Decision Quality Using Ranked Attribute Weights

Abstract

Three published approximation formulae for selecting the best multiattribute alternative based on rank-ordered weights are evaluated. All formulae are surprisingly efficacious in determining the best multiattribute alternative. Rank order centroid (ROC) weights are more accurate than the other rank-based formulae; furthermore, the ROC formula generalizes to incorporate both other forms of partial information about attribute weights and partial rank order information as well. Because a ROC-based analysis is so straightforward and efficacious, it provides an appropriate implementation tool.