Journal of S port & Exe rcise Psychology, 2010, 32, 483-4 98
© 2010 Human Kinetics, Inc.
Christian Unkelbach is with the Psychologisches Institut, Universität Heidelberg, Heidelberg, Germany.
Daniel Memmert is with the Institut für Bewegungswissenschaft, Deutsche Sporthochschule Köln,
Crowd Noise as a Cue
in Referee Decisions Contributes
to the Home Advantage
Christian Unkelbach1 and Daniel Memmert2
1University of Heidelberg; 2German Sport University Cologne
The home advantage is one of the best established phenomena in sports (Courneya
& Carron, 1992), and crowd noise has been suggested as one of its determinants
(Nevill & Holder, 1999). However, the psychological processes that mediate
crowd noise inuence and its contribution to the home advantage are still unclear.
We propose that crowd noise correlates with the criteria referees have to judge.
As crowd noise is a valid cue, referee decisions are strongly inuenced by crowd
noise. Yet, when audiences are not impartial, a home advantage arises. Using soccer
as an exemplar, we show the relevance of this inuence in predicting outcomes
of real games via a database analysis. Then we experimentally demonstrate the
inuence of crowd noise on referees’ yellow cards decisions in soccer. Finally, we
discuss why the focus on referee decisions is useful, and how more experimental
research could benet investigations of the home advantage.
Keywords: sport psychology, home advantage, referee decisions, cue learning
Sport teams as well as individual athletes performing “at home” have higher
success rates than teams or athletes performing “away.” This home advantage is
one of the best established phenomena in sports and offers a highly interesting
research eld for psychology in general and sport psychology in particular (for
reviews, see Courneya & Carron, 1992; Carron, Loughhead, & Bray, 2005). Next
to other factors (e.g., familiarity, territoriality, travel fatigue), crowd noise has been
suggested as one of the major determinants of the home advantage (e.g., Clarke &
Norman, 1995; Nevill & Holder, 1999; Pollard, 1986). However, the psychological
processes that mediate the impact of crowd noise on performance outcomes are
still unclear and sometimes contradictory (e.g., compare Wallace, Baumeister, &
Vohs, 2005; Pollard & Pollard, 2005; and Holder & Nevill, 1997).
While many studies investigated the crowd inuences on individual athletes
(see the review by Carron et al., 2005), we will focus on crowd inuences on judges
and referees (Mascarenhas, O’Hare, & Plessner, 2006). We believe this focus is
useful for two reasons: First, judges and referees have a pivotal role in determining
484 Unkelbach and Memmert
performance and competition outcomes; if they are inuenced by the presence of
crowds and crowd noise, we have identied a strong contribution to the variance in
the home advantage. Second, previous studies showed that home advantages and
thereby possible crowd noise effects are reduced or absent when performances are
scored objectively rather than subjectively (e.g., weight lifting vs. gure skating;
see below, Nevill & Holder, 1999), again demonstrating the importance of referees
and judges. In other words, we believe that crowd noise contributes to the home
advantage by inuencing referees’ decisions and judges’ verdicts.
How might crowd noise inuence judgments and decisions in sport perfor-
mances? Building on a Brunswikian perspective (Brunswik, 1957), we proposed
that referees use crowd noise as a proximal cue to judge distal criteria (e.g., How
hard was a foul? or How aesthetically pleasing was a performance?). Brunswik’s
approach is best illustrated with depth perception: Although images on the retina
are strictly two dimensional, humans learn to infer the distal criterion of depth from
proximal cues, such as feature overlap or disparities between the images on both
eyes. For example, overlap correlates perfectly with depth, because objects that are
closer to the perceiver will hide more distant objects. Similarly, referees might use
crowd noise because they learn the correlation between crowd noise and specic
criteria in a given situation. For judgments in sports, this cue-learning hypothesis
is easily illustrated: For example, when aesthetic performance is judged, good
performances will elicit more cheering from the audience than bad performances.
Similarly, in team sports, clear fouls elicit more distinct responses from the audience
than less clear fouls. Even more, any foul elicits more crowd responses than no foul.
This leads to a substantial correlation between crowd noise and the criterion to be
judged. And just as people learn to use depth cues in perception to judge depth,
referees might learn to use crowd noise in referring to judge the respective criteria
(performance, fouls, etc.). Because home performers in gure skating receive
greater overall responses from the crowd, they should receive higher ratings, and
because the home crowd reacts more strongly toward fouls committed against the
home team, more free kicks are awarded to the home team, and the away team is
punished more frequently for playing rough. Consequently, a distinct and predictable
home advantage arises. We will discuss this cue-learning hypothesis in more depth
below, but rst we examine some of the available evidence on the contribution of
crowd noise on the home advantage.
