JudgED: Comparison between Kickboxing Referee Performance at a Novel Serious Game for Judging Improvement and at World Championships

Abstract

The particular responsibility of referees in combat sports lies in their decision-making to enforce the rules of the sport, which requires considerable experience and a multitude of skills, including perception, categorization, memory processing, and information integration. As a cost-effective alternative to in-tournament training, this research aims to evaluate the novel video-based serious game called “JudgED” to train martial arts referees’ decision-making processes through immediate feedback. The effectiveness of the JudgED game was assessed by (a) measuring decision accuracy and specific reaction time, (b) calculating a theoretical probability of correct scoring, and (c) comparing these results with real competition judging agreement data. A field study was conducted to analyze the performance of 16 kickboxing referees. The study involved two video-based tests in the serious game. The performance data for JudgED were obtained via a procedure that compares the players’ inputs in the serious game with expert-defined decisions. The results were compared to real-competition data gathered through qualitative analysis of kickboxing fights (n = 400 fights/1200 bouts) at the WAKO World Championships 2021. The findings showed an average decision accuracy of 43.011% and an average reaction time of 1.022 s. For further comparison, binominal distribution for the probability of correct final decisions (between 15.3% and 67.2%) in JudgED and Fleiss’ Kappa interrater reliability for JudgED (Ring: κ = 0.371; Tatami: κ = 0.398; p < 0.001) and tournament decisions (by bout: κ = 0.114; by fight κ = 0.063; by outcome κ = 0.166; p < 0.001) were calculated. The results suggest that more training is required to improve referee decision accuracy, and JudgED bears the potential to work as a suitable supporting system.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 13 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.