Skip to Content
MDPIMDPI

Advanced

Applied SciencesApplied Sciences
PDF
  • Article
  • Open Access

13 October 2026

12 Pages

Associations Between GPS-Derived External Match Load and Refereeing Performance in Elite Football Referees

,
,
and
1
Faculty of Kinesiology, University of Zagreb, Horvacanski zavoj 15, 10000 Zagreb, Croatia
2
Croatian Football Federation, Ulica grada Vukovara 269 A, 10000 Zagreb, Croatia
3
Faculty of Civil Engineering, University of Zagreb, Fra Andrije Kacica-Miosica 26, 10000 Zagreb, Croatia
*
Author to whom correspondence should be addressed.
Appl. Sci.2026, 16(20), 10133;https://doi.org/10.3390/app162010133
This article belongs to the Special Issue Biomechanics and Technology in Sports
Version Notes
Order Reprints

Abstract

This study examined associations between GPS-derived external match-load variables and official refereeing performance in elite football referees. The cohort comprised all 17 main referees eligible to officiate in the highest Croatian national competition during the 2022/2023 season, including 10 UEFA/FIFA-listed referees. One referee was excluded from the field analysis because of injury, leaving 84 official-match observations from 16 referees. Refereeing performance was based on confidential official match grades additionally reviewed by seven experts appointed by the Croatian Football Federation Referees Committee. Because repeated matches were nested within referees, the primary analysis used referee fixed-effects models with referee-clustered standard errors and exact wild cluster bootstrap inference. Holm correction was applied across six GPS variables. The match-level grade was 8.27 ± 0.25 (range 7.30–8.50). Deceleration count showed a positive within-referee association with grade (B = 0.0068, 95% exact wild-cluster-bootstrap CI = 0.0036 to 0.0093, exact bootstrap p = 0.006, Holm-adjusted p = 0.038). Total distance, high-intensity running distance, sprint count, acceleration count, and maximum speed were not significant after correction. Sensitivity analyses showed that some estimates depended on model specification. GPS variables should therefore be interpreted as measures of match demands, not as direct measures of refereeing quality.

1. Introduction

Football referees must be physically prepared to follow the intensity and rhythm of the game while making decisions throughout the match. The main referee moves between lower- and higher-intensity activities, changes direction, accelerates, decelerates, and has to reach positions from which play can be observed clearly. Endurance, speed, agility, and repeated high-intensity running are part of referee preparation because the referee has to follow players without interfering with play and still be close enough to evaluate important incidents [1,2,3,4,5].
GPS technology makes it possible to record these movement demands during official matches. The system provides information on total distance, high-intensity running, sprint actions, accelerations, decelerations, and maximum speed [6,7]. Studies of professional and elite referees show that main referees usually cover about 9–12 km per match and perform repeated high-intensity actions [8,9]. GPS monitoring was used because it provides data on the physical demands of referees during official matches. Laboratory and field tests provide information on physical capacity, while GPS records the physical demands during competition. Short periods of very high running demand can be hidden when only whole-match averages are examined [9].
Referee movement during a match depends on events on the field. A physically prepared and experienced referee can anticipate the development of play and reach an appropriate position without unnecessary high-speed running. A referee who reacts later or starts from a less favourable position may need more sprinting to follow the same action. The same applies to acceleration and deceleration counts. A greater number of high-intensity actions does not necessarily represent better movement or better refereeing. Positioning is important because distance from the incident, viewing angle, and line of sight affect the information available before a decision is made [10,11,12,13].
Studies connecting physical performance with refereeing decisions have used different criteria and different settings. Mascarenhas et al. analysed movement and physiological variables together with expert-coded incidents during competitive matches [14]. McEwan et al. examined the internal and external load immediately before decisions [15], while Castillo-Rodriguez et al. related physical fitness and match running to the success of selected decisions in national-level referees [16]. Acute vigorous exercise can change response-suppression measures without producing the same change in football-specific decision performance [17], and response-suppression ability has recently been associated with some aspects of the accuracy of disciplinary decisions [18]. These studies show that physical fitness, total match load, and the physical load immediately before a decision should be considered separately.
The physical demands placed on a referee are not the same in every match. Elite referees adapt their running performance to the technical, tactical, and physical characteristics of the game, with experience contributing to this adaptation [19]. Match pace, rapid transitions, team style, scoreline, and the location of decisive incidents can change the distance and intensity required from the referee. The same referee can record different GPS values in two matches without any change in underlying fitness. Match context is necessary when external load is related to refereeing performance.
Official referee grades provide a different criterion from individual decision tasks. These grades are rarely available to researchers because they are confidential. Carvalho et al. examined the variables that contribute to formal referee match-performance assessments [20] but did not combine a complete GPS match-load profile with the grading procedure used in the present study. Each official grade in this project was first assigned through the standard referee-observer process. Seven experts appointed by the Croatian Football Federation Referees Committee then independently reviewed the referee’s performance, after which the grade could remain unchanged or be corrected. To the best of our knowledge, a complete elite national referee cohort combining official-match GPS data with confidential grades subjected to an additional seven-expert review and correction process has not previously been reported in football-referee research. GPS variables in this study were related to an official criterion of refereeing quality obtained during real competitive matches.
The aim of this study was to determine the association between GPS-derived external match-load variables and official refereeing performance in elite football referees. Total distance, high-intensity running distance, sprint count, acceleration count, deceleration count, and maximum speed were analysed. The analysis was primarily exploratory because six GPS variables were examined and no single variable was defined as a primary predictor. We expected that the GPS variables would not show the same relationship with official performance grades because movement during a match is influenced by positioning, anticipation, experience, and the demands created by the game.

