Team quality and game location effects in English professional soccer.

Bray, Steven R. ; Law, Jon ; Foyle, Jesse 等

The home advantage has been examined in sport contests for over 20 years. Results have consistently demonstrated that there is a performance advantage associated with competing at home among major professional and collegiate leagues, individual teams, and individual sport athletes. More than a decade ago, Courneya and Carron (1992) synthesized the results of 16 published and unpublished studies representing over 260 seasons of competition and concluded the home advantage phenomenon was robust and varied significantly in magnitude from sport to sport. For example, average home winning percentages varied from a low of 53.3% (effect size = .07) in baseball to a high of 69% in soccer (effect size = .38). More recently, Nevill and Holder (1999) extended Courneya and Carron's review to include 23 studies showing similar findings. Thus, in terms of describing the home advantage phenomenon, there seems to be ample evidence to support its existence.

Researchers have also attempted to explain why the home advantage exists by examining a number of factors that could be associated with game location. Summaries of research on explanatory factors related to the home advantage have also been provided by Courneya and Carron (1992) and Nevill and Holder (1999). In addition to providing an extensive integration of home advantage research, Courneya and Carron's review also introduced a conceptual model consisting of various factors that could influence the home advantage. Factors identified by Courneya and Carron are specific characteristics associated with game location as well as the psychological and behavioral states of athletes, coaches, and officials. Game location factors include the crowd, travel, familiarity with the venue, and rules that might favour the home team (e.g., batting last in baseball). Among these potential explanations for the home advantage, the influence of the crowd on home team performance has been the most studied (Agnew & Carron, 1994; Nevill & Holder, 1999).

Recent research has also shown that athletes' psychological states can vary depending on game location. Terry, Walrond, and Carron (1998) and Bray, Jones, and Owen (2002) found that athletes had lower state anxiety and higher self-confidence before home games compared to away. Terry et al. also found that athletes reported more positive mood profiles prior to playing at home.

Although several factors have been shown to play a role in explaining the home advantage, Courneya and Carron (1992) suggested that research should also focus on factors that moderate the game location-game outcome/performance relationship. One factor that has been shown to affect the nature and extent of home advantage is team quality, in their seminal work on home advantage Schwartz and Barsky (1977) showed that evenly-matched teams had similar home winning percentages of 50% with tie games included. However, the magnitude of home advantage was affected when teams played opponents of higher or lower quality. For example, in professional ice hockey when lower quality home teams played superior visitors, home advantage dropped to 37% and when higher quality home teams played inferior visitors, home advantage increased to 74%.

Snyder and Purdy (1985) and Madrigal and James (1999) obtained results consistent with those of Schwartz and Barsky (1977) in collegiate basketball indicating that team quality and game location interact. Madrigal and James found that the high quality teams in their sample won 70% of their home games played against other high quality teams and 95% of their home games against low quality teams. However, these results are limited in that they are confined to an examination of home advantage using only winning percentages of the home team as the critical statistic. Specifically, although the home winning percentage measure does provide some evidence of the interaction of team quality and game location, for each team it puts into perspective only half the games the team plays (e.g., in a balanced home-away schedule) and takes out of the equation its record of play when it competes away from home. This limitation was highlighted in a recent study carried out by Bray (1999) who operationalized home advantage in terms of the differential between each team's home and away winning percentages. Findings from that study showed that National Hockey League teams won an average of 17.3% more of their games played at home compared to games played away, but that the effect was not consistent across teams. Indeed, while most teams experienced a home advantage, several teams actually had a home disadvantage in some seasons.

Bray's (1999) findings offer an alternative perspective on the home advantage (i.e., from individual teams); however, the results are also clearly relevant to the discussion of the team quality--game location interaction as they showed that the magnitude of the homeaway differential was similar regardless of team quality. Specifically, high quality teams won 18.4% more games at home than on the road, while medium and low quality teams won 17.5% and 15.1% more games at home, respectively. In other words, the findings allow for an alternative hypothesis regarding the interaction of game location and team quality. That is, while research shows the performance of home teams appears to be affected by the extent to which their opponents are mismatched, generally the game location effect is similar regardless of how good they are. The answer depends on whether one considers all the games that are played by a team or only their home games.

