Introduction

The English Premier League introduced the Video Assistant Referee (VAR) at the beginning of the 2019/2020 season after clubs voted unanimously in favour of it in November 2018. Since then, controversy still arises [1–3] but the Premier League defend that the introduction of VAR has facilitated football and that teams are in favour of keeping it [4]. There is also evidence that the number of correct decisions has increased across nations since its introduction [5]. However, VAR has its critics over the negative impact on the matchday fan such as fans “being kept in the dark”, it “ruining the moment” and the number of stoppages it causes to a game [6,7]. While spectators and pundits often voice their views, the effect of the VAR protocol on referee decision making is still uncertain. It is crucial for referees to make accurate and timely decisions [8] under pressure where errors can have significant consequences [9]. Therefore, it is important to understand the external factors that may influence decision making during the current VAR protocol.

The primary roles of the VAR, and Assistant VAR (AVAR) are to support the on-field referee when a potential error has occurred. They are described as video match officials (VMOs) and assist the referee in accordance with the Laws of the Game [10] and the VAR protocol [11] which are both set by the International Football Association Board (IFAB). The VAR can assist the referee using replay footage for a “clear and obvious error” or a “serious missed incident” relating to a goal/no goal, penalty/no penalty, direct red card (not a second caution) or a case of mistaken identity [12]. The AVAR supports the VAR by observing the television footage while the VAR is occupied with a check or a review of another incident as well as other communication responsibilities [12]. The introduction of the VMOs has therefore influenced the role of the referee in that they may be advised to review one of their decisions at the pitch-side monitor by the VAR or AVAR.

Naturally, when it comes to decision making, particularly in sport, the accuracy is of specific interest to many. Spitz et al. [5] are among those that have attempted to investigate the impact that the introduction of VMOs have had on the accuracy of decision making. This group of researchers reviewed 2195 competitive football matches across thirteen countries (Australia, Belgium, China, Czechia, England, France, Germany, Italy, The Netherlands, Poland, Portugal, South Korea, and USA) over the 2016/17 and 2017/18 seasons. It was found that the VAR conducted 9732 checks for potential match-changing incidents (goals, penalty decisions, direct red cards, and mistaken identity) with this leading to 795 on-field reviews at the monitor. Accuracy of correct decisions appeared to increase to 98.3% from the initial 92.1% prior to the VAR introduction. Given football associations and governing bodies are interested in the accuracy of decision making by officials, work like Spitz et al.’s [5] acts as justification for the introduction of VAR. However, as this work reviewed data in matches between 2016 and 2018, the findings could be considered outdated due to the ever-changing and volatile environment of football and the VAR protocol.

The present study was less interested in the accuracy of referee decisions, but instead the psychology of the on-field referee as they review their decision at the pitch-side monitor. While Spitz et al. [5] provide evidence that the VAR process leads to improved decision accuracy, this could still be questioned. Spitz et al. [5] admit themselves that the laws of the game leave room for interpretation at times, and while they took steps to mitigate this by omitting “grey zone” decisions, this may still be problematic. Removing decisions where not all referees agreed on the “correct decision” was a sensible approach, but when this is with 5–10 referees, it is conceivable that their conclusions could be considered contentious. This example may be seen in practice where six officials come to the same conclusion, four on-field officials and the VAR and AVAR, yet the decision sparks controversy among coaches, pundits, and spectators, reiterating how the laws of the game can be interpreted differently. Therefore, while the accuracy of decision making is important, the present study was more concerned with examining the external factors that may influence the decision-making process during the VAR protocol.

Since VAR was implemented in the English Premier League at the beginning of the 2019/2020 season, the Professional Game Match Officials Limited (PGMOL), the body responsible for refereeing games in English professional football, have claimed to be consistently reviewing the use of VAR in the Premier League and often introduce changes at the start of each season. For example, at the beginning of the 2024/25 season, “referee’s call” was introduced [13]. Following this, in-stadium announcements began at the start of the 2025/26 season where referees announce and explain their decisions at the pitch-side monitor over the PA system, in a bid to increase transparency with the crowd in attendance [14]. Also, for the 2025/26 season, the assistant referee will now remain on the touchline in line with the penalty spot when a penalty kick is taken. Prior to this, they would take a position on the goal-line (between one goalpost and the edge of the penalty area) to identify and alert the referee of goalkeeper encroachment. However, they are no longer required to do so, as the VAR is now solely responsible for identifying and alerting the referee if the goalkeeper has encroached during the penalty kick. While this final example is not explicitly relevant to the VAR protocol, it does highlight the growing role and responsibilities of the VAR/AVAR and how this is changing season-to-season making it necessary to assess the differences between decision-making across seasons.

