Designs specifically tailored to quantitative factors


This GUI covers classical response surface designs (central composite and Box-Behnken) as well as so-called latin-hypercube samples for factors that can and should be set to many different levels. This help file is about when to apply which of these.

Central composite designs

Central composite designs can be generated from scratch or as an extension to 2-level full or fractional factorials generated with the 2-level dialogue or with function FrF2. Their rationale is explained here. They usually have five different levels per experimental factor.

Box-Behnken designs have only three levels for each factor. They are explained here. They can not be generated by augmenting an existing design.

Latin hypercube designs - if used with optimization which is strongly recommended - try to fill the experimental space with points in an efficient way. Note that they are NOT trying to optimize the design w.r.t. a specific statistical model. Such optimal plans are available outside the R Commander with package AlgDesign. Simple versions of these have already been implemented into the R Commander.


Ulrike Groemping


Box G. E. P, Hunter, W. C. and Hunter, J. S. (2005) Statistics for Experimenters, 2nd edition. New York: Wiley.

See Also

See Also and ccd.augment for the functions behind the central composite designs, for the function behind the Box-Behnken designs, and for the function behind the latin hypercube samples.

Want to suggest features or report bugs for Use the GitHub issue tracker.

comments powered by Disqus