Abstract

[This abstract is based on the authors' abstract.] A mixture experiment entails the mixing of different proportions of two or more components to make different compositions of an end product. Complete Scheffe quadratic (CSQ) models are frequently used to model responses as functions of the component proportions. At times, models containing fewer quadratic terms the CSQ are called for, and reduced Scheffe quadratic models consisting of linear terms augmented by selected quadratic crossproduct terms are used. A partial quadratic mixture model is proposed that contains linear terms augmented by selected squared terms with selected quadratic crossproduct terms and/or squared terms. The model is compared to restricted and reduced Scheffe models and demonstrated using data from two constrained mixture experiments with simulated waste glass.