Stat > Power and Sample Size > Plackett-Burman Design > Design
Number of main effect
terms omitted from model
Enter a nonnegative integer to specify the number of main effect terms to omit from the model. The number of main effect terms is the same as the number of factors. If the design has 1 corner point more than the number of factors, 1 replicate, and 0 center points, you must omit at least 1 term from the model. If the design has 1 corner point more than the number of factors, 1 replicate, 1 center point, and you include the center point term in the model, you must omit at least 1 term from the model. In other cases, you can omit main effect terms to do calculations for reduced models. For example, if you plan to study 15 factors but expect that only 5 of them will be statistically significant, you omit 10 terms. The calculations are most conservative when you omit the fewest terms possible.
Include term for
center points in model
Usually, you include a term for center points when you have center points in the design. The center points term separates the variation due to curvature from the variation the the model does not explain. This separation produces a more accurate estimate of the variation the model does not explain when the variation that the curvature explains is non-zero. You might exclude the term for center points if you will not use the center points to model variation due to curvature. For example, you replicate the center points to provide an estimate of pure error when you cannot replicate all of the corner points.
