ABSTRACT
In this chapter, the author defines a mixture and provides practical advice on how to put the statistical design of experiments to good use for optimizing reader’s formulation. He provides various excuses to avoid the peculiarities of mixture design and analysis by denying the true nature of the experimental variables. Unfortunately, many formulators, particularly those who are only taught the standard DOE tools for factorials and response surface methods, do not take to mixture design. However, the mixture design produced an unexpected result—consumers liked getting some cherries flipped upside down in any given box—hence the sharp upward slope to component A. After screening previously unknown components, the strategy calls for combining the vital few with the important ones reader set aside earlier into an optimization design that models nonlinear blending effects. Due to the collinearity of components that is inherent in mixture designs power calculations fall short.
