The cauldron has cooled down: a systematic literature review on home advantage in football during the COVID-19 pandemic from a socio-economic and psychological perspective

Following the work by Courneya and Carron (1992), we understand the home advantage as “the consistent finding that home teams in sport competitions win over 50% of the games played under a balanced home and away schedule” (p. 13). First empirical evidence for the home advantage in sport competition was provided by Schwartz and Barsky (1977) who analyzed the major leagues in baseball, American football, hockey, and basketball as well as the U.S. college competition in football and basketball. There are various influencing parameters, which Courneya and Carron (1992) systematize as follows:

Game location factors comprise four determinants that can affect the visiting and home team differently. This includes the crowd factor (1.), according to which the home team receives greater support at home than the visiting team. Learning/familiarity factors (2.) assume that the home team is better acquainted with the location and also has the possibility to redesign it at short notice. In football, the watering of the lawn could be an example. Travel factors (3.) means the consideration that visiting teams have to travel to the venue and thus have to take on inconveniences to which the home team is not exposed. Finally, under rule factors (4.), components of the set of rules are subsumed that provide for favoring the home team. An example is the last line change in ice hockey (Carron et al. 2005).

These four game location factors directly affect the critical psychological, behavioral and performance states of the three relevant groups (competitors, coaches and officials). Coruneya and Carron (1992) differentiate here between three levels: The primary level describes the basic level of performance (such as the distances covered in football or the number of sprints). The second level describes the intermediate or scoring aspect of the performance (e.g. the number of goals scored). The third level records the result measure (in football this would be the number of points that a team scores in a game).¹

While, for example, travel efforts for teams stayed identical during the COVID-19 pandemic,² crowd influence is of particular interest when analyzing the effects of the COVID-19 pandemic on sporting events. Relevant parameters here are crowd size or density (e.g., Dowie 1982; Goumas 2012; Pollard 1986). Structure and properties of the stadium—the basis for the production of atmosphere by fans and spectators—is also considered as a significantly influencing factor in previous studies (Unkelbach and Memmert 2010).

In addition to their classic function as demanders and consumers of the entertainment service produced, fans also play a role in other parts of the value creation process. Woratschek et al. (2014) emphasize the value contribution for the production by, e.g., creating banners or preparing creative choreographies or supporting their team with battle chants. In this context, the ghost games of the COVID-19 pandemic particularly illustrate the high relevance of fans for atmospheric and emotional sports entertainment.

The atmosphere within a stadium can be an important and beneficial factor for football fans (Flatau and Emrich 2016) and—although this finding may be intuitive—it is important to state that the typical stadium atmosphere is an original output of the fans themselves. In this context, we further follow the argument of Ednesor (2015) who states that “atmosphere is ( …) a co-production that involves players, match organisers, and fans” (p. 82). Both sides contribute to the production of the game atmosphere through mutual interaction. In Mauss' (1966) sense, the relation can even be interpreted as an exchange. From the player’s perspective, the stadium represents a stage, similar to that of an actor in a theater. Basically, it is about the self-presentation of the players in public. But the player needs the spectator, the big stage, to present himself and his performance (Gebauer 1972; Goffman 1959; Horky 2020). The fans cheer on the team and in return the team fights on the pitch, so that the marketable entertainment service "football" is produced as a result. This makes the unique situation of the COVID-19 pandemic especially interesting as there are no fans in the stadium, creating a highly unusual and mostly first-time situation for many players and clubs. Furthermore, there is the commonly known circumstance that supporters are considered as the "12th man" who influence the outcome of the game in favor of their team (Edensor 2015), making ghost games especially intriguing for socio-economic and psychological analysis. Indeed, analyses of football matches played behind closed doors in the pre-COVID-19 era evidenced the impact of the fans on referee decisions (Pettersson-Lidbom and Priks 2010; Reade et al. 2020a, b).

The opportunities presented by the COVID-19 pandemic to study home advantage have been used for different team sports. McHill and Chinoy (2020) used the COVID-19-related changes of the game settings to analyze mainly the effect of traveling on the team performance. The NBA ‘s 2019/2020 season was continued after an approximately six-month break with 22 teams in Orlando ("the bubble"). In the first games, spectators were completely excluded. McHill and Chinoy (2020) showed that traveling across time zones has a negative effect on winning percentage, team shooting accuracy, and turnover percentage. Furthermore, traveling in general has a negative effect on offensive rebounding and increases the number of points the opposing (home) team scores.

Higgs and Stavness (2021) showed on the basis of a Bayesian multilevel regression model that in the NHL and the NBA the home advantage decreased due to the exclusion of spectators, while in the MLB and the NFL hardly any effects were found. These results concerning the NBA were confirmed by Leota et al. (2021). In their study, in which they used mixed models, they showed that in games with crowds, the home team won 58.65% of the games. In games without crowds, the home team only won 50.60% of the games.