2. Materials and Methods

2.1. Participants

The target population consisted of 17 main football referees eligible to officiate in the highest Croatian national competition during the 2022/2023 season. Their mean age was 34.53 ± 5.69 years, mean body height was 186.61 ± 5.09 cm, and mean body mass was 84.81 ± 5.72 kg. Ten referees were listed by UEFA and/or FIFA, while seven officiated only under the Croatian Football Federation. The study included 100% of the main referees eligible for the highest national competition in that season.
One referee sustained an injury and did not provide a complete field-performance dataset. The field analysis included 84 official matches from 16 referees. These 84 observations represented all matches monitored with GPS during the selected half-season of the 2022/2023 competition for which complete GPS data and a final post-review match grade were available; no complete monitored observation was excluded from the analysis. The number of matches contributed by individual referees ranged from 1 to 11. The absolute number of referees is small, but the sample could not be increased within the predefined target population because all eligible main referees had already been included.

2.2. Study Design and Ethical Considerations

This study is part of a broader project in which anthropometric characteristics, motor and functional abilities, match load, and official refereeing performance were measured in the same elite referee population. The present paper analyses only the GPS-derived match-load variables and their relationship with official refereeing performance. Laboratory measures from the broader project were not included in the present analysis.
Before data collection, responsibilities of the research team and the Croatian Football Federation Referees Committee were defined, including procedures for handling confidential official grades. The study was conducted in accordance with the Declaration of Helsinki and approved by the Committee for Scientific Research and Ethics of the Faculty of Kinesiology, University of Zagreb (approval number 13/2023; approved on 22 February 2023). All GPS data analysed in the present paper were collected after this approval. All referees provided written informed consent before participation.

2.3. GPS Monitoring and External-Load Variables

External match load was monitored during official SuperSport Croatian Football League matches using STATSports APEX Pro GPS units (STATSports Group Limited, Newry, Northern Ireland, UK) operating at a 10 Hz sampling frequency. An individual profile was created for each referee in Sonra 4.0 software. The GPS unit was placed in a vest on the upper back between the scapulae before the match. It was switched on immediately before the referee left the tunnel for the pitch, switched off after the match, and returned to the measurement coordinator. The device was then connected to the STATSports system, and the recorded data were downloaded and stored separately for each referee. The firmware version used during data collection was not available in the archived study records.
The GPS variables were total distance (m), high-intensity running distance (m), sprint count, acceleration count, deceleration count, and maximum speed (km/h). High-intensity running was defined as running above 19.8 km/h. A sprint was recorded when speed exceeded 25 km/h for more than one second. These speed thresholds were defined in the original monitoring protocol before the present analysis, were applied unchanged to all matches, and were not selected on the basis of the study results. These cut-offs are consistent with velocity bands used in elite football-referee research, where high-speed running has been defined above 19.8 km/h and sprinting above 25.2 km/h [5,21]. Acceleration and deceleration counts were obtained using the default absolute thresholds in Sonra 4.0, with acceleration events defined as >3 m·s−2 and deceleration events as <−3 m·s−2. These default thresholds were not modified during the study. Maximum-speed values were checked for clearly implausible values before the revised analysis. All recorded values were between 23.69 and 31.40 km/h, and no match was excluded on this basis.

