The logarithmic least squares priorities and ordinal violations in the best-worst method

Abstract

The best–worst method is an increasingly popular approach to solving multi-criteria decision-making problems. However, the usual prioritisation techniques may result in an ordinal violation if the best (worst) alternative identified in the first step does not receive the highest (lowest) weight. The current paper gives two sufficient conditions for the logarithmic least squares method, applied to an incomplete best–worst method matrix, to guarantee the lack of ordinal violations. Our results provide another powerful argument for using the logarithmic least squares priorities in the best–worst method.