Identifying outlier scores and outlier jurors to reduce manipulation in classical music competitions

Abstract

Manipulations in classical music competitions are a proven problem that has as yet not been eradicated from the competition practice. Several examples are provided. We aim to compare and analyze different methods of reducing manipulation in classical music competitions, focusing on outlier scores and on outlier jurors. First, we investigate the typical approach of correcting or discarding individual jurors' scores that significantly deviate from the mean or median of scores received by a given candidate. We then introduce a new method that involves the exclusion of outlier jurors (EOJ). This approach implies that all scores of jurors with ratings that substantially differ from those of other jurors are removed and not taken into account when determining the ranking of candidates. The properties of both approaches are discussed in hypothetical voting scenarios, where one or more jurors assign scores that deviate markedly from those awarded by other jurors. Finally, we present examples of applying various methods to real-world data from classical music competitions, demonstrating the potential effectiveness and implications of each approach in reducing manipulation within these events. Two examples are taken from the International Karol Szymanowski Music Competition, which took place in September 2023 in Katowice, Poland, where EOJ was adopted for the first time as the official scoring system in four competition categories: piano, violin, vocal, and string quartet.