MULTICOLLINEARITY AND LEVERAGE IN MIXTURE EXPERIMENTS

The impact of leverage, influential observations, and multicollinearity in mixture experiments is discussed. By example, it is shown that consideration of these factors is important in the practical selection of a mixture design. Both standard simplex-type mixture designs and computer-generated designs for constrained hyperpolyhedron regions can have very non-uniform distributions of leverage, and consequently, some design points may exert disproportionate influence on the least squares fit. Generally, multicollinearity is not a serious problem in simplex-type designs, provided that when constraints are present the pseudocomponent transformation is used. However, in computer-generated designs, high levels of multicollinearity may be present. How these potential problems can be evaluated and considered in selecting a mixture design is illustrated.