2.4. Assessment of Refereeing Performance

Refereeing performance was obtained from the official referee-observer report used in the highest Croatian competition. The starting grade for a match is 8.4. Correct and incorrect decisions can increase or decrease the grade according to the importance of the decision. Evaluated incidents include disciplinary sanctions, penalty decisions, goals, offside-related situations, and other events that require referee intervention.
After every competition round, seven experts appointed by the Croatian Football Federation Referees Committee independently reviewed the complete match recording and relevant VAR footage for every delegated main referee, and each expert assigned a numerical grade. When individual numerical grades differed, the experts discussed their evaluations until agreement was reached on a single final numerical grade. The experts did not have access to the original referee-observer report or to the GPS data. After all seven independent evaluations were completed, the final official grade was determined by consensus of the seven experts and could remain unchanged or be corrected upward or downward relative to the observer grade. The corrected grade was subsequently used in the official referee ranking lists. The research database contained only the final committee-approved grade and not the seven individual expert ratings; therefore, inter-rater reliability could not be calculated for the present study. The final post-review grade was used as the criterion variable for each match. A season-long grade was calculated for descriptive purposes as the mean of the final grades from all matches officiated by each referee.

2.5. Statistical Analysis

Data management and descriptive statistics were performed using STATISTICA 14.3. Descriptive statistics included the arithmetic mean, standard deviation, minimum and maximum values, and range of results. The revised inferential analyses were performed in Python 3.13.5 using statsmodels 0.14.6 and custom code for exact wild cluster bootstrap inference.
The field database contained 84 official-match observations from 16 referees. Because several matches were officiated by the same referee, the primary analysis used referee fixed-effects linear regression models. A separate model was calculated for each GPS variable, with the final post-review match grade as the dependent variable and referee indicator terms included in every model. This approach controls for all stable differences between referees and estimates the association between match-to-match changes in a GPS variable and the official grade within the same referee. Four referees contributed only one match. These singleton observations were retained in the dataset and descriptive analyses, but they did not contribute within-referee variation to the GPS slope estimates. The fixed-effects coefficients were therefore identified by 80 match observations from 12 referees with at least two matches. The number of matches officiated by each referee was not included in the primary model because unequal numbers of repeated observations are accommodated by the fixed-effects specification and because match count may itself be related to appointment frequency or referee performance.
Standard errors were clustered by referee using the CR1 finite-sample correction. For inference on the within-referee GPS coefficients, the 12 referees with repeated observations were treated as the informative clusters, giving 11 cluster-based degrees of freedom (G-1). Because the number of informative clusters was small, exact wild cluster bootstrap-t p-values were used for the primary inference. Rademacher weights were assigned at referee level and all 212 = 4096 possible sign combinations were evaluated. The six GPS variables were treated as exploratory tests, and Holm adjustment was applied to the six exact bootstrap p-values. Statistical significance was set at adjusted p < 0.05. Sensitivity analyses repeated the pooled referee-clustered models both without and with the number of observed matches per referee as a covariate. The Durbin-Watson test from the original analysis was not used for inference in the revised repeated-measures models. The fixed-effects models assume a linear within-referee association between each GPS variable and match grade. Referee clusters were treated as independent, while observations from the same referee were allowed to be correlated.

3. Results

3.1. Participant and Refereeing-Performance Characteristics

Seventeen elite main referees were included in the complete project and 16 had complete GPS and official match-grade data. The mean season-long official grade was 8.21 ± 0.26, with a minimum of 7.30 and a maximum of 8.40. Across the 84 match observations used in the regression models, the mean official match grade was 8.27 ± 0.25, the median was 8.40, and the range was 7.30–8.50. The match-level grade distribution was 7.3 (n = 1), 7.4 (n = 2), 7.8 (n = 1), 7.9 (n = 10), 8.0 (n = 1), 8.1 (n = 2), 8.2 (n = 4), 8.3 (n = 20), 8.4 (n = 31), and 8.5 (n = 12). The 16 referees contributed 7, 2, 1, 10, 1, 10, 4, 2, 11, 9, 8, 1, 9, 3, 1, and 5 matches, respectively. The small range of grades is consistent with a highly selected group from the highest competition level. The characteristics of the referee cohort are presented in Table 1.
Table 1. Characteristics of the elite referee cohort.

