you need a better model.