Guérette et al. (2021) examining the influence of crowds on the number of penalties called by referees in the NHL, came to similar conclusions as Higgs and Stavness (2021) with regard to the NHL. They showed that in the presence of crowds, home teams are favored by referees concerning penalties. If games take place without crowds, the number of penalties awarded does not differ significantly between away teams and home teams.

The results of Higgs and Stavness (2021) for the MLB are supported by Losak and Sabel ‘s (2021) research. According to their analysis, the presence of the crowd is not a driver of the home field advantage. Likewise, according to their research, travel fatigue does not seem to have any influence on home advantage. Zimmer et al. (2021) come to comparable results in their study of the MLB; although there might be differences between star-batters and nonstar-batters (Jane 2021).

Fioravanti et al. (2021) examined 1027 rugby union matches from 11 tournaments in 10 countries. They came to the conclusion that the exclusion of spectators has the effect that home teams win less matches and that their points difference decreases.

A large number of studies on ghost games have been conducted and published during the COVID-19 pandemic. In order to provide a comprehensive overview on these studies’ data, methods and findings, we decided to analyze the potential effects of ghost games on football with the help of a systematic literature review.³ This approach is frequently used in psychology, economics and management research to present the state of the art in a certain field of research (Fisch and Block 2018; Frank and Hatak 2014; Webster and Watson 2002).⁴

On April 28th, 2021 we used the databases EBSCO-Host, EconStor, SURF, Emeraldinsight, JSTOR, Sciencedirect, Springerlink and Google Scholar to search with the terms COVID-19, football, soccer, behind closed doors, ghost game, and home advantage for related studies. These databases were selected because we assume that the relevant topic is comprehensively covered due to the breadth of these databases. We used the following search strategies:

The keywords mentioned were only searched for in the titles of the articles. Newspaper articles and comments were excluded,⁵ so that in the end only articles from journals and working papers remained. Papers that were not relevant for the scope of the present work (e.g., investigating sport-medical consequences of ghost games) were also excluded.⁶ This was particularly the case with the EconStor and Springerlink databases. Against this background, Table 1 shows the following results that were obtained.

Additionally, we manually worked through the references of these studies in order to find manuscripts that had been overlooked by our initial search. Here, one study provided only descriptive data and the focus was not on home advantage, and one paper was published in a journal whose doi was not accessible. After we removed duplicates, originally 16 studies (10 peer-reviewed) formed our sample to be analyzed. In this context, the Econstor database in particular showed that sometimes several hits related to the same article. Especially the work from Reade et al. (2021) provided valuable information to fully complete our dataset. After the review process—which was finalized in December 2021—26 studies (20 peer-reviewed) were finally included into the present analysis.⁷ Note that all studies included are marked with a hashtag (#) in the references section.

Besides basic properties and information (year, authorship, journal, peer-review status, etc.) the following characteristics of every study were extracted and documented in detail: number of reported countries, number of analyzed leagues, method of comparison (e.g., pooled and/or not pooled leagues), number and properties of analyzed factors (e.g., goals, cards, fouls, etc.), number and properties of evaluation methods and statistical approaches, results and (central) conclusion.

In order to quantify these factors, we created an individual “Spectrum score” for every included study—used as a proxy that allows a comparison in terms of breadth and depth of analysis, methodological design, statistics used and factors included. In this way we assessed the studies’ crucial components: “Leagues”, “Leagues comparison”, “Seasons”, “Main factors”, “Side factors” and “Evaluation”. To achieve a final score for each work the studies were analyzed individually for similarities and differences in their analytical approach, and weightings were used respectively (also see supplementary material) to provide a balanced assessment of the included characteristics. The weighting was based on calculated mean values across all studies. This approach prevented individual study characteristics from having a biasing influence on the respective overall picture, represented in the concluding “Spectrum score”, which calculation is based on the following factors:

The sum of the individual values from the list from A to E forms the "Spectrum score" and thus reflects a concise and objective comparative value—based on included factors and methods—between the different studies.

Besides calculating an individual “Spectrum score” for every study we additionally set another individual value—the “Conclusion score”—representing the central conclusion of each study in terms of the magnitude of the impact of ghost games on home advantage. In order to set this score, we screened every paper for crucial text passages (usually found in the discussion or conclusion section of the manuscript) indicating the authors' verdict on the magnitude of the home advantage in ghost games, based on their findings. A 7-point non-directed likert scale (ranging from (1) “strongly increased home advantage in ghost games” to (4) “no change in home advantage in ghost games” and (7) “strongly reduced home advantage in ghost games”) was used to represent the studies’ conclusion as precisely as possible.

