SOCCER VAR: Development of an observational instrument for the analysis of VAR related incidents in soccer

Abstract

Existing literature evidence reveals a significant gap in the use of Video Assistant Referees (VAR) in soccer, particularly regarding the availability of VAR observational methodology resources as a rigorous tool for analysing incidents. The aim of this study was to fill this gap by developing an observational instrument grounded in the laws of the game, the procedures for using VAR, the dynamics of the game, and the contexts that may influence referee decisions. The study underwent the following procedures: a literature review, construction of the instrument, discussions with experts, subsequent adjustments to the instrument, and observer training. To analyse the instrument criteria, an invitation was extended to six referee analysts (53.16 ± 5.19 years old) from the Brazilian Football Confederation (CBF), who are experts in the subject. All met the following criteria: a) be available to analyse the instrument criteria and procedures; b) have at least 15 years of experience in soccer professional refereeing; c) be an analyst, referee, or refereeing instructor; and d) have completed the VAR course. The final VAR observational instrument comprised four criteria and their respective categories: game contexts (n=5), pre-incident (n=7), incident (n=6), and VAR (n=5), totalling 23 sub-criteria with 181 categories. Intra-observer reliability was demonstrated by a Cohen's Kappa value of 0.93, indicating high reliability of the instrument.

The instrument was tested in the analysis of 211 incidents involving VAR in the 2019 Brazilian championship. Of the match situations analysed, 100% involved a review, 146 of which were on-field and 65 factual. Initially enabling studies on the analysis of temporal patterns, the presence of many incomplete T-patterns showed a high variability of occurrences. The following sequential analysis showed statistically significant data calculated from the prospective and retrospective adjusted values (Z>1.96; p<0.05). This culminated in the investigation of polar coordinates, during which several significant associations (Z > 1.96; p < 0.05) involving the four focal behaviours were recorded: penalty, goal, red card, and misidentification. This study contributed to the development of the SOCCER VAR observational tool, which was considered suitable for scanning and analysing contexts and incidents with VAR interaction. This research offers a valuable approach by providing methodological rigour in the observation and analysis of VAR incidents, thereby contributing to filling the gap in the existing literature.