3.2. GPS-Derived Match Demands

The mean total distance was 10,987.42 ± 653.19 m. High−intensity running distance averaged 585.70 ± 204.43 m and ranged from 116 to 1150 m. The referees recorded 3.93 ± 2.91 sprints, 36.63 ± 18.56 accelerations, and 42.45 ± 14.89 decelerations. Maximum speed averaged 27.25 ± 1.36 km/h and ranged from 23.69 to 31.40 km/h. The wide ranges in several variables (Table 2) show that physical demands differed considerably between matches.
Table 2. GPS-derived external match-load characteristics.

3.3. Associations Between GPS Variables and Refereeing Performance

In the primary referee fixed-effects models (Table 3), deceleration count showed a positive within-referee association with the final official match grade (B = 0.00676, CR1 SE = 0.00118, 95% exact wild-cluster-bootstrap CI = 0.0036 to 0.0093, exact wild cluster bootstrap p = 0.006). This association remained significant after Holm correction for six GPS variables (adjusted p = 0.038). An increase of 10 decelerations within the same referee corresponded to approximately 0.068 points higher official match grade (95% CI = 0.036 to 0.093). Total distance, high-intensity running distance, sprint count, acceleration count, and maximum speed were not significant after correction. Maximum speed was not significant in the small-cluster bootstrap analysis (B = −0.08298, exact p = 0.248). These results should be considered exploratory because only 12 referee clusters contributed within-referee variation and match context was not included in the models.
Table 3. Primary referee fixed-effects models of match-level GPS variables and official match grade (84 matches, 16 referees; within-referee slope estimates informed by 80 matches from 12 referees).
The sensitivity analyses (Table 4) showed that the isolated maximum-speed association from the original model depended on inclusion of the match-count covariate. Without this covariate, maximum speed was not associated with grade (p = 0.702), whereas deceleration count was associated with grade (p = 0.015). After adding match count, maximum speed reached p = 0.021 and deceleration count changed to p = 0.117. The other four GPS variables remained non-significant in both specifications.
Table 4. Sensitivity analyses using pooled referee-clustered models with and without the number of observed matches per referee as a covariate.