While the aforementioned results open up a broader interpretation of the potential for game location to impact on team performance, there is an omission in those analyses which raises a concern regarding Bray's (1999) data. Bray's analysis operationalized home advantage as a winning percentage differential; it did not include tied games. While tied games are relatively uncommon in professional ice hockey (i.e., only 13.6% of games in Bray's 20-year sample were tied), ties are a relatively common outcome and deserve consideration in analyses of home advantage in many other sports. For example, in English professional soccer (i.e., Football League), drawn matches are a regular occurrence, representing as many as one-third of all regular-season match outcomes (Smailes, 2000). Thus, considering that the magnitude of home advantage varies by sport and the frequency of drawn matches is high in some sports compared to others, investigation of the role of draws in relation to game location is clearly needed.

Therefore, the general purpose of the present study was to extend Bray's (1999) analysis of the home-away winning differential to English professional soccer and to examine game location and team quality in relation to team winning percentages as well as tied games. The extension of Bray's (1999) analyses of the home advantage to English soccer represented a minor aspect of the current study. In fact, analyses of individual team statistics in English soccer have been reported by Clarke and Norman 0995) and Clarke 0996). Clarke and Norman showed that during the 10-year period of 1980/1981-1990/1991, teams in the four divisions of the English Football League generally won 24% more of their home games compared to away. However, while the effect was consistent across divisions, the magnitude of the advantage was highly variable from year to year and from team to team. The current investigation builds on those initial findings by utilizing a larger sample consisting of match results from 1981/1982-1999/2000, nonetheless, it was anticipated that a similar pattern of results as those found by Clarke and Norman would emerge.

The major objective of the study was to examine the potential interaction of game location and team quality on performance outcome with regards to home and away winning and draw percentages. Based on the findings of Bray (1999), it was hypothesized that soccer teams would show a similar home-away winning percentage differential regardless of team quality (i.e., no interaction).

Although tied games have been examined as a performance outcome in many studies of the home advantage, they are normally lumped in with win/loss statistics to gain an overall picture of the game location effect (cf. Courneya & Carron, 1992). This analysis strategy is reasonable, when one considers that from the perspective of the entire league's results, there can be no such thing as a home advantage when it comes to draws; it is a statistical inevitability that there will always be the same number of home draws as away draws within the league. This fact is also apparent from the overall perspective of individual teams in the league when all games are considered. However, one issue that should not be overlooked is that from an individual team perspective, it is clearly possible for any one team to have (a) a greater percentage of home draws compared to away draws, (b) an equal percentage of draws both home and away, or (c) a greater percentage of draws away than at home. Furthermore, the draw outcome may not be value-neutral and in fact may have dramatically different meaning depending on the quality of the team as well as where the draw occurs. For example, high quality teams that win a large percentage of their home games night consider a draw at home a poor result. On the other hand, for low quality teams that seldom win at home or away, a draw should certainly represent a positive result regardless of game location.

Hypotheses regarding the home-away draw percentage differential across varying team qualities were necessarily conservative. One limiting factor was the fact that this issue has not been addressed in previous research, therefore no previous findings could be used as reference points. Furthermore, (as noted above) the value of a drawn match may differ depending on team quality. For high quality teams, a home draw may represent a poor result, therefore, it could seem reasonable to hypothesize that high quality teams should draw a greater percentage of their away games than home games. However, because a home draw is still valuable in that the team earns a point result, high quality teams might be expected to draw a greater percentage of their home games compared to away. Thus, no hypotheses regarding the effect of game location on draws for high quality teams were advanced. For low quality teams that seldom win at home or away, a draw should represent a positive result regardless of game location. Consequently, low quality teams were expected to draw a greater percentage of their games when playing at home compared to away.

Method

Data

Archival data for the four divisions comprising the English Football League were obtained from the Breedon Book of Football Records (Smailes, 2000) and a total of 19 seasons (1981/1982-1999/2000) of team results were compiled representing over 77,000 matches. The sample of seasons was selected because of a change in points scoring (i.e., from 2 points for a win, I point for a draw, and 0 points for a loss to 3 points for a win, 1 point for a draw, and 0 points for a loss) was introduced in the 1981-1982 season.

Measures

Team season. Team season was the unit of analysis. One team season represented one season of regular-season competition for each team. There were a total of 1748 team seasons in the dataset.

Team quality. Team quality was assessed in order to examine hypotheses regarding the interaction of this variable with game location. Three groups were created within each division, based on the percentage of available points teams had earned in each season. High quality teams were those whose earned point percentage record was greater than one standard deviation above the sample mean of their division for the 19-year period. Low quality teams were those whose earned point percentage record was less than one standard deviation below that of their division's mean. Teams were classified as average quality if their earned point percentage record was within one standard deviation of the sample means of their respective divisions. The resulting sample split yielded 282 high quality, 246 low quality, and 1220 average team seasons.

Home winning percentage minus away winning percentage differential (H/AD). Consistent with the formula described by Bray (1999), H/AD was computed by subtracting each individual team's away winning percentage (i.e., # of away games won / # of away games played X 100) from its home winning percentage (i.e., # of home games won / # of home games played X 100) for each season.

Home draw percentage minus away draw percentage differential (H/ADD). In order to account for the results of drawn matches, H/ADD was computed by subtracting each individual team's away draw percentage (i.e., # of away games drawn / # of away games played X 100) from its home draw percentage (i.e., # of home games drawn / # of away games played X 100) for each season.

Results

Descriptive statistics for home and away winning and draw percentages are presented in Table 1. The average game winning percentage for the sample was 36 %, while the average percentage of drawn matches was 27 %. Because the sample was comprised of four independent divisions, we examined the data for possible variation from one division to another. A one way MANOVA showed that there were no significant differences in the study measures across the four divisions, Wilks' lambda = .994, F(12, 4606.55) = 0.82, p = .63, /.z = 0.00. Consequently, the data from all four divisions were pooled for further analyses.

Consistent with the first hypothesis, results of a repeated measures ANOVA showed that on average teams won a greater percentage of their home games compared to away games, F(1,1747) = 4176.84, p < .001, with a large associated effect size (At = 0.71). As expected, results also showed that overall, teams drew the same percentage of games both at home and away, F(1,1747) = 0.02, p = 0.89, [micro] = 0.00.

As pointed out above, the major focus of the analyses was on the team quality data. As shown in Table 1, high quality teams won a greater percentage of both their home (66%) and away games (43%) than average or lower quality teams. Results of the H/AD analysis, computed as a 3 (Team Quality) X 2 (Game Location) ANOVA with repeated measures on the latter factor, showed significant main effects for both team quality and game location as well as a significant interaction (See Table 2). For ease of interpretation, these results are depicted in Figure 1. Results clearly show a dramatic advantage in winning percentage at home versus away regardless of team quality, however, the H/AD for the low quality teams was not as pronounced (l 6.30 %) as for high (24.19 %) and average (22.81%) quality teams.

[FIGURE 1 OMITTED]

Analyses of the drawn (H/ADD) match percentages were also carried out using a 3 (Team Quality) X 2 (Game Location) ANOVA with repeated measures on the game location factor. Results showed a main effect for team quality and no main effect for game location. More importantly; however, there was also a significant Team Quality X Game Location interaction (See Table 3). These results are presented graphically in Figure 2, where it is evident that average teams drew virtually the same percentage of games both at home and away. However, high quality teams drew 6.58 %fewer home games compared to away games While the reverse was true for low quality teams who drew 5.80 % more games when they played at home.

[FIGURE 2 OMITTED]

Discussion

The purpose of the present study was to examine the home-away winning differential in English professional soccer and investigate game location and team quality in relation to winning percentages as well as tied games. Based on the findings of previous research by Bray (1999) and Clarke and Norman 0995), it was hypothesized that professional soccer teams would have a home advantage when their home and away performances were compared. Insofar as that hypothesis was concerned, the results were supportive and clearly showed a home winning percentage differential of greater than 20% for individual teams. As expected, the home--away draw differential showed no significant advantage in terms of playing at home. Furthermore, these effects were consistent across the four divisions comprising professional soccer in England over a recent 19-year span.

The major objective of the study focused on the issue of team quality as it related to game location effects. In line with the findings of Bray (1999), it was expected that there would be no differences in the home advantage (as defined by the H/AD) across high, medium, and low quality teams. While the results showed a similar trend of home--away differences, there was a clear main effect for team quality indicating better quality teams won a greater percentage of matches both at home and away than those of lesser quality. In addition, there was a significant interaction indicating that low quality teams have a slightly smaller H/AD than medium or high quality teams. Interestingly, these results are similar to those of Bray (1999), which showed a non-significant trend of lower quality teams having a smaller H/AD than medium or higher quality teams. However, Bray's sample was considerably smaller than the one examined in the present study (i.e., 409 vs. 1748 team seasons) and while sufficiently powerful (Cohen, 1992), was less likely than the current larger sample to show a significant effect. Considering the small effect size associated with the Team Quality X Game Location interaction in the present study, the results may be seen as consistent with those of Bray's earlier findings. Taken together, the combined results show that when the home advantage is operationalized as a home-away winning percentage differential, there is very little difference in the magnitude of the home advantage effect across team qualities.

Perhaps the most intriguing result from the present study was the significant Game Location X Team Quality interaction found when examining the home-away draw percentage differential. Those results indicated that for average quality teams, virtually the same percentage of games were drawn at home as on the road. However, low quality teams drew a greater percentage of their home games compared to away and high quality teams drew a smaller percentage of their home games compared to away.

As pointed out above, any firm hypotheses about results pertaining to draws and team quality were difficult to justify. Nonetheless, the fact that lower quality teams drew more frequently at home supports the hypothesis that because a draw result is likely to be a positive outcome for low quality teams, the home advantage should manifest itself in a greater percentage of draws at home compared to away.

There was certainly no clear foundation for a hypothesis that higher quality teams should draw more frequently at home compared to away and indeed, the opposite effect was found. One important consideration when examining these results is the fact that because they win so frequently at home, high quality teams have only 35% of home matches in which to draw or lose as opposed to 59% of matches on the road (see Table 1). Thus, at first glance, it appears that high quality teams may have an increased scope of opportunity for drawing away compared to home. However, the bias apparent in these statistics does not necessarily represent a contextual constraint that limits the probability of the draw outcome occurring at home for high quality teams. Rather, the probability of a draw occurring at home or away may have much to do with differential approaches that high quality teams take to their home and away matches. For example, Birmingham City Football Club chairman David Gold recently reflected on his team's ascension from lower to higher quality status, stating: "We used to go away from home, aim for a point and try and nick three. Now we go for three points and settle for one if we have to" (p.34, The Guardian The Season 02/03, 12/08/02). What this quote serves to illustrate is that teams may view the value of a draw result quite differently depending on how good they are relative to the other teams in their league. For teams of higher ability, anything less than a win at home is a negative result; despite the single point gleaned from a draw, it is an outcome they strive to avoid. Because they may be more inclined to settle for a draw away, it makes sense that high quality teams achieve that outcome more often on the road. Future research should examine psychological and behavioural factors as they pertain to team strategies when playing at home and away.

Due to the fact that the results of the present study are based entirely on archival data from one sport, the extent to which findings can be generalized beyond the current data is limited. Therefore, we suggest that future research should examine the home/away differential for both wins and draws with regards to varying team qualities across other major sports such as baseball, North American football, and basketball. The archival nature of the data also limits our ability to make firm recommendations for coaches and practitioners in sport psychology. Nevertheless, coaches and applied sport psychologists should take note of the pattern of results shown in the data. For example, there was a consistently strong effect showing an advantage in winning performance at home compared to away for teams of all qualities; however, the effect was significantly less pronounced for low quality teams. Practitioners working with lower quality teams may need to educate the players on those teams as to the features of the home environment that could play to their advantage and try to exploit them fully.

From an applied perspective, practitioners should also consider the approaches teams may take towards their home and away matches. The David Gold quote presented above clearly identifies distinct strategic approaches taken by one team based on game location considerations. However, it appears that an important factor affecting their changes in strategic approach was that team's recent success. Because team performance has an impact on the team's beliefs in its collective capabilities, future research could focus on team's perceptions of their collective efficacy (Bandura, 1997) and how those beliefs affect both strategies and performance relative to game location. According to theory, efficacy beliefs can be determined by multiple sources including vicarious experiences and verbal persuasion as well as performance accomplishments. Therefore, nurturing a collective belief amongst players that their team is capable of winning on the road may play an important role in motivating players to invest higher levels of effort and persistence despite having to play on another team's home ground. Certainly, going into an away game looking for a win and settling for a draw must be seen as a strategic approach that reflect greater team confidence and should be more conducive to better away performance than going for a tie and just hoping for a win.

In summary, the results of the present study showed that winning percentages were consistently higher (i.e., by about 20%) when English professional soccer teams competed at home compared to away across teams of higher, average, and lower quality. However, there was a clear discrepancy in the percentages of home versus away drawn matches between higher and lower quality teams. Although numerous previous investigations that have examined game location effects on performance have excluded tie games from their analyses (cf. Courneya & Carron, 1992; Nevill & Holder, 1999), in English professional soccer it is apparent that the relevance of tied games to home advantage depends on the quality of the teams involved. The process of unearthing inconsistencies in game location effects helps bring us closer towards understanding the complex phenomenon of the home advantage.

Address Correspondence To: Steven R. Bray, Department of Kinesiology, University of Lethbridge, Lethbridge, Alberta, Canada, TIK3M4. E-mail: steven.bray@uleth.ca.

