Printer Friendly

Home advantage and crowd size in soccer: a worldwide study.

Twenty years ago Courneya and Carron (1992) described home advantage as "the consistent finding that home teams in sport competitions win over 50% of the matches played under a balanced home and away schedule", and since then its existence has been well established in a range of team sports including association football or 'soccer' (Nevill & Holder, 1999; Pollard & Pollard, 2005). Home advantage in soccer occurs worldwide, at almost every level of the game, and at the elite level about 60% of all competition points gained are at home, more than that in most other sports investigated (Pollard & Pollard, 2005).

Four major factors have been identified that may contribute to home advantage in sport: adverse effects of travel, familiarity with local conditions, rules favouring the home team, and crowd support (Courneya & Carron, 1992). In soccer, travel effects and familiarity may play some role in home advantage, rule factors play little or no role, and although home team crowd support appears to play a major role the mechanisms through which it operates are unclear (Pollard, 2008). The focus of the present study is on crowd size and how its contribution to home advantage in soccer may vary worldwide.

Although home advantage has been shown to vary greatly within continents--for example it is much higher in the Balkans than in the rest of Europe, and higher in the Andean nations than in the rest of South America--there appears to be little difference between continents as a whole (Pollard, 2006a). During the period 1998 to 2003 home advantage in domestic soccer leagues--in terms of the percentage of competition points gained by home teams--was 61% in Europe, Africa and Australia, 64% in South America, 62% in North and Central America, and 59% in Asia (Pollard, 2006a).

Several studies have investigated the effect of crowd size on home advantage in European soccer. An analysis of English soccer leagues (Pollard, 2006b) found that home advantage was greater in the top four leagues (about 60%) than in the lower five leagues (about 55%) where average crowd sizes were much lower. However, home advantage varied little between the top four leagues even though their average crowd size ranged from less than 5,000 to over 30,000. A study of Scottish soccer leagues (Nevill, Newell & Gale, 1996) showed a similar pattern of results, with the same level of home advantage (about 60%) observed in each of the top two divisions despite a four-fold difference in average crowd size, and little or no home advantage in the third division which attracted much smaller crowds. In a regression analysis of individual matches in the English Premier League (Boyco, Boyco & Boyco, 2007) home advantage increased significantly by 0.09 goals for every 10,000 person increase in attendance.

Although the abovementioned studies provide evidence that home team crowd support contributes to home advantage in soccer, there is no consensus on what the nature of this association may be. To investigate the role crowd size plays in home advantage this study used match data from international club soccer competitions in four confederations of the International Federation of Association Football (FIFA). These tournaments provide a rich source of data for investigating crowd effects on home advantage as crowd sizes range from a few hundred to over 50,000 in each competition, as opposed to domestic leagues where there tends to be much less variation in attendance. The objectives of this study were two-fold: (1) to compare the level of home advantage in soccer competitions in four different continents; and (2) to investigate how home advantage in each of these continents varies according to crowd size.



The data used in this study were all matches from recent seasons (up to 2011) of the major international club soccer leagues in four FIFA confederations representing Europe, Asia (including the Middle East), North America (including Central America and the Caribbean) and South America, and are summarised in Table 1. Only those seasons with sufficiently complete (> 90%) crowd size data were included. Each league follows a similar format, with entry based on a team's performance in their domestic competitions the previous season. Each tournament begins with one or more qualifying rounds, followed by a round-robin group stage and then a finals stage. Almost all matches are played in pairs, one at each team's home ground. Each of the group stage matches are decided on their own, whereas matches in the qualifying and finals stages (except for some final matches) are played as 'two-legged ties' won by the team scoring the most goals over two matches. Seven final matches were excluded from the analysis as they were played at neutral venues. The qualifying round in the Asian Champions League was also excluded due to the small number of matches (< 5 per season).

For each match the total number of goals scored was available for each team, as well as venue and crowd size. All match data were obtained from (the official confederation websites provided limited data in terms of individual match results and crowd size). Summary data for each league were checked for accuracy against those from the official confederation websites; no discrepancies were found.


