The fish patty experiment: a strip-plot look

Abstract

In this article, I provide a detailed discussion of the well-known fish patty experiment introduced in the literature by the late John A. Cornell in the first edition of his famous textbook on the design and analysis of mixture experiments. Cornell used the fish patty experiment as the motivating example for an article discussing that, for logistical reasons, many mixture-process variable experiments are run using a split-plot experimental design. More specifically, he described two possible ways in which the fish patty experiment might have been performed, both of which require a split-plot analysis of the data. These descriptions were not followed by the corresponding analyses of the fish patty data. Moreover, Cornell did not discuss the most convenient way in which the fish patty experiment could have been run, namely using a strip-plot design. In this article, I discuss the logistics leading to a strip-plot design, conduct the corresponding strip-plot analysis and contrast it with the two split-plot analyses.