Summing up, we assigned two values to every included study, the “Spectrum score” and the “Conclusion score”. While the "Spectrum score" represents a central statement regarding the depth, breadth, and empirical amplitude of the respective study, the "Conclusion score" reflects the central statement of the authors regarding the influence of ghost games on the home advantage.

Based on the results from our systematic literature review we state the following main findings:

With regard to existing explanatory approaches in the literature, theories emerging from social pressure and conformity seem to distill as an empirically conclusive explanatory model for the found effect of decreased home advantage in ghost games. Most studies suggest that a relationship between spectator absence and home advantage can be inferred via the influence on the referee(s) and from a missing effect on players’ emotions, originating from the (home) crowd. Although a less emotional behavior (e.g., Webb 2020), especially a lower level of aggression towards referees, may be desirable from a sporting perspective, it seems to come at the expense of the home advantage and is reflected in a lower sporting success of the home team. The extent of decrease of the home advantage seems to depend on different factors. For example, Fischer and Haucap (2021) analyze matches from the first, second and third league in Germany. Surprisingly, with respect to COVID-19 ghost games, they find a significant decrease in home advantage only for the first Bundesliga when matches are played excluding spectators, possibly due to the larger delta between matches with spectators and ghost games in the first Bundesliga. Furthermore, it is remarkable that a stadiums’ atmospheric conditions seem to significantly depend on its properties (e.g., number of standing places, distance to the field) (Unkelbach and Memmert 2010). If one considers that stadiums are mostly occupied by fans who actively produce atmosphere by singing and cheering for the home team, it can be assumed that the quality of the spectators in terms of their input is relevant for the teams’ success. In this context the power of fans is also becoming increasingly important regarding upcoming reforms in European soccer. A possible European Super League has been discussed for years (e.g., Follert 2019; Follert and Daumann 2021; Follert and Emrich 2020; Littkemann, Geyer and Schmitz, 2021), and time and again the preferences of large parts of the fan scene have been neglected. In the light of our findings from the ghost games of the COVID-19 pandemic, the importance of fans becomes clear not only as a production factor in an economic sense, but also as a significant variable from a sport-psychological perspective.

The question now arises as to what implications the empirical evidence might have regarding the club’s governance. In the sports economics literature, it has been discussed for several years to explicitly consider fans in the club's objective function ("fan welfare maximization", e.g., Madden 2012; Madden and Robinson 2012). Although this demand may be criticized with reference to the importance of ownership in a market economy system and the advantages of having an investor as “sugar daddy” (Franck 2010; Richau et al. 2021), it is worth considering compensation for fans for providing the factor of production (Follert et al. 2020). Based on our findings in this present study, we even take a step further and argue that a discussion regarding compensation of fans after sporting success (e.g., winning the championship) is plausible and worth considering.

That fans also hold a majority stake in a football club is not unheard of. For example, the Exeter Supporter's Trust holds a majority stake in Exeter City FC, which plays in the English League Two (4th division). Although the supporters achieve a consumption benefit from that football match as it is, there is already reciprocity here in that they pay the admission price for this. The bargaining position of the fans is certainly strengthened by the empirical data provided by the studies that we found in our review, so that they could possibly demand greater influence on strategic and operational club decisions. However, it must be considered that fans have strongly limited alternatives for time allocation, provided that they want to spend their free time-consuming football. It is easy to see that a fan of FC Bayern Munich will not switch to Borussia Dortmund if “his” or “her” club denies him or her recognition. Thus, due to their preference structure, fans suffer considerable utility losses when choosing "migration" (Hirschman 1970). Therefore it can be concluded that the stronger the bond between club and fan, the harder the exit will be, since the alternative provides only a comparatively small benefit. With respect to the differentiation provided by Giulianotti (2002), a spectator who is classified as hot and traditional will almost never switch to another club. This does not mean, of course, that the clubs can act without paying any attention to the fans. Rather, it seems necessary for clubs to produce a minimum level of sporting success in order to keep fans in line in the medium and long term, which in turn has corresponding implications for sports management (signing of players, ticket pricing, etc.).

Besides that, the market power of the fans must not be overestimated from a different perspective: The market power of the fans essentially depends on their level of organization. If the fans can confront the clubs as a closed cartel, they are certainly able to assert their interests. Social media enable such an organization and reduce the corresponding communication costs, but on the one hand, the number of fans is very high and on the other hand, their interests are often very differentiated. Thus, an appearance of the fans as a closed cartel seems rather unlikely, which means that the price-setting scope of the clubs should remain comparatively high.

However, the role of fans can also be used for strategic purposes. It can be interesting for the organizer of national as well as international competitions to choose the venue and/or matchdays (Goller and Krumer 2020) in such a way that the role of the fans is marginalized, and that a “bias-free athletic performance” of the teams dominates the result of the game. At the same time, it should be noted that this can of course also have effects on the other sub-markets such as the market for sponsoring, TV broadcasting rights and merchandising.

It could also be interesting for the visiting team to purchase tickets for away games and distribute them free of charge to their own fans, in the hope that this subsidy will increase the number of their own fans in the away game and thus at least partially eliminate the home advantage.

In line with the classical conceptual framework for research on home advantage (Courneya and Carron 1992), we found strong evidence for the crowd as a crucial game location factor. In our systematic literature review the effect of the missing supporters was not only evident with respect to performance and outcome measures (i.e., win ratio, points, goals) in the majority of the studies, but also in measures more directly related to the critical behavioral states of players and referees, such as match dominance (i.e., shots, shots on target), fouls, and awarded cards (Dilger and Vischer 2020; Sors et al. 2020).

From a behavioral science perspective, the consequent relevant question would be about the underlying psychological states of the players and referees. At the moment, the relationship between behavioral and psychological states can only be indirectly inferred from the present data (Webb 2020). Anecdotal evidence from interviews with players (Guardian Football 2020; Hamilton 2021) and referees (UEFA.tv 2020; ZDFsport 2020), however, indicates a substantial impact of the missing supporters on the subjective experience of these sports professionals.

Besides qualitative interviews and self-report questionnaires, a promising approach was recently put forward by Leitner and Richlan (2021b). Their “Analysis System for Emotional Behavior in Football “ (ASEB-F) is a video-based categorical analysis system of nonverbal behavior during football matches. It assumes that emotions can be observed and described as an organized psychophysiological reaction to specific events in the environment, rising to overt actions and leading to human (nonverbal) behavior. The ASEB-F was used to video analyze the behavior of players and officials in 20 games of FC Red Bull Salzburg before and during the COVID-19 pandemic. There were about 20% fewer emotional situations in matches without supporters compared to matches with supporters. In addition, referees were markedly less actively involved in these emotional situations. The results indicate that the absence of supporters has a substantial influence on the experience and behavior of players and officials alike (Link and Anzer 2021).

Possibly related to the psychological effects is the question of whether there are particularly home strong or home weak teams, and, if yes, what the underlying mechanisms are. In addition, our systematic literature review revealed studies which reported differences in the effect of the missing supporters on the home advantage between leagues within countries (e.g., first vs. second divisions) and differences between leagues across countries (e.g., German Bundesliga vs. English Premier League). Therefore, not only psychological but also socio-cultural explanations for the effects in question have to be taken into account (e.g., Sánchez and Lavín, 2021).

Apart from football, there is of course the question of how home field advantage changes in other sports. If one follows the model of Courneya and Carron (1992), then a large number of other sports would have to be affected by the loss of the audience. In order to draw a complete picture of the actual impact of the COVID-19 pandemic on home field advantage in general, studies are needed that are not limited to individual sports but combine as many different disciplines as possible.

In summary, pending questions for future research on the home advantage in football concern (among others) are: (a) the psychological basis of the behavioral effects, (b) differences between teams within leagues, (c) differences between leagues within countries, (d) differences between leagues across countries, and (e) the effects of partial attendance (i.e., only a limited number of—primarily—home fans allowed in the stadiums as a measure of normalization after the COVID-19 pandemic) on the home advantage.

When analyzing the data and interpreting the conclusions of the individual studies, the question of a potential publication bias (sometimes also referred to as “file-drawer problem”) came up when looking at the results of our review. Our analyses show that none of the 26—in some cases significantly different—studies included, conclude an increase in home advantage during ghost games. Likewise, the strong clustering in categories 6 & 7 (decreased & strongly decreased home advantage) of the "Conclusion score" is striking (69.2% of all studies). In a similar vein, only 6 of 26 studies concluded that the home advantage did not change significantly as a result of the ghost games in European football. In the light of these findings, it is quite possible that the ghost games might indeed have brought a significant reduction in home advantage. Nevertheless, the possibility must be considered that results from other studies—that were not published due to “scientifically unpopular results”—would potentially weaken the effect of significantly reduced home advantage during the COVID-19 related ghost games in our review study.

Another key issue to consider when addressing the question of the impact of ghost games on home advantage is how to operationalize home advantage. In this context, different approaches and various constructs can be found in the scientific literature (e.g., Leitner and Richan, 2021a; Matos et al. 2020). While some studies choose the win-ratio or gained/lost points to represent home advantage, other study authors decided to rather analyze the distribution of yellow and red cards to the home and away teams. There is also the empirical approach to analyze match specific aspects, such as match dominance (e.g., characterised by ball possession, shots on target or successful tackles) or other sport-performance related characteristics. Especially in a review study, this divergent approach creates potential problems in an inferential statement. However, our present work does not attempt to make a judgement on these different approaches. Rather, we argue that these circumstances need to be taken into account in a next evaluation, but that at the same time it does justice to a broad overall picture of the influence of ghost games on home advantage in professional football.