Multifactor and multidimensional data quality analysis of judge scoring in diving competition.

Abstract

In sports competitions, judge scoring data serve as an objective measure of an athlete's performance level. However, research has indicated the unreliability of objective measurements. Controversy often arises regarding the quality of judge scoring data, undermining fairness and justice in sports competitions. This paper proposes a method utilizing the Kendall covariance coefficient and the Kendall correlation coefficient for the thorough evaluation of judging data quality in diving events. The analysis is structured around four key elements: overall competition, individual divers, specific rounds, and distinct diving techniques. Each element is analyzed across three dimensions: the collective data quality from the judging panel, interjudge data quality comparisons, and the alignment of individual judges' scores with the final tallied scores. Two case studies serve to illustrate the application of this method. The Kendall covariance coefficient is employed to assess the data quality from the judges as a unified entity, whereas the Kendall correlation coefficient is utilized to evaluate the data quality from individual judges. Results show that the data quality of the judge group's scoring is high, while the data quality of the judge group's scoring for the 6th diver, the 5th round, Dive No. 5152B, Judge 5 and 6 in the Competition 1, and the 1st diver, the 3rd round, Dive No. 6245D, Judge 4 in the Competition 2 is inconsistent with the others. This approach uncovers disparities in data quality attributed to the judges' panel across each diver, each round, and the various diving maneuvers. However, the Kendall correlation coefficient may not be suitable for evaluating data quality when both the data differences and the sample size are small.