4. Discussion

The revised analysis showed that the relationship between GPS match-load variables and official refereeing performance depends on how repeated observations are modelled. In the primary within-referee analysis, deceleration count showed a positive association with the official match grade and remained significant after Holm correction. Maximum speed was not significant in the exact small-cluster bootstrap analysis. Sensitivity analyses showed that both deceleration count and maximum speed changed according to whether the number of observed matches was included as a covariate. These isolated findings should therefore be interpreted as exploratory rather than as direct indicators of refereeing quality. The criterion variable was the official match grade after additional review by seven experts, rather than an isolated decision obtained from a laboratory or video-based test.
The mean total distance was 10,987.42 ± 653.19 m, with values from 9176 to 12,511 m. These results are close to values reported for referees at the highest national and international levels [8,9,10,22]. Martinez-Torremocha et al. reported 10,417 ± 860 m in the Spanish first division and 10,420 ± 729 m in the second division [8]. High-intensity running in the present sample averaged 585.70 ± 204.43 m, with a wide range from 116 to 1150 m. Sprint count was 3.93 ± 2.91. This was lower than the values reported by Martinez-Torremocha et al. for Spanish first- and second-division referees, approximately 10 sprints per match [8]. The definition used in the present project was demanding because a sprint was recorded only above 25 km/h for more than one second. Some matches contained no registered sprint. Differences in speed zones, competition level, and the movement of players during a match need to be considered when these values are compared across studies [8,9,23].
A low sprint count does not mean that the referee was unable to follow the match. During some matches, there is no need for the main referee to exceed 25 km/h. One possible explanation is that a physically prepared referee with good anticipation and a clear view of play may predict the next movement zone and control the match without frequent maximal-speed running. A referee who starts from a less favourable position may need more sprinting to recover distance after the action has developed. The same explanation may apply to accelerations and decelerations, but positioning and anticipation were not measured directly in this study. The present referees recorded 36.63 ± 18.56 accelerations and 42.45 ± 14.89 decelerations per match. The positive within-referee association observed for deceleration count may indicate more frequent braking or positional adjustment in some matches, but the present data cannot identify the mechanism. In practical terms, 10 additional decelerations within the same referee corresponded to an estimated 0.068-point-higher match grade, although this exploratory estimate should be interpreted cautiously. Match-running studies in top referees have found weak or no relationships between standard fitness-test performance and high-intensity or acceleration demands recorded during matches [24].
Maximum speed averaged 27.25 ± 1.36 km/h, and the highest recorded value was 31.40 km/h. Martinez-Torremocha et al. reported mean maximum speeds of 28.76 ± 2.06 km/h in the Spanish first division and 28.56 ± 1.82 km/h in the second division [8]. The lowest maximum speed in the present sample was 23.69 km/h, which is below the 25 km/h sprint threshold and explains why a referee could finish a match without a registered sprint. Match maximum speed is not the same as maximal sprint capacity measured in a separate test. Previous research showed that football match officials do not necessarily reach their laboratory or field-test maximum sprint speed during competition [25].
Maximum speed was not associated with the official grade in the revised primary analysis. The negative association appeared only in the sensitivity model that included the number of observed matches as a covariate and was not present when that covariate was removed. High speed may still be necessary when the referee follows a counterattack or another rapid transition and must remain close to the next incident. The same high value could also occur when a referee starts from a less favourable position and has to recover distance quickly. These are possible explanations because positioning was not measured directly. The present results do not show that high-speed running lowers decision quality. Acute physical strain immediately before a decision may be relevant, but that question cannot be answered from one maximum-speed value for the whole match [15,17].
The referees in this study were already a selected group from the highest competition level. Before the season, they had to satisfy official fitness requirements and medical criteria, and 10 of the 17 referees were listed by UEFA and/or FIFA. The small variability in physical characteristics and official grades is expected in such a selected group. Once the required physical standard has been reached, differences in refereeing performance may depend more on experience, decision strategy, and the effective use of physical capacity than on one isolated physical measure [18,26,27,28,29,30]. Anticipation and positioning are possible contributors, but they were not measured in the present study. Similar findings have been reported in referee research from other sports, where single anthropometric or physiological variables did not consistently explain differences in official performance grades [28,29,30].
The timing of physical load may be more important than the total amount recorded over 90 min. McEwan et al. found a greater likelihood of decision errors when very high internal or external load was present in the short period before a decision [15]. Acute vigorous exercise can change response-suppression measures [17], and response-suppression ability has been linked with some aspects of disciplinary decision accuracy [18]. The present GPS database contains match-level movement values and cannot connect each sprint or high-intensity action with the exact time of a correct or incorrect decision. A referee can have a moderate total match load and still experience a very demanding period immediately before a key incident.
Whole-match averages can hide the most demanding periods of refereeing [9]. GPS data should be examined together with the time and context of specific match events. Short periods of 10, 30, or 60 s before a decision could show whether the referee was exposed to unusually high speed, acceleration, or deceleration at that moment. Match context is particularly important for the present results because match pace, rapid transitions, team style, scoreline, and the location of decisive incidents were not available for the regression models. Technical and tactical characteristics of the game influence the running performance of elite referees, and experience contributes to how referees adapt their movement [19]. A difficult high-tempo match could increase running demands and make refereeing more difficult at the same time. Therefore, the present associations cannot be interpreted independently of possible match-context effects.
The criterion variable in this study differs from the decision outcomes used in many previous referee studies. Individual decisions have been evaluated with expert panels, simulations, video tasks, or selected incident categories [14,15,16]. Carvalho et al. studied variables within formal referee performance assessments [20]. The present study used the official post-match grade from the national federation. After every round, seven experts independently reviewed the complete match recording and relevant VAR footage, assigned numerical grades, and reached the final grade by consensus. They did not have access to the original referee-observer report or to GPS data. The study database contained only the final committee-approved grade, so inter-rater reliability could not be calculated. This criterion reflects full-match performance and the consequences of important decisions, not only the accuracy of one selected incident.

4.1. Practical Implications

GPS monitoring is useful for referee preparation, but the values should not be interpreted as quality scores. Total distance, sprint count, acceleration count, deceleration count, and maximum speed describe the physical work completed during a match. A referee who covers more distance or reaches a higher speed is not necessarily officiating better. Anticipation and positioning may reduce unnecessary movement while allowing the referee to maintain a clear view of important events, but these factors were not measured directly in the present study.
GPS data can be used to quantify match demands, plan individual conditioning, identify unusually demanding matches, and support recovery. They can be compared with the referee’s usual match profile and used to prepare training that includes short accelerations, high-intensity running, braking, turning, and recovery movements. The practical goal is not to maximise every GPS variable. Referees need enough physical capacity to follow the game and to use that capacity efficiently during different match situations. GPS data can support referee preparation, but they should not be used independently to evaluate or rank refereeing quality.

4.2. Strengths and Limitations

The main strength of this study is the level and completeness of the sample. All 17 main referees eligible to officiate in the highest Croatian competition during the 2022/2023 season took part in the project, including 10 referees listed by UEFA and/or FIFA. The study covered 100% of the target referee population for that season, and the field analysis included 84 official matches rather than simulated tasks. A second strength is the criterion used to assess refereeing performance. Official match grades were reviewed after every competition round by seven experts appointed by the Croatian Football Federation Referees Committee. Each expert independently reviewed the complete match recording and relevant VAR footage without access to the original referee-observer report or GPS data, and the final official grade was reached by consensus. This procedure reduced dependence on a single observer and provided an additional expert-review process before the final grade was used as the criterion variable.
The main limitation is the small absolute number of referees. The 84 match observations came from 16 referees and individual referees contributed between 1 and 11 matches. Four referees contributed only one match and therefore did not contribute within-referee variation to the GPS slope estimates. The fixed-effects analysis accounts for repeated observations within referees, and exact wild cluster bootstrap inference was based on the 12 informative referee clusters, but the small number of independent referees still limits statistical precision. The group was homogeneous and the official grades had a small range. The GPS field analysis covered one national competition and one half-season. Match context was not included in the models; match pace, transitions, team style, scoreline, and the location of important incidents could influence both GPS values and refereeing difficulty. GPS variables do not show viewing angle, distance from the incident, line-of-sight obstruction, or the tactical reason for a sprint, acceleration, or deceleration. Anticipation and positioning were not measured directly. The study database contained only the final committee-approved grade and not the seven individual expert ratings, so inter-rater reliability could not be calculated. The experts did not have access to GPS data during the grading procedure. STATSports APEX Pro units operating at 10 Hz were used, and data were processed in Sonra 4.0 with the default absolute acceleration and deceleration thresholds (>3 m·s−2 and <−3 m·s−2, respectively); the firmware version was not available in the archived study records. Finally, sensitivity analyses showed that the estimates for deceleration count and maximum speed changed when the number of observed matches was included as a covariate, which confirms that these isolated associations should be interpreted cautiously.

4.3. Future Research

Future studies should connect GPS timestamps with time-stamped match incidents and analyse the movement that occurs immediately before important decisions. Running speed, acceleration, deceleration, distance from the incident, viewing angle, and the final decision could then be examined together. This would provide a more direct test of the relationship between acute external load and decision quality.
Future research should include referees from several leagues and more than one season to increase the variability of referee abilities and official grades. Repeated measurements should be used to separate differences between referees from changes within the same referee between matches. GPS data should also be combined with physiological measures and video analysis of referee positioning.

5. Conclusions

The revised analysis showed that GPS-derived external match-load variables were not uniformly associated with official refereeing performance in elite football referees. In the primary within-referee analysis, deceleration count showed a positive association with official match grade and remained significant after Holm correction (B = 0.00676, 95% exact wild-cluster-bootstrap CI = 0.0036 to 0.0093, exact bootstrap p = 0.006, adjusted p = 0.038). Total distance, high-intensity running distance, sprint count, acceleration count, and maximum speed were not significant after correction. The maximum-speed association reported in the original model was not robust across the revised analyses and depended on inclusion of the match-count covariate.
The present findings do not support the use of greater physical output as a direct indicator of better refereeing performance. The isolated deceleration result should also be interpreted as exploratory because the number of referees was small and match context was not available. GPS data are most useful when they are interpreted together with the demands of the match and the situations in which movement occurs. The use of GPS data together with official expert-reviewed grades can provide additional information about the relationship between physical match demands and the quality of refereeing.

Author Contributions

Conceptualization, J.T., I.B. and M.D.; methodology, J.T. and M.D.; software, I.K. and M.D.; validation, J.T., I.B. and M.D.; formal analysis, J.T., I.B. and M.D.; investigation, I.K. and J.T.; resources, J.T.; data curation, I.K.; writing—original draft preparation, I.K., J.T., I.B. and M.D.; writing—review and editing, I.K., J.T., I.B. and M.D.; visualization, J.T., I.B. and M.D.; supervision, I.K. and I.B.; project administration, I.K. and I.B. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Committee for Scientific Research and Ethics of the Faculty of Kinesiology, University of Zagreb (approval number 13/2023; approved on 22 February 2023).

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request. Access to official referee performance grades is subject to the confidentiality requirements of the Croatian Football Federation.

