Modelling home advantage in the Summer Olympic
* and A.M. WILLIAMS
Research Institute for Sport and Exercise Sciences, Liverpool John Moores University, 15–21 Webster Street, Liverpool
L3 2ET and
School of Sport, Performing Arts and Leisure, University of Wolverhampton, Walsall Campus, Gorway
Road, Walsall WS1 3BD, UK
Accepted 24 February 2003
Home advantage in team games is well proven and the inﬂuence of the crowd upon ofﬁcials’ decisions has been
identiﬁed as a plausible cause. The aim of this study was to assess the signiﬁcance of home advantage for ﬁve
event groups selected from the Summer Olympic Games between 1896 and 1996, and put home advantage in
team games in context with other sports. The ﬁve event groups were athletics and weightlifting (predominantly
objectively judged), boxing and gymnastics (predominantly subjectively judged) and team games (involving
subjective decisions). The proportion of points won was analysed as a binomial response variable using
generalized linear interactive modelling. Preliminary exploration of the data highlighted the need to control for
the proportion of competitors entered and to split the analysis pre- and post-war. Highly signiﬁcant home
advantage was found in event groups that were either subjectively judged or rely on subjective decisions. In
contrast, little or no home advantage (and even away advantage) was observed for the two objectively judged
groups. Ofﬁciating system was vital to both the existence and extent of home advantage. Our ﬁndings suggest
that crowd noise has a greater inﬂuence upon ofﬁcials’ decisions than players’ performances, as events with
greater ofﬁciating input enjoyed signiﬁcantly greater home advantage.
Keywords: crowd noise, ofﬁciating, subjective judging.
Baron Pierre de Courbertin proposed the revival of the
Olympic Games in 1892, with the modern Olympic
Games beginning in Athens in 1896. Since then, the
Olympic Games has grown in number of events,
number of competing nations, number of competitors
and number of home competitors (although not
proportionally to all competitors). There are now more
than 40 times more competitors than in 1896, 14 times
more nations competing and nearly seven times more
events (43 in Athens vs 300 in Sydney). Although such
growth has led to concerns about gigantism (Walle-
chinsky, 2000), the Olympic Games has evolved into
perhaps the world’s foremost sporting event, and
provides a rare opportunity to examine differences in
home advantage between sports.
The existence of home advantage in major team
sports has been well established (for a review, see
Nevill and Holder, 1999). Examples include baseball
(Adams and Kupper, 1994), football (Pollard, 1986;
Nevill et al ., 1996), ice hockey (Agnew and Carron,
1994) and basketball (Moore and Brylinsky, 1993).
Courneya and Carron (199 2) proposed four possible
factors that are thought to account for home
advantage: travel factors, learning/familiarity, rule
factors and crowd factors. The present study focuses
primarily upon crowd factors and speciﬁcally the
inﬂuence of ofﬁcials upon home advantage. For major
league baseball, Schwartz and Barsky (1977) identi-
ﬁed increasing home advantage with crowd density,
while also attempting to control for team quality.
Subsequently, Nevill et al. (1996) demonstrated
frequency of penalties and sendings-off to favour the
home side, this discrepa ncy increasing with crowd
size. The authors concluded that the crowd might
either inﬂuence away players to play more recklessly
or affect the match ofﬁcials’ decisions in favour of the
home side. There is now growing experimental
support for the latter hypothesis, with Nevill et al.
(1999, 2002a) showing the pr esence of crowd noise
to result in an imbalance of refereeing decisions in
favour of the home side.
* Author to whom all correspondence should be addressed.
Journal of Sports Sciences, 2003, 21, 469–478
Journal of Sports Sciences ISSN 0264-0414 print/ISSN 1466-447X online
2003 Taylor & Francis Ltd
The prevalence of home advantage in both individual
sport and unbalanced competition is less clear. Some
evidence of home advantage has been identiﬁed in cross-
country running (McCutcheon, 1984), wrestling (Gay-
ton and Langevin, 1992) and World Cup alpine skiing
(Bray and Carron, 1993). In contrast, once quality of
athlete had been accounted for, home advantage was not
found to be a major inﬂuence on performance in
individual ‘grand slam’ tennis or ‘major’ golf tourna-
ments (Nevill et al., 1997). Holder and Nevill (1997)
conﬁrmed these ﬁndings, suggesting that any apparent
home advantage is mainly an artefact of selection
procedures that favour increased entry of lower ranked
home competitors. The authors suggested that lack of
home advantage could be due to both objective scoring
systems and relatively little subjective input from
ofﬁcials. Multi-event com petition such as the Olympic
Games allows examination and comparison of home
advantage between team and individual events, as well
as between events relying on differing scoring systems.
Previous research on home advantage in the Winter
Olympic Games (Balmer et al., 2001) has shown home
advantage to vary between events, with a signiﬁcantly
greater advantage for disciplines in which ofﬁcials
directly judge outcome. It is suggested that ofﬁcials in
such sports (e.g. ﬁgure skating, freestyle skiing) may be
scoring home competitors disproportionately highly.