As about one third of matches in this study were decided by the total number of goals scored over two home and away legs, the conventional method of calculating home advantage in terms of competition points gained was not appropriate (Pollard, 1986). Instead, home advantage was expressed as the percentage of all goals scored at home; that is, the total number of goals scored by home teams divided by the total number of goals scored by both home and away teams, multiplied by 100.

To investigate home advantage in the soccer leagues represented in this study a paired design was used whereby each match contributed two observations, one for the home team and one for the away team. A repeated measures regression analysis using log-link Generalised Estimating Equations (Liang & Zeger, 1986) in STATA 11 (2009) was used to estimate the mean number of goals scored by home and away teams. Repeated measures analysis is used when observations occur in pairs or groups and the outcome of interest is likely to be correlated within each group. In the present study the 'groups' were the individual matches and the 'observations' were the number of goals scored by each of the two opposing teams.

As the dependent variable (goals scored) is a discrete count Poisson errors were specified for the regression model. This modelling strategy has previously been used to investigate home advantage in terms of disciplinary sanctions issued by referees (Goumas, 2012). Home advantage (HA) was estimated by applying the regression coefficient (beta) for match location (0 = Away, 1 =Home) to the following equation:

HA = [exp(beta)/[exp(beta) + 1]] x 100

Home advantage was estimated for each continental league and for all leagues combined. To test for difference in home advantage between leagues interaction terms between match location and league were added to the regression model and a chi-square test of their joint effect carried out.

To investigate how home advantage may vary according to crowd size (and to control for crowd size when comparing home advantage between continents) an interaction term between match location and crowd size was added to the regression model described above. Linear, quadratic and logarithmic terms for crowd size were explored, with the natural logarithmic transformation providing the best fit in each league. The coefficient (beta) for the interaction term between match location and log-transformed crowd size was interpreted as the rate of change in home advantage per each 10% increase in crowd size using the following equation:

Rate of Change = exp (beta x ln(1.1))

Interaction terms between match location and season were added to the regression model to control for variation in home advantage over time. Season was coded as the year the finals took place.

When comparing home advantage across different levels of a predictor variable (in this case crowd size) the home/away balance in the data is usually lost and team ability becomes a potential confounder which needs to be controlled for (see Clarke & Norman, 1995). To measure team ability each team was given points based on the stage of competition it reached in each of the last five seasons. For example, if a league consisted of two qualifying rounds, a group stage, three knockout rounds and a final (seven rounds in total) then a team reaching the second qualifying round was given two points for that season, whereas a team reaching the final was given the maximum seven points. The number of points was then summed across seasons. If a team did not participate in a given season then it was given zero points for that season. As the number of rounds played varied between leagues, team ability was standardised within each league to a mean of zero and standard deviation of one to allow for comparison between leagues (range: -1.0 to 3.9). A linear term for team ability was added to the regression model.

Crowd size analyses were conducted separately for each continental league and for all leagues combined; in the combined analysis interaction terms between match location and league were included in the model to control for potential confounding. To test for difference between leagues in the effect of crowd size on home advantage three-way interaction terms between match location, crowd size and league were added to the model and a chi-square test of their joint effect carried out.

P-values for statistical significance of individual regression parameter are from two-sided z-tests. P-values less than 0.05 were considered to be significant.


Thirty five (1.8%) matches with missing crowd size data were excluded from the analysis. All results are based on the remaining 1,900 matches. Table 2 shows home advantage, with and without adjustment for crowd size, and median crowd size in the league representing each continent. Unadjusted home advantage varied somewhat but non-significantly ([x.sup.2.sub.3] = 6.5; p = 0.09) between continents, ranging from 58.4% in Europe to 62.8% in North America. After adjusting for crowd size, however, there were highly significant ([x.sup.2.sub.3] = 27; p < 0.001) differences, with home advantage dropping to 56.2% in Europe and increasing to 67.1% in North America. There did not appear to be any relationship between median crowd size and home advantage across continents.

