Variance dispersion graphs for comparing blocked response surface designs

• Autor(a) principal: Trinca, Luzia A. [UNESP]
• Data de Publicação: 1998
• Outros Autores: Gilmour, Steven G.
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1080/00224065.1998.11979868; http://hdl.handle.net/11449/224097
• topic: Blocking; Box-Behnken Designs; Central Composite Designs; D-Optimality; Prediction

In this paper, we extend the use of the variance dispersion graph (VDG) to experiments in which the response surface (RS) design must be blocked. Through several examples we evaluate the prediction performances of RS designs in non-orthogonal block designs compared with the equivalent unblocked designs and orthogonally blocked designs. These examples illustrate that good prediction performance of designs in small blocks can be expected in practice. Most importantly, we show that the allocation of the treatment set to blocks can seriously affect the prediction properties of designs; thus, much care is needed in performing this allocation.