Center Point
Extra runs at the middle of the design region — for curvature test and pure-error estimation of the standard deviation.
Also known as: centre point, center run
Center points are runs with every factor at its middle level (coded 0). They are added to the corner points of a 2ᵏ or fractional design — usually 3–5 replicates.
- Curvature test — is the response between −1 and +1 linear or curved? A significant gap between the centre-point mean and the corner-point mean indicates curvature — then upgrade to CCD or Box-Behnken
- Pure error — replicated centre points provide a model-independent estimate of process variation
- Lack-of-fit test — without this variance estimate the classical lack-of-fit test is unavailable
See also
Used in
In the Algorithm Lab
Sources
- Montgomery, D. C. — Design and Analysis of Experiments, 10th Edition, Chapter 6
- Myers, Montgomery & Anderson-Cook — Response Surface Methodology, 4th Edition