Central Composite Design for the Experiments with Replicate Runs at Factorial and Axial Points

摘要

In this paper, we have considered for study the central composite design (CCD) with a view to extending the related second-order response surface models. In the CCD, there are three types of design points, such as the factorial, axial and center points. We devised a new type of CCD for the experiments with replicate runs at such design points, not just at the design center. This type of experimental design, named as CCD-R, proves to be flexible in the sense that there exist a myriad of exactly orthogonal and nearly rotatable designs. It is also demonstrated that the proposed CCD-R is useful, especially when a number of replicate runs are desired in the experiment for one reason or another. Since the second-order generalized regression models provide a good framework of the method, an attempt is given to the problem of logistic regression analysis for categorical data based on the CCD-R.