Table 3 presents results from the regression analysis of home advantage on crowd size. Home advantage in all continents combined increased by 1.5% per each 10% increase in crowd size (p < 0.001). Such a logarithmic association means that although home advantage increases with increasing crowd size the rate of increase slows as crowd size increases. Significant (p < 0.05) logarithmic associations were also observed within each continent. The rate of increase in home advantage per 10% increase in crowd size in North America (2.5%) was about twice that in each of the other continents (1.2% to 1.3%), although there was no statistical evidence of overall variation in the crowd size effect between continents ([x.sup.2.sub.3] = 3.1; p = 0.38). Figure 1 plots home advantage by crowd size in each continent, based on the results of the regression analysis; trend curves cover the middle 95% of crowd size values in each continental league. Home advantage reached 63% in Europe at crowd sizes of 77,000, 67% in Asia and South America at crowds of 42,000 and 53,000, respectively, and 73% in North America at crowd sizes of 25,000. At attendance levels of less than one thousand there was no home advantage in any continent. Although appreciable levels of home advantage (> 55%) were observed in North America, South America and Asia once crowds reached about two thousand, such levels were not evident in Europe until attendance reached about ten thousand.


The aims of this study were to determine the level of home advantage in major international club soccer competitions representing four continents, and to investigate the role crowd size plays in home advantage and how its effect may vary between continents.

Home advantage

Combined home advantage across the four continental leagues in this study (60.4%) was close to the worldwide figure for domestic leagues (61.5%; Pollard, 2006a). Unadjusted home advantage varied non-significantly between continents, ranging from about 58% to 63%; these figures are similar to those previously reported for domestic leagues in the same regions (Pollard, 2006a). It should be noted, however, that home advantage in the present study was based on goals scored whereas those in domestic leagues are usually based on competition points gained; comparisons therefore need to be made with caution. Also, the home advantage estimate for all continents combined is weighted heavily in favor of Europe, which comprised almost half the matches in the study.

Although there appeared to be little variation in home advantage between continents, adjusting for differences in crowd size revealed highly significant variation, with home advantage ranging from 56% in Europe to 67% in North America.

Crowd size and home advantage

In the combined analysis of all continental leagues there was strong evidence of a positive association between home advantage and crowd size. The crowd size effect was logarithmic in that the rate of increase in home advantage slowed as crowd size increased. Similar trends were observed within each continent although the level of home advantage at any given crowd size varied considerably. For example, in North America home advantage reached 73% at attendances of just 25,000, whereas in Europe home advantage reached only 63% for crowds three times that size. The effect of crowd size on home advantage appeared to be much stronger in North America than in other continents, although there was no statistical evidence of overall variation in the crowd size effect between continents. A post-hoe test did, however, show a significant difference ([x.sup.2] = 4.9; p = 0.03) between North America and the other three continents combined in the association between home advantage and crowd size.

These results show that the apparent similarity in the magnitude of home advantage previously observed worldwide (Pollard, 2006a) may have been confounded by differences in crowd size. In the present study controlling for crowd size revealed highly significant differences in home advantage between continents, with the range increasing more than two-fold. This occurred because the continent with the highest unadjusted home advantage (North America) also had the lowest average crowd size, whereas the continent with the lowest unadjusted home advantage (Europe) had the highest average crowd size. Controlling for crowd size, which was positively associated with home advantage, therefore had the effect of increasing the difference in the estimated levels of home advantage between these two continents.

The logarithmic nature of the association between home advantage and crowd size observed in the present study supports the results of previous comparisons of home advantage and average crowd size between different soccer leagues. Figure 2 plots the results of studies of eight soccer leagues in England and Scotland in 1992/93 (Nevill et. al, 1996) and nine soccer divisions in England in 1996/97 to 2001/02 (Pollard, 2006b). A logarithmic curve provides a reasonably good fit to both sets of observed data, with most of the increase in home advantage occurring at crowd sizes of less than ten thousand.

