Avoiding rank reversal in analytic hierarchy process decision-support models

Extending the work of a previous paper, which showed that rank reversal in AHP is caused by eigenvector normalization, this paper shows why methods proposed for avoiding rank reversal work, when they are unnecessary, and when they are futile. The paper also discusses the implicit, often unwarranted assumption that the DM’s preference structure can be represented by an additive value function and examines the ramifications when a nonadditive value function is required.