There is extensive evidence of ofﬁciating bias in
subjectively judged sports. Most of this evidence relates
to speciﬁc nationalistic or political bias, rather than a
more general home advantage. Nationalistic and/or
political bias has been demonstrated for a range of
subjectively judged events, including Olympic diving
(Park and Werthner, 1977), ﬁgure skating (Seltzer and
Glass, 1991) and gymnastics (An sorge and Scheer,
1988; Whissell et al., 1993). With the exception of
Balmer et al. (2001), who provided some evidence, it is
unclear whether the ofﬁciating bias observed in
subjectively judged sports extends to home advantage.
Several recent studies have examined performance in
the Olympic Games (e.g. Lozano et al., 2002; Morton,
2002; Nevill and Stead, 2003). All of these studies
examined performance at a nation level, reclassifying
success while controlling for gross national product and
population size (Lozano et al ., 2002), gross domestic
product and population size (Morton, 2002) and gross
national produc t alone (Nevill and Stead, 2003).
Adjusted indices of success produced dramatic changes
in ranking (compared with observed medal tables), with
Eastern European and island nations fairing particularly
well (Morton, 2002) and Cuba ﬁnishing top by a clear
margin (Morton, 2002; Nevill and Stead, 2003).
Although these three studies did not formally consider
home advantage, it was acknow ledged as a potential
inﬂuence (e.g. Morton, 2002), though evidently the
authors’ mai n conclusion (regarding Cuba’s perfo r-
mance) remains unaffected, since Cuba has never
hosted the Olympic Games. Clarke (2000) examined
home adva ntage in the Summer Olympic Games at a
nation level (across all events). Home nations were
shown to win approximately three times more medals in
home Olympics (compared with away) and, interest-
ingly, approximately two times more in Olympic Games
either side of their home Olympics. Clarke’s (2000)
primary aim was to assess Australia’s upcoming Sydney
2000 performance; although competitor participation
was discussed (as an inﬂuential factor in 1956) it was
not controlled for, as the paper was concerned with
predicting performance rather than explaining home
advantage. Indeed, Clarke (2000) was particularly
successful in this respect, as Australia’s predicted
success (of approximately 20 gold medals and 60
medals in total) was remarkably similar to their
observed medal tallies (16 gold medals and 58 medals
As in tennis and golf tournaments (Nevill et al.,
1997), team/competitor quality is also a major con-
sideration in the Olympic Games, given the highly
variable medal winning potential of host nations. In
Summer Olympic Games competition, for example, the
United States has won nearly 50 times more medals
than Mexico, even though both have been host nations.
Assessing home advantage must, therefore, involve only
a comparison of home and away performances of the
same nation or team. Previous research has highlighted
team quality as a likely inﬂuence, with ‘superior’ teams
tending to exhibit greater home advantage (Schwartz
and Barsky, 197 7), although some authors have made
reservations regarding the classiﬁcation of quality
(Madrigal and James, 1999). While earlier examples
concerned team games, in which an assessment of the
quality of home and away teams can be fairly easily
made, Olympic competition typically involves many
competitors with a diverse range of abilities, creating
some difﬁcul ty in formally assessing quality. Madrigal
and James (1999) proposed that a proportion of this
‘team quality effect’ may be due to superior teams
enjoying larger or denser and more supportive audi-
ences, although this is not the case for Summer
Olympic Games, as success is not related to large home
crowds. The year of the Olympic Games is more
signiﬁcant for spectator att endance, as numbers simply
increase over time. For Olympic competition, stronger
nations’ dominance in a given event would severely
reduce weaker host nations’ opportunities for home
advantage. Similarly, stronger nations, with more
competitors capable of winning medals, should be able
to produce more consistent home advantage. Evidently,
then, team quality must be considered to correctly
measure home advantage.
470 Balmer et al.
The aim of the present study was to compare home
advantage between groups of events with varying styles
of ofﬁciating. Realization of this aim would put the
consistent home advantage observed in team games in
context alongside both objectively and subjectively
judged individual competition. We hypothesized that
signiﬁcant home advantage would be observed in
subjectively judged events (gymnastics, boxing) in
which the potential for biased ofﬁciating is at its greatest
(Ansorge and Scheer, 1988). We also hypothesized that
signiﬁcant home advantage would be observed for team
games, in which the crowd may have an inﬂuence upon
subjective decisions made by ofﬁcials (Nevill et al.,
1999, 2002a). In contrast, we hypothesized that events
such as athletics and weightlifting, which have pre-
dominantly objective scoring and little input from
ofﬁcials, would sho w no home advantage.
Establishing unbiased home advantage
Inclusions and exclusions
Given the vast amount of Summer Olympic Ga mes
data, a subset of event groups was chosen, made up of
track and ﬁeld athletics, gymnastics, team games,
boxing and weightlifting. The rationale for the choice
of events is presented in the next subsection.