One aspect of the VAR protocol that has remained constant is the location of the pitch-side monitor. As referees must remain visible during the review process to ensure transparency [11], it has meant that the pitch-side monitor is often positioned on one side of the pitch close to the half-way line. In most cases, this is close to the dugout of both the home and away teams and usually positioned in front of home fans, making the current research team curious as to whether home advantage influences the decisions at the pitch-side monitor.

Home advantage is the tendency for sporting teams to win more matches played at their home ground than those played away from home [15]. While this advantage may be due to the familiarity of the pitch, as well as dressing rooms, other facilities, and less travel, it could also be due to the support from a home crowd [16]. In fact, 60% of all competition points, across sports, are scored at home [17], as well as away teams receiving more yellow and red cards than home teams in football [18–21]. When investigating crowd effects, researchers have observed an increase in this apparent home team bias with increasing crowd size [19] and crowd density [21]. These findings are supported by evidence that home bias was reduced during the COVID-19 pandemic where attendance restrictions were enforced [22]. Sors et al. [22] reviewed all first and second division matches in England (Premier League/ Championship), Spain (La Liga/ Segunda División), Italy (Serie A/ Serie B), Germany (Bundesliga/ 2. Bundesliga), and France (Ligue 1/ Ligue 2) in the 2020/21 season, where restrictions remained in place and found that there were fewer home victories and more away victories than the previous five seasons where restrictions were not in place.

Kocsoy [23] also found that referees favour the home teams in terms of additional time awarded. When reviewing 9018 football matches across six European football leagues (England, France, Germany, Italy, Spain, Türkiye) between 2018 and 2023 they observed that 3550 of these matches went into additional time with a one-goal difference. A primary finding in this study was that referees kept the game forty-four seconds shorter when the home team were leading by a single goal compared to when the away team are leading by one goal at the end of the 90^th minute. This figure is elevated to fifty seconds when in the presence of a crowd, and the more home fans in the stadium, the more likely the referee is to shorten the amount of additional time that is played. Thus, it appears that decision making can be influenced by the attendance in the stadium, as well as whether the decision is regarding the home or away team.

The time of the match the decision is taking place should also be considered as there is evidence that, like players, referees may also grow into a game. Mascarenhas et al. [24] assessed five FIFA-qualified referees across seven matches in the New Zealand Football Championship in the 2005/06 season. When analysing 144 incidents, it was found that referees made accurate decisions 64% of the time. However, this accuracy was lower in the first fifteen minutes of each half (51%) compared to all other parts of the match (70%). Although this is a relatively small sample in just one league, there is evidence that referees frequently make errors with correct decision rates ranging from 50% to 93.1% [25] of which Mascarenhas et al.’s [24] findings fall between. These findings, therefore, suggest that the time of the match can influence decision making and therefore could do so during the VAR protocol.

Decision making in football has historically and predominantly been the responsibility of the referee, however, now with the introduction of the VAR protocol, the decision making of the VAR, is now under scrutiny too. Recent research has found that age, experience, and status asymmetries between officials influence decision making [26]. When assessing VAR interventions in the Eredivisie, the Dutch top football league, between 2019 and 2021, Garretsen et al. [26] discovered that younger, less experienced, and lower-ranked VARs recommended more interventions and are more likely to have their recommendations dismissed by the on-field referee. These findings suggest that there are cognitive and social factors that contribute to decision making under uncertain, high-pressure situations and therefore, the current study was interested in examining the likelihood of VAR intervention across seasons, as well as whether attendance in the stadium played a role, given these factors are yet to be explored.

The current study, therefore, aimed to make three contributions to the literature. Firstly, to provide prevalence rates of maintain and overturn decisions at the pitch-side monitor between 2021–2025 in the English Premier League. Secondly, to explore the likelihood of referees maintaining their decision at the pitch-side monitor based on external factors such as the season in which the decision took place, whether the decision was regarding the home or away team, the scoreline at the time of the decision, the time in which the decision took place, and the attendance in the stadium. Finally, the current study also tested whether the season in which the decision took place, and stadium attendance influenced the likelihood of VAR intervention.

Method

Assumptions

As our outcome variables were categorical with two levels (overturn/maintain, intervened/did not intervene), we conducted binary logistic regressions. This analysis type requires little to no multicollinearity between independent variables, with Table 1 depicting this is the case for our data.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Pearson correlations between independent variables.

https://doi.org/10.1371/journal.pone.0345704.t001

Results

Binary logistic regression

The initial Binary Logistic Regression was performed to assess the impact of a set of predictor variables on the odds that the referee would overturn their original decision. The first model contained five independent variables (season, home advantage, scoreline, time of decision, and crowd attendance). The full model containing all predictors was not statistically significant χ² (8, N = 245) = 9.946, p = .269, indicating that the model was unable to distinguish between the referee’s decision to overturn or maintain their original decision. The model as a whole correctly classified 95.1% of cases. As shown in Table 2, none of the independent variables made a unique statistically significant contribution to the model.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Logistic Regression predicting the likelihood of the referee maintaining their original decision at the pitch-side monitor.

https://doi.org/10.1371/journal.pone.0345704.t002

A second Binary Logistic Regression was performed to assess the impact of a set of predictor variables on the odds that the VAR would intervene. The model contained two independent variables (season and crowd attendance). The full model containing both predictors, again, was not statistically significant χ² (4, N = 1535) = 2.424, p =.658, indicating that this model was also unable to distinguish between whether the VAR intervened or not. The model as a whole correctly classified 84% of cases. As shown in Table 3, neither of the independent variables made a unique statistically significant contribution to the model.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Logistic Regression predicting the likelihood of the VAR intervening.

https://doi.org/10.1371/journal.pone.0345704.t003

Discussion

The aims of this study were i) to calculate the prevalence of overturn and maintain decisions at the pitch-side monitor, ii) to calculate the odds that the referee would overturn their decision based on external factors, iii) to calculate the odds that the VAR would intervene based on external factors. This study has evidenced that referees in the English Premier League overturned their original decision over 95% of the time across the four-season period and that referees are consistent in these decisions across seasons. It was also found that home advantage, scoreline, time of decision, and crowd attendance do not sway this decision, and that season and attendance did not influence whether the VAR would intervene.

Referees overturning their original decision in 95% of cases is high but may be expected, as the reason they are being advised to consult the pitch-side monitor is due to a potential error. However, prior to this study, the prevalence of overturn decisions was unknown, and therefore this gap is filled. Future researchers are better informed when developing their hypotheses regarding this area with the knowledge that referees change their original decision nineteen in twenty times.

Without this knowledge, the current research team utilised an exploratory approach to investigate the prevalence of overturn and maintain decisions at the pitch-side monitor. The research team decided to explore the impact of potential variables that have been found to influence referee decisions in the past such as home advantage [15,23], scoreline, the time of the match [24], and match attendance [19,21], and due to the ever-changing nature of the VAR protocol, we assessed whether season influenced the referee’s decision. Season and match attendance were also investigated to determine whether they influenced if the VAR would intervene or not. As there was such a discrepancy in classification of our dependent variables i) 5% maintain decisions v. 95% overturn decisions, ii) 15% VAR intervention v. 85% no VAR intervention, the binary logistic regressions were rendered less useful in determining the likelihood of the referee maintaining their decision or the VAR intervening based on the external factors we were interested in. This finding provides support for PGMOL by suggesting that their officials, both on-field and VMOs are consistent in their decisions and are not influenced by external factors that have in the past been reported to affect decision making.

From this, the low prevalence of maintain decisions may be the more interesting aspect of these findings. For example, three quarters of maintain decisions were regarding the home team, and while this is a relatively small sample (9 of 12), this finding could be explained by referees implicitly favouring the home team as has been reported in the past [15,23]. As the sample of this occurrence is very small, it could simply be explained by coincidence. However, the location of the pitch-side monitor must also be considered. As referees must remain visible during the review process [11], it leaves few options for the location of the pitch-side monitor. Four enclosed stands comprise all 25 stadia included in the current study, which means the pitch-side monitor will always be set up near a set of supporters. Invariably, the position is close to the dugout of both the home and away teams which is often just yards away from home supporters, which could invoke some proximity bias in the on-field referee. Proximity bias is the tendency to favour those that are closer to us, with a popular example being office managers favouring those working in the office over those working remotely [27]. In the current context, it is possible that the home supporters and dugout have an influence on the rate of maintain decisions regarding the home team simply because they are closer to the referee. Likewise, decisions that regard the away team may be easier to overturn when the supporters are not close to the referee as they make their decision at the pitch-side monitor. This proximity bias may also explain why home bias reduced during COVID-19 restrictions where there was no crowd to influence the referee [22], whether implicitly or explicitly.

Conclusion and future directions

To conclude, the present study aimed to i) calculate the prevalence of overturn and maintain decisions at the pitch-side monitor, ii) calculate the odds that the referee would overturn their decision based on external factors, iii) calculate the odds that the VAR would intervene based on external factors. The current study was successful in achieving these aims as data analysis revealed a prevalence of 95% overturned decisions in the English Premier League between 2021/22 and 2024/25. It was also found season, home advantage, scoreline, time of decision, and crowd attendance, do not influence this decision. VAR/AVAR intervention was also not influenced by season or crowd attendance. Due to the large imbalance between overturn and maintain decision, the research team have discussed some of the potential reasons for rarer occurrence of referees maintaining their decision at the pitch-side monitor, such as proximity bias. Practical implications based on these findings could include referees reviewing their decision at a monitor away from the pitch, but with this still recorded for transparency purposes. Following on from this work, it would also be useful to conduct qualitative interviews with English Premier League officials to investigate their thoughts and feelings towards external factors that may or may not influence decision making associated with the current VAR protocol.