Table 1.
Descriptive Statistics for the Study Measures

 All teams Average teams (a)
Variable M SD M SD

Total winning
percentage 36.41 10.89 35.42 (b,c) 6.63

Home winning
percentage 47.47 14.21 46.82 (b,c) 10.64

Away winning
percentage 25.35 11.72 24.01 (b,c) 8.78

Home/away
winning
percentage
differential 22.12 14.31 22.82 (c) 14.31

Total draw
percentage 27.09 6.68 27.66 (c) 6.73

Home draw
percentage 27.11 9.61 27.87 (b) 9.68

Away draw
percentage 27.07 9.45 27.45 (c) 9.24

Home/away
draw percentage
differential 0.05 13.61 0.42 (b,c) 13.30

 High quality teams (b) Low quality teams (c)
Variable M SD M SD

Total winning
percentage 53.69 (a,c) 5.45 21.57 (a,b) 4.74

Home winning
percentage 65.78 (a,c) 8.82 29.72 (a,b) 9.13

Away winning
percentage 41.59 (a,c) 9.06 13.42 (a,b) 6.77

Home/away
winning
percentage
differential 24.19 (c) 14.17 16.30 (a,b) 13.00

Total draw
percentage 25.59 (a) 5.89 25.99 (a) 6.89

Home draw
percentage 22.30 (a,c) 7.98 28.90 (b) 9.31

Away draw
percentage 28.89 (c) 9.43 23.09 (a,b) 9.51

Home/away
draw percentage
differential -6.58 (a,c) 12.90 5.80 (a,b) 12.81

Note: N = 1748 team seasons. For each variable, means scores
not sharing a common subscript differ at p<.05 (Tukey's HSD)
between average (n = 1220), high (n = 282), and low (n = 246) quality
teams.

Table 2
ANOVA Table for Differences Between Home and Away Winning Percentages
Across High, Average, and Low Quality Teams in Four Divisions of
English Soccer

Source df F

 Between subjects
Team quality 2 833.39
 Within subjects
Game location 1 2293.24
Team quality X game location 2 24.64
Error 1745

Source [micro] p

Team quality .68 .001

Game location .57 .001
Team quality X game location .03 .001
Error

Note: N = 1748 team seasons.

Table 3.
ANOVA Table for Differences Between Home and Away Draw
Percentages Across High, Average, and Low Quality
Teams in Four Divisions of English Soccer

Source df F [micro] p
 Between subjects
Team quality 2 15.46 .02 .001
 Within subjects
Game location 1 .09 .00 .76
Team quality X 2 59.25 .06 .001
game location
Error 1745

Note: N = 1748 team seasons.

Footnotes

(1) In English professional soccer, each team earns one point for a draw, a loss is worth zero points to the losing team, and win is worth three points to the winning team.

(2) The H/AD and H/ADD statistics were represented as percentages of the number of games played due to the fact that the number of games played at home and away over the course of a season was not always consistent within or across the four divisions.

References

Agnew, G. A., & Carron, A. V. (1994). Crowd effects and the home advantage. International Journal of Sport Psychology, 25, 53-62.

Bandura, A. (1997). Self-efficacy: The exercise of control. New York: W. H. Freeman.

Bray, S. R. (1999). The home advantage from an individual team perspective. Journal of Applied Sport Psychology, 11, 116-125.

Bray, S. R., Jones, M. V., & Owen, S. (2002). The influence of competition location on athletes' psychological states. Journal of Sport Behavior, 25, 231-242.

Clarke, S. R., (1996, October). Home advantage in balanced competitions: English soccer 1991-1996. Proceedings of the 3rd Conference on Mathematics and Computers in Sports. Queensland, Australia, 111-116.

Clarke, S. R., & Norman, J. M. 0995). Home ground advantage of individual clubs in English soccer. The Statistician, 44, 509-521.

Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155-159.

Courneya, K. S., & Carron, A. V. 0992). The home advantage in sport competitions: A literature review. Journal of Sport & Exercise Psychology, 14, 13-27.

Madrigal, R., & James, J. (1999). Team quality and the home advantage. Journal of Sport Behavior, 22, 381-398.

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.

Smailes, G. (2000). The Breedon book of football records, 1871-2000. Derby: Breedon Books.

Schwartz, B., & Barsky, S. F. (1977). The home advantage. Social Forces. 55, 641-661.

Snyder, E. E., & Purdy, D. A. (1985). The home advantage in collegiate basketball. Sociology of Sport Journal, 2, 352-356.

Terry, P. C., Walrond, N., & Carron, A. V. (1998). The influence of game location on athletes' psychological states. Journal of Science and Medicine in Sport 1(1), 29-37.

Steven R. Bray

University of Lethbridge

Jon Law and Jesse Foyle

University of Birmingham