The results of this study suggest that although crowd size plays an important role in home advantage there are other factors also involved. Previous studies (Pollard, 2006a; Pollard & Gomez, 2009) have shown that home advantage varies greatly between regions differing in culture and history, being higher in isolated and ethnically distinct areas where the intensity of home crowd support may be greater. This may partly explain the relatively low home advantage observed in Europe--a more unified continent (with the exception of parts of eastern Europe) than the other continents represented in this study.

For sensitivity purposes the main analyses were repeated including the group stage of competition only (1,171 matches), as this stage is qualitatively different from the qualifying and finals stages in that group stage matches are decided individually, whereas almost all matches in the other stages are decided over two legs. Unadjusted home advantage in group stage matches differed from that for all stages combined by no more than an absolute 2% in each continent, and adjusting for crowd size had the same effect in the group stage as for all stages, with home advantage increasing in North America and decreasing in Europe. The significant logarithmic association between home advantage and crowd size in each continent was maintained when restricting the analysis to group stage matches, with the strongest association still being observed in North America. These results show that the main findings in this study are internally consistent and not dependant on how matches are decided.


A potential confounder of the association between crowd size and home advantage that this study was unable to control for was stage of competition (qualifying, group, and finals). Adjusting for this covariate caused instability in the regression model for Europe, possibly due to it being highly correlated with crowd size. Adding stage of competition to the models for Asia and North America had a negligible effect on the crowd size coefficients, whereas for South America the crowd size coefficient shifted a little away from the null. Therefore, any bias in the estimated effect of crowd size on home advantage caused by the exclusion of stage of competition from the model should have been minimal and/or towards the null. The similar pattern of results observed when restricting the analysis to group stage matches provides further evidence that stage of competition was not exerting a confounding effect.

Crowd factors other than absolute size, such as the ratio of crowd size to stadium capacity (crowd density) and how close the crowd is to the playing field (crowd proximity) may also play a role in home advantage in soccer, but were not able to be investigated in this study as such data were not available. A study of Greek football (Armatas & Pollard, 2012) found that home advantage was less in stadiums surrounded by a running track, where the crowd is further from the field of play. Although previous studies (Boyco et al., 2007; Pollard, 1986) have failed to show an association between crowd density and home advantage in terms of match outcome, home advantage in terms of referee decision making was found to increase with increasing crowd density (Boyco et al., 2007; Garicano, Palacios-Huerta & Prendergast, 2005; Goumas, 2012). The association between crowd size and home advantage may, therefore, be influenced by other crowd factors such as density and proximity.


This study shows that: (1) home advantage in soccer varies considerably worldwide when taking into account differences in crowd size; and (2) home advantage increases logarithmically with increasing crowd size and this trend is consistent across continents. Although crowd size plays a significant role in home advantage in soccer, other aspects of crowd support relating to geographic and cultural factors may also be important. Future research into home advantage could explore the effect of other crowd factors such as density and proximity.


Armatas, V., & Pollard, R. (2012). Home advantage in Greek football. European Journal of Sport Science, advance online publication.

Boyko, R. H., Boyko, A. R., & Boyko, M. G. (2007). Referee bias contributes to home ad-vantage in English Premiership football. Journal of Sports Sciences, 25, 1185-1194.

Buraimo, B., Forrest, D., & Simmons, R. (2010). The 12th man?: Refereeing bias in English and German soccer. Journal of the Royal Statistical Society (A), 173, 431-449.

Clarke, S. R., & Norman, J. M. (1995). Home ground advantage of individual clubs in English soccer. The Statistician, 44, 509-521.

Courneya, K. S., & Carron, A. V. (1992). The home advantage in sport competitions: A literature review. Journal of Sport and Exercise Psychology, 14, 28-39.

Garicano, L., Palacios-Huerta, I., & Prendergast, C. (2005). Favoritism under social pressure. The Review of Econimics and Statistics, 87, 208-216.

Goumas, C. (2012). Home advantage and referee bias in European football. European Journal of Sport Science, advance online publication.

Liang, K. Y., & Zeger, S.L. (1986). Longitudinal data analysis using generalized linear models. Biometrika, 73, 13-22.