Acknowledgments

The authors thank all referees who participated in the study and the Croatian Football Federation for its cooperation and support. During the preparation of this manuscript, the authors used ChatGPT (OpenAI, GPT-5.6) for language editing, manuscript structuring, and statistical-result verification. The authors reviewed and edited all outputs and take full responsibility for the content of the publication.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Reilly, T.; Gregson, W. Special populations: The referee and assistant referee. J. Sports Sci. 2006, 24, 795–801. [Google Scholar] [CrossRef] [Scilit]
  2. Castagna, C.; Abt, G.; D’Ottavio, S. Physiological aspects of soccer refereeing performance and training. Sports Med. 2007, 37, 625–646. [Google Scholar] [CrossRef] [Scilit]
  3. Krustrup, P.; Helsen, W.; Randers, M.B.; Christensen, J.F.; MacDonald, C.; Rebelo, A.N.; Bangsbo, J. Activity profile and physical demands of football referees and assistant referees in international games. J. Sports Sci. 2009, 27, 1167–1176. [Google Scholar] [CrossRef] [Scilit]
  4. Martinho, D.V.; Field, A.; Rebelo, A.; Gouveia, E.R.; Sarmento, H. A systematic review of the physical, physiological, nutritional and anthropometric profiles of soccer referees. Sports Med. Open 2023, 9, 72. [Google Scholar] [CrossRef] [Scilit]
  5. Zhang, L.; Geok, S.K.; Wazir, M.R.W.N.; Qin, L. Physical demands and physiological response of soccer referees in high-level matches: A systematic review. PLoS ONE 2025, 20, e0315403. [Google Scholar] [CrossRef] [Scilit]
  6. Cummins, C.; Orr, R.; O’Connor, H.; West, C. Global positioning systems (GPS) and microtechnology sensors in team sports: A systematic review. Sports Med. 2013, 43, 1025–1042. [Google Scholar] [CrossRef] [Scilit]
  7. Varley, M.C.; Fairweather, I.H.; Aughey, R.J. Validity and reliability of GPS for measuring instantaneous velocity during acceleration, deceleration, and constant motion. J. Sports Sci. 2012, 30, 121–127. [Google Scholar] [CrossRef] [Scilit]
  8. Martinez-Torremocha, G.; Martin-Sanchez, M.L.; Garcia-Unanue, J.; Felipe, J.L.; Moreno-Perez, V.; Paredes-Hernandez, V.; Gallardo, L.; Sanchez-Sanchez, J. Physical demands on professional Spanish football referees during matches. Sci. Med. Footb. 2023, 7, 139–145. [Google Scholar] [CrossRef] [Scilit]
  9. Yousefian, F.; Zafar, A.; Peres, P.; Brito, J.; Travassos, B.; Figueiredo, P. Intensity demands and peak performance of elite soccer referees during match play. J. Sci. Med. Sport 2023, 26, 58–62. [Google Scholar] [CrossRef] [Scilit]
  10. Mallo, J.; Navarro, E.; Garcia-Aranda, J.M.; Helsen, W. Activity profile of top-class association football referees in relation to fitness-test performance and match standard. J. Sports Sci. 2009, 27, 9–17. [Google Scholar] [CrossRef] [Scilit]
  11. Mallo, J.; Frutos, P.G.; Juarez, D.; Navarro, E. Effect of positioning on the accuracy of decision making of association football top-class referees and assistant referees during competitive matches. J. Sports Sci. 2012, 30, 1437–1445. [Google Scholar] [CrossRef] [Scilit]
  12. Johansen, B.T.; Erikstad, M.K. A preliminary analysis of the importance of distance, angle, and insight when soccer referees make penalty decisions. Front. Sports Act. Living 2021, 2, 595703. [Google Scholar] [CrossRef] [Scilit]
  13. Vater, C.; Schnyder, U.; Muller, D. That was a foul! How viewing angles, viewing distances, and visualization methods influence football referees’ decision-making. Ger. J. Exerc. Sport Res. 2024, 54, 476–485. [Google Scholar] [CrossRef] [Scilit]
  14. Mascarenhas, D.R.D.; Button, C.; O’Hare, D.; Dicks, M. Physical performance and decision making in association football referees: A naturalistic study. Open Sports Sci. J. 2009, 2, 1–9. [Google Scholar] [CrossRef] [Scilit]
  15. McEwan, G.P.; Unnithan, V.B.; Easton, C.; Glover, A.J.; Arthur, R. Decision-making accuracy of soccer referees in relation to markers of internal and external load. Eur. J. Sport Sci. 2024, 24, 659–669. [Google Scholar] [CrossRef] [Scilit]
  16. Castillo-Rodriguez, A.; Alejo-Moya, E.J.; Figueiredo, A.; Onetti-Onetti, W.; Gonzalez-Fernandez, F.T. Influence of physical fitness on decision-making of soccer referees throughout the match. Heliyon 2023, 9, e19702. [Google Scholar] [CrossRef] [Scilit]
  17. Albaladejo-Garcia, C.; Caballero, C.; Asencio, P.; Moreno, F.J. Acute effects of a vigorous-intensity warm-up on response suppression and decision-making of football referees. J. Sports Sci. 2024, 42, 1081–1089. [Google Scholar] [CrossRef]
  18. Albaladejo-Garcia, C.; Ignoul, E.; Demarez, M.; Caballero, C.; Helsen, W.F.; Moreno, F.J. Response suppression and decision-making in association football referees. Psychol. Sport Exerc. 2025, 80, 102908. [Google Scholar] [CrossRef] [Scilit]
  19. Martinez-Torremocha, G.; Alonso-Callejo, A.; Garcia-Unanue, J.; Gallardo, L.; Sanchez-Sanchez, J.; Felipe, J.L. How do technical and tactical demands of football matches affect the physical performance of elite football referees? Exploring the role of experience. J. Strength Cond. Res. 2024, 38, e664–e669. [Google Scholar] [CrossRef] [Scilit]
  20. Carvalho, V.; Esteves, P.T.; Nunes, C.; Helsen, W.F.; Travassos, B. The assessment of the match performance of association football referees: Identification of key variables. PLoS ONE 2023, 18, e0291917. [Google Scholar] [CrossRef] [Scilit]
  21. Weston, M.; Drust, B.; Gregson, W. Intensities of exercise during match-play in FA Premier League referees and players. J. Sports Sci. 2011, 29, 527–532. [Google Scholar] [CrossRef] [Scilit]
  22. Costa, E.C.; Vieira, C.M.A.; Moreira, A.; Ugrinowitsch, C.; Castagna, C.; Aoki, M.S. Monitoring external and internal loads of Brazilian soccer referees during official matches. J. Sports Sci. Med. 2013, 12, 559–564. [Google Scholar]
  23. Gomes, R.; Mendes, R.; Ferreira, A.; Mendes, R.; Dias, G.; Martins, F. Physical and physiological demands of amateur Portuguese field and assistant football referees. Sports 2024, 12, 133. [Google Scholar] [CrossRef] [Scilit]
  24. Riiser, A.; Andersen, V.; Castagna, C.; Pettersen, S.A.; Saeterbakken, A.; Froyd, C.; Ylvisaker, E.; Kjosnes, T.N.; Moe, V.F. The construct validity of the CODA and repeated sprint ability tests in football referees. Int. J. Sports Med. 2018, 39, 619–624. [Google Scholar] [CrossRef] [Scilit]
  25. Castillo, D.; Camara, J.; Castellano, J.; Yanci, J. Football match officials do not attain maximal sprinting speed during matches. Kinesiology 2016, 48, 207–212. [Google Scholar] [CrossRef] [Scilit]
  26. Weston, M.; Castagna, C.; Impellizzeri, F.M.; Bizzini, M.; Williams, A.M.; Gregson, W. Science and medicine applied to soccer refereeing: An update. Sports Med. 2012, 42, 615–631. [Google Scholar] [CrossRef] [Scilit]
  27. MacMahon, C.; Helsen, W.F.; Starkes, J.L.; Weston, M. Decision-making skills and deliberate practice in elite association football referees. J. Sports Sci. 2007, 25, 65–78. [Google Scholar] [CrossRef] [Scilit]
  28. Belcic, I.; Ruzic, L.; Marosevic, A. Influence of functional abilities on the quality of refereeing in handball. Balt. J. Health Phys. Act. 2020, 12, 23–34. [Google Scholar] [CrossRef] [Scilit]
  29. Belcic, I. Does age, experience and body fat have an influence on the performance of handball referees? Appl. Sci. 2022, 12, 9399. [Google Scholar] [CrossRef] [Scilit]
  30. Belcic, I.; Ruzic, L.; Krakan, I. The effect of physiological load during matches on the quality of refereeing in handball referees. Ann. Appl. Sport Sci. 2022, 10, e1082. [Google Scholar] [CrossRef] [Scilit]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Article Metrics

Citations

Article Access Statistics

Journal Statistics
Multiple requests from the same IP address are counted as one view.
Fluence-Dependent Multimodal Response of NdGaO3(110) Single Crystals to 147 MeV Kr Ions
Previous
Stage-Aware Optimization of Llama 3 Model Inference for Language Services on Many-Core CPUs
Next