A new method for scoring additive multi‐attribute value models using pairwise rankings of alternatives

Abstract

We present a new method for determining the point values for additive multi‐attribute value models with performance categories. The method, which we refer to as PAPRIKA (P otentially A ll P airwise R an K ings of all possible A lternatives), involves the decision‐maker pairwise ranking potentially all undominated pairs of all possible alternatives represented by the value model. The number of pairs to be explicitly ranked is minimized by the method identifying all pairs implicitly ranked as corollaries of the explicitly ranked pairs. We report on simulations of the method's use and show that if the decision‐maker explicitly ranks pairs defined on just two criteria at‐a‐time, the overall ranking of alternatives produced by the value model is very highly correlated with the true ranking. Therefore, for most practical purposes decision‐makers are unlikely to need to rank pairs defined on more than two criteria, thereby reducing the elicitation burden. We also describe a successful real‐world application involving the scoring of a value model for prioritizing patients for cardiac surgery in New Zealand. We conclude that although the new method entails more judgments than traditional scoring methods, the type of judgment (pairwise rankings of undominated pairs) is arguably simpler and might reasonably be expected to reflect the preferences of decision‐makers more accurately. Copyright © 2009 John Wiley & Sons, Ltd.