In defense of group fuzzy AHP: A comparison of group fuzzy AHP and group AHP with confidence intervals

Abstract

Group aggregation of the Analytic Hierarchy Process (AHP), both with crisp and fuzzy numbers, is a widely researched field in group multiple criteria decision-making. Typically, if uncertainty needs to be reflected in the outcome of an AHP, triangular fuzzy numbers are used, although this is a controversial method, as many argue that the traditional 1–9 scale used in crisp AHP introduces uncertainty to the decision-making process, leaving no need for fuzzification. This research introduces the use of confidence intervals around the aggregated group mean to estimate the score ranges generated through group fuzzy AHP (GFAHP), as confidence intervals are simpler to calculate and more familiar to many. A comparison of the score ranges resulting from GFAHP and group aggregation using crisp numbers with confidence intervals around the group mean score, aggregating on individual priorities using both the arithmetic mean and geometric mean, was completed. The concept of using confidence intervals on crisp scores to generate score ranges was implemented through a case study to select a new all-wheel drive crossover vehicle. This study shows that confidence intervals introduce excessive uncertainty to the decision-making problem.