Assessment of home advantage concerns only
nations who have hosted an Olympic Games (hosting
nations) with intra-nation comparison (home vs away
by nation) being central to a fair assessment. The
inclusion of a large number of weaker non-hosting
nations as away data would simply lead to unrealis-
tically large home advantage. To produce the max-
imum number of hosting nations, only male data
were analysed, as females typically did not compete
until 1928 and, in many cases, much later (prohibit-
ing a thorough pre-Second World War analysis).
Having no data before 1928 alone exclu des all data
for Greece, France, Belgium and Holland, while the
omission of women’s hockey (ﬁrst played in 1980) or
basketball (ﬁrst played in 197 6) would result in
signiﬁcant loss of data.
Additionally, if a given hosting nation has neither
won home nor away in an event, at any Olympic
Games, they were exclu ded. This resulted in a
differing number of hosting nations between events,
both as a result of which Olympic Games each event
was held at and hosting nation performance. A single
medal won, home or away, qualiﬁes a hosting nation
for inclusion in the analysis, classifying them as a
successful hosting nat ion for a given event or group
Selection of events
Event groups were chosen on the basis of longevity (i.e.
how many Olympic Games events have been contested)
and, most importantly, to allow contrasts between
ofﬁciating style. Ofﬁcials have little or no input in
weightlifting or athletics. In contrast, they decide most
outcomes in boxing and all outcomes in gymnastics. In
team games, although the outcome is decided by goals,
points or baskets, ofﬁcials make many subjective
decisions. The following groupings were included:
1. Track and ﬁeld athletics were included as they form
the focal point of modern Olympic Games, as well
as a comprehensive and continuous source of data.
Athletics as a whole contributes 12 of the 16
individual events contested at all Olympic Games
(Greenberg, 2000). Most importantly, events have
objectively measurable outcomes and little subjec -
tive input from ofﬁcials. As a result of this, track
and ﬁeld athletics should have little or no home
advantage assuming our research hypothesis about
subjective decisions le ading to home advantage is
2. Weightlifting is also an almost entirely objective
discipline. Outcom e is generally decided by an
aggregate measure of weight lifted in two or three
disciplines, which have changed on several occa-
sions over time (notably the abolition of single-
handed lifts in 1928). Although judges adjudicate
on the success of each lift, it is assumed that their
input is minimal, with most decisions clear-cut.
Moreover, any ambiguity was further reduced when
the ‘press’ component was removed in 1976 owing
to judging difﬁculty (Greenberg, 2000). Given its
predominantly objective ofﬁciating, weightlifting
should again have little or no home advantage.
3. For gymnastics, judge s have a large subjective input
and directly assess performance. Ansorge and
Scheer (1988) claimed that the ‘effects of biased
ofﬁciating are potentially most dramatic in sports in
which the ofﬁcials actually score the points through
judging the performance of competitors with some
combination of objective and subjective criteria’ (p.
103). Gymnastics provides an example of such
artistic events and should exhibit highly signiﬁcant
4. Historically, boxing has generated the most con-
troversy of any Olympic sport, including such
connected events as attacks on ofﬁcials, sit-down
protests and full-scale riots. Much of this contro-
versy has focused upon the ﬁve ringside judges and
referee. Indeed, recent measures (since Barcelona
1992) have included banning ofﬁcials from cocktail
parties and the administration of daily alcohol tests
471Modelling home advantage in the Summer Olympic Games
to ensure they are ‘out of reach’ of inﬂuence from
national associations and their ofﬁcials (W allechins-
ky, 2000). Despite such concerns, little research has
addressed possible inﬂated home advantage in
boxing. Although a small proportion of boxing
matches are decided by knockout, most Olympic
bouts rely on the subjective assessment of judges
(85.95% of Olympic bouts; Lyberg, 1999c). Given
this subjective judging, boxing should display highly
signiﬁcant home advantage.
5. Olympic team games have an objective scoring
system (e.g. goals, baskets), although ofﬁcials
have substantial subjective input during matches.
Home advantage for major team games is well
known (Nevill and Holder, 1999), although it has
not been put into context alongside other major
individual sports. Given the proven inﬂuence of
crowd noise upon referees in an experimental
setting (Nevill et al., 1999, 2002a), team games
should be associated with signiﬁcant home ad-
vantage, although perhaps not as much as artistic
event groups, where judge s directly decide out-
Points and maximum points
Performance was measured using a simple points
system, with 3 points allocated for a gold medal, 2
points for silver and 1 point for bronze. A fair
estimation of performance requires not only points
scored, but also a consideration of the number of points
available. Points were considered as a proportion of
maximum points available, this maximum varying with
event. In most team games or events (e.g. football,
hockey, 46100 m) the maximum is 3 points (i.e. a
single gold), as only one team may be entered.
Elsewhere, the maximum is typically 6 points, as three
or more competitors of the same nationality could win
gold, silver and bronze (6 points) in a given event and
Both points and maximum points were combined for
each event group (deﬁned in the previous subsection).
Combined maxim um points also considered number of
competitors or teams, so they never exceeded the
number of points achievable by the number of
competitors or teams entered. The result is a sum of
points for all relevant events, for each Olympic Games
at which the events were contested. Event groups and
number of constituent events or weight categories are
presented in Table 1. Each observation is a proportion
for a given country, in a given event group and Olympic
Games (e.g. Germany won 11 of 48 boxing points at the
1936 Olympic Games). A nation was removed if it had
never contested the events or had never won a point
(home or away) in a given event group. All data were
obtained from the Olympic museum, Lausanne (Ly-
berg, 1999a,b,c,d,e,f,g,h,i). The results from Sydney
2000 were not included, since although medal winning
performance was avai lable, detailed partici pation ﬁg-
ures were only available up to 1996.
The response or dependent variable was the number of
points won by each nation as a ratio of the maximum
number of points available.
Home versus away
A binary home versus away indicator variable was
entered to allow assessment of the difference between
home and away performance.
Proportion of competitors or teams competing
The number of competitors or teams for each nation,
entered as a proportion of the total competitors or
teams, was included as a covariate likely to inﬂuence the
proportion of points won. Most pre-Second World War
Table 1. Event groups included in the analysis and number of constituent events
No. of events/
Observations in analysis (n)
No. of Olympic
Event group weight divisions Dates competed Pre-war Post-war Games
Track and field 25 1896–1996 71 144 23
Gymnastics 8 1896–1996 32 86 23
Weightlifting 10 1896, 1904, 1920–1996 25 115 20
Boxing 12 1904–1908, 1920–1996 29 142 20
Team games 5 1900–1996 44 103 22
Total 201 590
472 Balmer et al.
Olympic Games had few limits on the maximum
number of competitors, leading to inﬂated home team
sizes, or even additional teams in some early team
games. At the 1904 St. Louis Olympic Games, the host
nation entered two of a total of three competing football
teams (Wallechinsky, 2000). Even after the instigation
of maxima, many nations used the opportunity of a
home Olympic Games to reach the maximum when
otherwise further competitors at considerable cost
would not have been worthwhile. Evidently further
competitors at home would enhance the given host
nation’s ability to win medals (however slightly). To
reach an unbiased home advantage, a competitor
(competitor or team) covariate indicating the number
of competitors or teams entered by each nation in each
event group was included in the analysis.
This covariate was equal to the number of competi-
tors or teams divided by the total number for all nations
(i.e. the proport ion of competitors entered by each
‘hosting nation’, in each event or event group, at each
Olympic Games). Using this value rather than raw
number of competitors or teams accounts for changing
(generally growing) competitor or team participation
over time. This covariate will be referred to as
‘proportion of competitors’ in the analyses.
Evidently, stronger host nations win more points and
may enjoy greater home advantage (e.g. Schwartz and
Barsky, 1977; Madrigal and James, 1999). Considera-
tion of differences between host nations as a repeated-
measures ‘within-subject’ factor (‘host’) allows differ-
ences in nation quality to be evaluated and accounted
Differences over time exist both in the number of
competitors entered (especially home competitors) and
number of nations entering. Generally, after 1936,
restrictions were placed upon the number of competi-
tors entering, preventing vast numbers of home
competitors. An increased number of away nations
has also increased competition and may further reduce
home advantage. For this reason, separate analyses
were conducted pre- and post-war to allow for
differences in the proportion of competitors covariate.
The full rationale for this split is explained in the
Analysis section below.
Event group/ofﬁciating style
Consideration of the above variables allows both
accurate (unbiased) measurement of home advantage
in each event group and a comparison between groups
with different ofﬁciating style s. A ﬁve-category event
group factor was included in the analys is to highlight
differences in home advantage between groups (home/
away 6 event group interaction). Analysis was then
split by event group to examine absolute home
advantage for each group.
Two major medal-winning nations that have also hosted
Summer Olympic Games have seen considerable
changes in composition. First, Germany split into East
and West Germany after the Secon d World War, before
re-unifying in time for the 1992 Olympic Games.
Similarly, the Soviet Union collapsed before the 1992
Olympic Games (though a uniﬁed team competed in
1992). For simplicity, Germany is deﬁned as West
Germany between 1952 and 1988 (Federal Republic of
Germany from 1968 on wards), although it should be
noted that Germany did enter a combin ed team on
three occasions within this period (1956–1964). After
the collapse of the Soviet Union, Russia is deﬁned as
Russia and not former constituents of the Soviet Union.
We attempt to control for changes in the proportio n of
total athletes or teams entered (and subsequent success)
as a result of these changes (and of boycotts) by using
the proportion of competitors or teams as a covariate.
The proportion of medals won by each nation was
analysed using two separate methods. First, traditional
analysis of covariance (ANCOVA) was used, with
measurements relating to the proportion of competitors
entered as a covariate. The general linear model used
assumes normality. However, the response variables
consisting of proportions of points (of total points)
resulted in predictable departures from normality due
to many especially large or small proportions (Zar,
1998). An arcsine transformation (Winer, 1972) to
stabilize variances and produce a more acceptable
normal distribution failed to signiﬁcantly impro ve
deviations from normality in either the pre-war or
post-war analyses. Preliminary exploration of the data
also highlighted the need to split the analysis pre- and
post-war. This was mainly a result of post-war
competitor restrictions leading to a far less inﬂuential
‘proportion of competitors’ covariate, with a markedly
different slope (see Fig. 1).
Proportions of points won by each nation were
analysed as a binomial response variable (e.g. if a
nation won a silver medal in the 100 m athletics ﬁnal, 2
of 6 points would be allocated to that nation for that
Olympic Games) using generalized linear interactive
473Modelling home advantage in the Summer Olympic Games
modelling (GLIM; Aitkin et al., 1989; Nevill et al.,
2002b). Although proportion of points won is not
strictly a binomial proportion (as it is not generated
from repeated yes/no responses), it is a discrete
distribution with an upper bound (whi ch differed with
event, Olympic Games and number of competitors
entered). In such a case, the binomial distribution
provides the most accurate approximation (rather than
the Poisson distribution where there exists no upper
bound). Rather than assuming the response variable has
a linear function of the covariates with an approximate
normal error, GLIM is able to assess the effect of all the
explanatory variables on the proportion of points won
assuming the exact binomial error distribution (r points
from a possible n). Separate GLIM analyses were
performed on the pre- and post-war data because of
differing ‘proportion of competitors’ covariates. Ana-
lyses were then split by event group both pre- and post-
Fitting the covariate of ‘proportion of competitors’, the
main effects of ‘host’, ‘event group’ and ‘home vs away’
plus the interaction between ‘event group’ and ‘home vs
away’ term explained a loss in deviance of 71628.9 with
17 degrees of freedom (P 50.0001). When we at-
tempted to remove the interaction term, the covariate or
any of the main effects from the model, the increase in
deviance was too large in all cases (P 50.01). Conse-
quently, all terms were retained in the ﬁnal pre-war
model describing the proportion of points won by the
eight ‘hosting’ nations. The unadjusted and adjusted
proportions of points won (obtained by calculating the
mean intercepts for proportion of points won having
ﬁtted the covariate ‘proportion of competitors’) by
hosting nations pre-war are presented in Figs 2 and 3,
respectively. All error bars in the ﬁgures denote the
standard error of the estimate.
Five simpliﬁed models were then ﬁtted for each event
group, with the single ‘proportion of competitors’
covariate and main effects ‘host’ and ‘home vs away’.
This allowed absolute measurement of home advantage
for each group. Table 2 presents the direction (home
advantage/away advantage) and signiﬁcance (change in
scaled deviance as a result of removal from ﬁnal model)
of the facto r ‘home vs away’ for both pre- and post-war
Fig. 1. Proportion of competitors entered versus percentage
of points won, for each Olympic Games, nation and event
group. O, pre-war; 6, post-war.
Fig. 2. Percentage of points (+standard error of the esti-
mate) won by all successful hosting nations pre-war.
home; &, away.
Fig. 3. Adjusted percentage of points (+standard error of the
estimate) won by all successful hosting nations pre-war.
home; &, away.
474 Balmer et al.
With the factor ‘host’ and covariate ‘proportion of
competitors’ entered, no home advantage was found for
‘track and ﬁeld’ , ‘weightlifting’ or ‘boxing’. In contrast,
‘gymnastics’ yielded signiﬁcant home advantage
= 5.24, P = 0.022), as did ‘team games’
= 19.38, P 50.001). For each event group, as with
the global analysis, increase in deviance was too large to
remove either ‘host’ or ‘proportion of competitors’,
conﬁrming the need to retain them in each pre-war
Adopting the same methodological approach to analys-
ing the pre-war results, the covariate of ‘proportion of
competitors’, the main effects of ‘host’, ‘event group’
and ‘home vs away ’ were ﬁtted, plus the interaction
between ‘event group’ and ‘home vs away’, to explain
the proportion of points won by the 12 hosting nations.
These terms explained a loss in devia nce of 72657.1
with 21 degrees of freedom (P 50.0001). As with the
pre-war analysis, when we tried to remove the interac-
tion term, the covariate or any of the main effects from
the model, the increase in deviance was too large in all
cases (P 50.01). All terms were retained in the ﬁnal
post-war model describing the proportion of points won
by the 12 ‘hosting’ nations. The unadjusted and
adjusted (for proportion of competitors) proportions
of points won by hosting nations post-war are presented
in Figs 4 and 5, respectively.
Separate models were ﬁtted for each event group,
using an identical procedure to that of the pre-war
analysis. As for the pre-war analyses, Table 2 presents
the direction (home advantage/away advantage) and
signiﬁcance ( change in scaled deviance as a result of
removal from ﬁnal model) of the factor ‘home vs away’.
As with the pre-war analysis, ‘track and ﬁeld’ yielded
no home advantage. Interestingly, once ‘proportion of
competitors’ had been accounted for, ‘weightlifting’
Table 2. Extent and direction of home advantage for each of the ﬁve event groups pre-war (Pre) and post-war (Post)
Direction of advantage
Change in scaled
Degrees of freedom P-value
Event group Pre Post Pre Post Pre Post Pre Post
Track and field Away Home 0.22 0.20 1 1 0.64 0.66
Gymnastics Home Home 5.24 25.23 1 1 0.022 5 0.001
Weightlifting Home Away 1.46 6.21 1 1 0.23 0.013
Boxing Home Home 2.47 42.92 1 1 0.12 5 0.001
Team games Home Home 19.38 9.99 1 1 5 0.001 0.0015
Fig. 4. Percentage of points (+standard error of the esti-
mate) won by all successful hosting nations post-war.
home; &, away.
Fig. 5. Adjusted percentage of points (+standard error of the
estimate) won by all successful hosting nations post-war.
home; &, away.
475Modelling home advantage in the Summer Olympic Games
now exhibited signiﬁcant away advantage (w
P = 0.013). Meanwhile, a large signiﬁcant home ad-
vantage was observed for ‘gymnastics’ (w
P 50.001), ‘team games’ (w
= 9.99, P = 0.0015) and
= 42. 9, P 50.001). As with previous
analyses, increase in deviance was too large to remove
either ‘host’ or ‘proportion of competitors’, conﬁrming
their importance in the model.
The present study had two objectives. The ﬁrst was to
assess the signiﬁcance of home advantage in a subset of
Summer Olympic Gam es event groups, while identify-
ing and controlling for confounding factors. The
second was to examine differences in home advantage
between groups of events relying on differing ofﬁciating
styles. We hypothesized that sports requiring subjective
judgement (boxing, gymnastics) or subjective decisions
(team games) woul d yield highly signiﬁcant home
advantage. In contrast, little or no home advantage
was expected for sports for which ofﬁciating is less overt
or predominantly an objective process (track and ﬁeld,
In both the pre-war and post-war analyses, overall
home advantage was found to be signiﬁcant, high-
lighted by the large change in deviance when attempting
to remove the ‘home vs away’ main effect. The
‘proportion of competitors’ covariate was also found
to be highly inﬂuential both pre- and post-war,
indicating the importance of proportion of competitors
or teams to successful performance. Controlling for this
covariate proved central to a fair measure of home
advantage, highlighted by the marked difference be-
tween unadjusted (Figs 2 and 4) and adjusted
performance (Figs 3 and 5). A similar measure of
participation could prove valuable in the further
development of al lometric models of Olympic perfor-
mance (e.g. Lozano et al., 2002; Morton 2002; Nevill
and Stead, 2003).
Signiﬁcantly different slopes for the ‘proportion of
competitors’ covariate illustrated the need to split the
analysis pre- and post-war, as shown by Fig. 1.
Similarly, signiﬁcant differences were noted between
‘host’ nations’ point-winning performances. Entry of a
nation main effect accounted for a large proportion of
variance (due to substantial differences between na-
tions), conﬁrming ‘team quality’ concerns of previous
research (e.g. Schwartz and Barsky, 1977; Holder and
Nevill, 1997; Nevill et al., 1997; Madrigal and James,
1999). The success of different nations is also likely to
have been inﬂuenced by numerous boycotts. These
include Spain, Switzerland, Egypt, Iraq and Lebanon in
1956, various Africa n nations in 1976, the United
States in 1980 and the Soviet Union at the following
Olympic Games. These boycotts, particularly the ﬁnal
two, clearly enhanced the performance of host nations
and, therefore, increased home advantage. Fitting the
‘proportion of competitors’ covariate attempted to
control for such boycotts (by modelling increases in
proportion of competitors), alth ough it is unlikely that it
completely accounted for the quality of competitors
lost. When we assessed differences in home advantage
between event groups, the conclusions remained
unaffected. However, analysis at the level of nations
(e.g. attempting to predict a particular nation’s future
medal-winning performance; e.g. Clarke, 2000) would
require further consideration of boycotts and quality of
The most important ﬁnding was the signiﬁcance of
the home/away6event group interaction, both pre-
and post-war. Subsequent analysis for each individual
event group revealed the source of this signiﬁcant
interaction term. Pre-war, the signiﬁcant home/
away6group interaction term was a result of a small
(non-signiﬁcant) away advantage in the two objec-
tively judged groups, compared with signiﬁcantly
greater positive home advantage for gymnastics and
substantial home advantage for team games. Post-war,
when competitor participation had stabilized to some
extent due to restrictions, the home/away6gro up
interaction was a result of signiﬁcantly greater home
advantage in event groups that are either subjectively
judged (gymnastics, boxing) or rely on subjective
decisions (team games). Home advantage for these
three groups was signiﬁcantly greater than that of the
two objectively judged groups (track and ﬁeld,
weightlifting) once we had controlled for the propor-
tion of competitors/teams.
Objectively judged groups showed no home advan-
tage either pre- or post-war (Figs 3 and 5) and, in the
case of post-war weightlifting, even showed some
indication of away advantage. This may be explained
by nations which have never hosted the Olympic Ga mes
beginning to enter participants in weightlifting and
becoming increasingly strong as time progresses.
Notable examples include Bulgaria, Romania and
China, none of whom featured in men’s weight lifting
medals tables between 1948 and 1968. Bulgaria then
topped the table with six medals in 1972 and won six
again in 1976. Similarly, Romania and China occupied
the top two weightlifting table spots in 1984, winning
14 of 30 medals between them. An increasing number
of strong away nations may reduce home nations’
chances of winning medals compared with earlier away
Olympic Games in which such nations were not present
With respect to subjectively judged events, our results
conﬁrm previous Winter Olympic Gam es ﬁndings
476 Balmer et al.
(Balmer et al., 2001) that such disciplines enjoy
signiﬁcantly greater home advantage than events with
little ofﬁciating input. Evid ently, this ofﬁciating com-
ponent is vital for home advantage in individual sports.
This could explain why signiﬁcant home advantage has
been observed in wrestlin g (Gayton and Langevin,
1992), even at high school level, but not international
tennis or golf (Holder and Nevill, 1997; Nevill et al .,
1997). It would appear that the potential for biased
ofﬁciating in subjectively judged events predicted by
Ansorge and Scheer (1988) is conﬁrmed in terms of
As hypothesized, team games demonstrated highly
signiﬁcant home advantage both pre- and post-war,
although the size of this imbalance was surprisingly
large. Team games demonstrated by far the largest
home advantage pre-war and highly signiﬁ cant home
advantage post-war (Table 2). Pre-war, a lack of
competitive away teams and failure to complete ly
account for additional home teams (‘proportion of
competitors’) may partially explain this imbalance.
Post-war, following instigation of entry restrictions,
home advantage remains highly signiﬁcant, although
somewhat less so than for the subjectively judged
Previous research has highlighted crowd factors as a
dominant cause of home advantage. Crowds are able to
inﬂuence players and ofﬁcials to alter performance to
favour the home side or nation (Pollard 1986; Nevill et
al., 1996 ). Competitors in all of the event groups enjoy
consistently la rge and vocal crowds. If these crowds
were able to inﬂuence competitors’ performance, home
advantage would be observed for all event groups,
which was not the case. Far greater home advantage in
the three event groups with substantial ofﬁciating input
supports the latter hypothesis, that the crowd is able to
inﬂuence ofﬁcials to favour the home side. Experi-
mental research has provided support for a crowd
inﬂuence upon ofﬁcials in association football (Nevill et
al., 1999, 2002a). However, while an imba lance of
decisions was identiﬁed, this was not quantiﬁed in
terms of home advantage. The results of the present
study suggest that the imbalance observed with crowd
noise in football translates to a sizeable home advan-
tage, signiﬁcantly larger than that for objectively judged
events and comparable to that for subjectively judged
Adams, R.D. and Kupper, S.J. (1994). The effect of
expertise on peak performance: the case of home-ﬁeld
advantage. Journal of Sport Behavior, 17, 108–119.
Agnew, G.A. and Carron, A.V. (1994). Crowd effects and
the home advantage. International Journal of Sport Psychol-
ogy, 25, 53–62.
Aitkin, M., Anderson, D., Francis, B. and Hinde, J. (1989).
Statistical Modelling in GLIM. Oxford: Oxford University
Ansorge, C.J. and Scheer, J.K. (1988). International bias
detected in judging gymnastic competition at the 1984
Olympic games. Research Quarterly for Exercise and Sport,
Balmer, N.J., Nevill, A.M. and Williams, A.M. (2001).
Home advantage in the Winter Olympics (1908–1998).
Journal of Sports Sciences, 19, 129–139.
Bray, S.R. and Carron, A.V. (1993). The home advantage in
alpine skiing. Australian Journal of Science and Medicine in
Sport, 25, 76–81.
Clarke, S.R. (2000). Home advantage in the Olympic games.
In Proceedings of the Fifth Australian Conference on Mathe-
matics and Computers in Sport (edited by G. Cohen and T.
Langtry), pp. 43–51. Sydney, NSW: University of Tech-
Courneya, K.S. and Carron, A.V. (1992). The home
advantage in sport competitions: a literature review.
Journal of Sport and Exercise Psychology, 14, 13–27.
Gayton, W.F. and Langevin, G. (1992). Home advantage:
does it exist in individual sports? Perceptual and Motor
Skills, 47, 706.
Greenberg, S. (2000). Whitaker’s Olympic Almanack: An
Encyclopaedia of the Olympic Games. London: The Station-
Holder, R.L. and Nevill, A.M. (1997). Modelling perfor-
mance at international tennis and golf tournaments: is
there a home advantage? Journal of the Royal Statistical
Society, Series D (The Statistician), 46, 551–559.
Lozano, S., Villa, G., Guerrero, F. and Cortes, P. (2002).
Measuring the performance of nations at Summer
Olympics using data envelopment analysis. Journal of the
Operational Research Society, 53, 501–511.
Lyberg, W. (1999a). The Athletes of the Summer Olympic
Games, 1896–1996; Athletics. Bruty: Allsport.
Lyberg, W. (1999b). The Athletes of the Summer Olympic
Games, 1896–1996; Basketball. Bruty: Allsport.
Lyberg, W. (1999c). The Athletes of the Summer Olympic
Games, 1896–1996; Boxing. Bruty: Allsport.
Lyberg, W. (1999d). The Athletes of the Summer Olympic
Games, 1896–1996; Football. Bruty: Allsport.
Lyberg, W. (1999e). The Athletes of the Summer Olympic
Games, 1896–1996; Gymnastics. Bruty: Allsport.
Lyberg, W. (1999f). The Athletes of the Summer Olympic
Games, 1896–1996; Hockey. Bruty: Allsport.
Lyberg, W. (1999g). The Athletes of the Summer Olympic
Games, 1896–1996; Volleyball. Bruty: Allsport.
Lyberg, W. (1999h). The Athletes of the Summer Olympic
Games, 1896–1996; Waterpolo. Bruty: Allsport.
Lyberg, W. (1999i). The Athletes of the Summer Olympic
Games, 1896–1996; Weightlifting. Bruty: Allsport.
Madrigal, R. and James, J. (1999). Team quality and the
home advantage. Journal of Sport Behavior, 22, 381–398.
477Modelling home advantage in the Summer Olympic Games
McCutcheon, L.E. (1984). The home advantage in high
school athletics. Journal of Sport Behavior, 7, 135–138.
Moore, J.C. and Brylinsky, J.A. (1993). Spectator effect on
team performance in college basketball. Journal of Sport
Behavior, 16, 77–84.
Morton, R.H. (2002). Who won the Sydney 2000 Olympics?
An allometric approach. Journal of the Royal Statistical
Society, Series D (The Statistician), 51, 147–155.
Nevill, A.M. and Holder, R.L. (1999). Home advantage in
sport: an overview of studies on the advantage of playing
at home. Sports Medicine, 28, 221–236.
Nevill, A.M. and Stead, D. (2003). The relationship
between national sporting success and gross national
product: a law of diminishing returns. Journal of Sports
Sciences, 21, 283–284.
Nevill, A.M., Newell, S.M. and Gale, S. (1996). Factors
associated with home advantage in English and Scottish
soccer. Journal of Sports Sciences, 14, 181–186.
Nevill, A.M., Holder, R.L., Bardsley, A., Calvert, H. and
Jones, S. (1997). Identifying home advantage in interna-
tional tennis and golf tournaments. Journal of Sports
Sciences, 15, 437–443.
Nevill, A.M., Balmer, N.J. and Williams, A.M. (1999).
Crowd inﬂuence on decisions in association football.
Lancet, 353, 1416.
Nevill, A.M., Balmer, N.J. and Williams, A.M. (2002a). The
inﬂuence of crowd noise and experience upon refereeing
decisions in football. Psychology of Sport and Exercise, 3,
Nevill, A.M., Atkinson, G., Hughes, M. and Cooper, S.-M.
(2002b). Statistical methods for analysing discrete and
categorical data recorded in sport performance and
notation analyses. Journal of Sports Sciences, 20, 829–844.
Park, R. and Werthner, P. (1977). Analysis of politicized
bias in judging at the Olympic diving competitions,
Montreal, 1976. Canadian Journal of Applied Sports
Sciences, 2, 225.
Pollard, R. (1986). Home advantage in soccer: a retro-
spective analysis. Journal of Sports Sciences, 4, 237–248.
Schwartz, B. and Barsky, S.F. (1977). The home advantage.
Social Forces, 55, 641–661.
Seltzer, R. and Glass, W. (1991). International politics and
judging in Olympic skating events: 1968–1988. Journal of
Sport Behavior, 14, 189–200.
Wallechinsky, D. (2000). The Complete Book of the Olympics.
London: Aurum Press.
Whissell, R., Lyons, S., Wilkinson, D. and Whissell, C.
(1993). National bias in judgements of Olympic-level
skating. Perceptual and Motor Skills, 77, 355–358.
Winer, B.J. (1972). Statistical Principles in Experimental
Design, 2nd edn. Tokyo: McGraw-Hill.
Zar, J.H. (1998). Biostatistical Analysis, 4th edn. Upper
Saddle River, NJ: Prentice-Hall.
478 Balmer et al.