Nevill, A. M., Newell, S. M., & Gale, S. (1996). Factors associated with home advantage in English and Scottish soccer matches. Journal of Sports Sciences, 14, 181-186.

Nevill, A. M., & Holder, R. L. (1999). Home advantage in sport: An overview of studies on the advantage of playing at home. Sports Medicine, 28, 221-236.

Pollard, R. (1986). Home advantage in soccer: A retrospective analysis. Journal of Sports Sciences, 4, 237-248.

Pollard, R. (2006a). Worldwide regional variations in home advantage in association football. Journal of Sports Sciences, 24, 231-240.

Pollard, R. (2006b). Home advantage in soccer: Variations in its magnitude and a literature review of the inter-related factors associated with its existence. Journal of Sport Behavior, 29, 169-89.

Pollard, R. (2008). Home advantage in football: A current review of an unsolved puzzle. Open Sports Sciences Journal, 1, 12-14.

Pollard, R., & Gomez, M. A. (2009). Home advantage in football in South-West Europe: Long-term trends, regional variation, and team differences. European Journal of Sport Sciences, 9, 341-352.

Pollard, R., & Pollard G. (2005). Long-term trends in home advantage in professional team sports in North America and England (1876-2003). Journal of Sports Sciences, 23, 337-50.

STATA 11 (2009). StataCorp, USA.

Address correspondence to: Chris Goumas, School of Public Health, The University of Sydney, NSW 2006, Australia. E-mail:

Chris Goumas

The University of Sydney

Table 1
Summary of data from the four international club soccer
competitions represented in this study

Continent       Confederation   ComGietition        Seasons

Europe          UEFA            Champions League    2007/08-2010/11
Asia            AFC             Champions League    2008-2011
North America   CONCACAF        Champions League    2008/09-2010/11
South America   CONMEBOL        Copa Libertadores   2009 -2011

Continent       nations   Teams (a)   Matches

Europe          52        163         848
Asia            13        69          444
North America   10        44          233
South America   11        74          410
Total           86        350         1,935

Note. UEFA = Union of European Football Associations; AFC = Asian
Football Confederation;

CONCACAF = Confederation of North, Central American and Caribbean
Association Football;

CONMEBOL = South American Football Confederation.

(a) Total number of teams participating over the period of study.

Table 2
Home advantage (%) and median crowd size in four continents (SE
in brackets)

                                               Home advantage (a)

Continent        Matches   Median                      Adjusted for
                           crowd size   Unadjusted       crowd size

Europe           828       28,674       58.4   (1.1)   56.2   (1.1)
Asia             444       8,344        61.0   (1.4)   63.5   (1.4)
North America    231       5,745        62.8   (1.9)   67.1   (1.8)
South America    397       14,828       62.5   (1.5)   62.9   (1.6)
All continents   1.900     14.064       60.4   10.71       --

(a) Estimated from Poisson regression models

Table 3 Results of regression analysis of home advantage on crowd
size in four continents

                 Regression statistics for crowd size (SE in brackets)

Continent        Coefficient (a)   z-score   p-value   Rate of change
Europe           0.133   (0.038)   3.49      <0.001    1.013   (0.004)
Asia             0.140   (0.053)   2.62      0.009     1.013   (0.005)
North America    0.262   (0.070)   3.73      <0.001    1.025   (0.007)
South America    0.121   (0.056)   2.16      0.031     1.012   (0.005)
All continents   0.154   (0.025)   6.16      <0.001    1.015   (0.002)

(a) Regression coefficient or interaction term between match
location anie log-transormed crowd size; estimated from log-link
Poisson regression models adjusted for team ability and season.
Coefficient for all continents combined is also adjusted for

(b) Rate of change in home advantage per 10% increase in crowd
COPYRIGHT 2013 University of South Alabama
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Goumas, Chris
Publication:Journal of Sport Behavior
Geographic Code:1USA
Date:Dec 1, 2013
Previous Article:Bullying experiences of individuals with visual impairment: the mitigating role of sport participation.
Next Article:Student athlete perceptions of academic success and athlete stereotypes on campus.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters