A decision aid for finding performance groups

Abstract

PurposeFinding groups of similarly performing organizations for strategic or tactical performance comparison has typically been done by either cluster or frontier analysis. Both present problems of definition and interpretation. Given the inevitable abstraction of any model it is desirable that as much as possible of the model specification be in easily understood language and that, where appropriate, the ambiguity of the model is recognised by giving a number of equally good solutions. The purpose of this paper is to describe a model which balances what is judged and what is calculated in a way different to that usual in performance analysis.Design/methodology/approachTwo linked optimisations find the minimum number of groups and the distribution of group sizes which identifies large groups, if they exist. A number of alternative optimal solutions are generated and must be considered in the final synthesis. The method is illustrated by finding performance groups among 35 international airports.FindingsIn total, 16 alternative optimal solutions, each of 12 groups, are found. Two thirds of the airports were found in the same groups in all solutions, forming the robust core. Of the remainder, a solution was chosen and justified on the grounds of similarity of country and jurisdiction.Practical implicationsThe model changes the focus of analysis from one of reacting to apparent patterns in the result of an analysis to one in which careful thought is given about the definitions of group which reflect the point of view of the user.Originality/valueThe paper illustrates how a novel methodology forms a decision aid for finding performance groups in which due scope is given for the interpretation of equally good solutions.