A Cursory Literature Review
Consistent with the suggested importance of referee decisions for the home
advantage, there is a greater home advantage for sports in which subjective referee
decisions inuence a competition’s outcome (e.g., scoring in boxing, gure skat-
ing; Balmer, Nevill, & Lane, 2005), compared with sports in which performance
is measured objectively (e.g., weightlifting, short-track speed skating; Balmer,
Nevill, & Williams, 2003). The importance of subjectivity is strongly highlighted
in the study by Balmer and colleagues (2005): When boxing competition between
evenly matched ghters ended in a knockout (an objective criterion), the prob-
ability of a “home” win was only .57; however, when the match was decided by
technical knockouts or points decisions (i.e., subjective criteria), the probability of
a home win increased to .66 and .74, respectively. Thus, they showed greater home
Crowd Noise and Referee Decisions 485
advantages within a sport due to increased importance of subjective judgments.
Similarly, Balmer, Nevill, and Williams (2001) investigated the home advantage
for the Olympic Winter games from 1908 to 1998; across all events, they found a
home advantage. This advantage, however, was moderated by disciplines. Disci-
plines with objective performance measures (e.g., speed skating) showed less of
a home advantage. These data support Holder and Nevill’s (1997) conclusion that
the home advantage is absent when performance is scored objectively, without
subjective input from judges or referees.
Consistent with the suggested impact of the crowd, Schwartz and Barsky’s
(1977) classic study found a correlation between the home advantage and crowd
density (i.e., percentage used of available venue spectator capacity). Similar results
were reported by Clarke and Norman (1995), as well as Agnew and Carron (1994).
Nevill, Newell, and Gale (1996) even found an absolute crowd effect on the home
advantage in English and Scottish soccer. However, there are also many research-
ers who report failures to nd crowd effects on the home advantage (e.g., Strauss,
2002), or even performance impairments due to crowd inuences (Baumeister &
These are only a few prominent ashlights of research on how crowd noise
and refereeing decisions impact home advantages. However, all of the studies cited
above use archival data to test the impact of crowd noise on performance and referee
decisions. Although statistical models and methods have improved considerably
and have become much more rened over the years (e.g., from Schwartz & Barsky,
1977, to Pollard & Pollard, 2005), archival analyses can almost never show causal
relationships. To show cause-and-effect relations, experiments are necessary; yet,
to our knowledge, there are only two experimental studies that investigated the
impact of crowd noise on refereeing decisions.
First, Nevill, Balmer, and Williams (1999) presented 52 challenge scenes from
one European Champions League soccer match to eleven experienced referees
in the game’s chronological sequence. Half of the challenges were initiated by a
home team player and half were initiated by an away team player. The referees’
task was to call a foul or not. Six referees judged the scenes without sound, and
ve with the original sound from the game. When sound and thus, crowd noise
was present, referees called more challenges by the away team players; this effect
disappeared when no crowd noise was present. The authors concluded that when
in doubt, judges “refereed to the crowd for guidance” (p. 1416).
Second, Nevill, Balmer, and Williams (2002) presented 47 challenge scenes
from one English Premier League soccer match to 40 referees of varying expertise
level; 18 referees watched the challenges without sound and 22 with the original
sound from the stadium. Their task was to classify the challenge as a home team
player foul, an away team player foul, no foul at all, or uncertain. When referees
could hear the crowd noise, they called fewer fouls on the home team compared
with the away team. This effect was moderated by referees’ expertise; more expe-
rienced referees were less inuenced by crowd noise.
Although both experiments are highly informative, they have some limita-
tions; rst, both studies used a sound/no sound manipulation, which allows the
reinterpretation of watching a game under “natural conditions” (i.e., with sound)
and “unnatural conditions” (i.e., without sound). This is a proper way to maximize
the variation on the independent variable, but it is a slight weakness of both designs
486 Unkelbach and Memmert
and it prohibits the direct conclusion that crowd noise inuences referee decisions
in favor of the home team (i.e., by the way of more noise in favor of the home
team). Rather, it tests the hypothesis that home crowd noise in comparison with
no home crowd noise biases decision toward the home team. Second, and more
importantly, both studies used only challenge incidents from one game each (for
example, Liverpool vs. Leicester City, in Nevill et al., 2002). In terms of stimulus
sampling, this could be seen as a serious aw, as it might be something about these
particular games or these particular crowds that causes the observed effects (cf.
Wells & Windschitl, 1999). Sutter and Kocher (2004, p. 463) already noted that it
is impossible in this design to distinguish a “home bias” from a “Liverpool bias.”
The problem is highlighted by the differential outcomes of otherwise very similar
experiments. The rst study shows an increase of fouls called on the away team,
while the latter shows an decrease of fouls called on the home team. Nevill and
colleagues (2002) tried to reconcile their contradictory ndings by introducing the
idea that referees tried avoiding a “bad call” against the home team, or in general,
they try to avoid unpopular decisions.
The hypothesis presented by Nevill and colleagues (2002) is a motivational
one because referees want to avoid displeasing the crowd, and more so when there
is crowd noise; thus, referees call fewer fouls on the home team when crowd noise
is present. This idea is substantiated by data from Balmer and colleagues (2007),
who showed that crowd noise is associated with increased anxiety and mental
effort in referees; therefore, referees tended to cope with such situations with more
popular decisions for the home team. Sutter and Kocher (2004) present a similar
motivational account, based on the idea that referees are informed agents who
want to balance two different goals: being impartial to their employers (i.e., the
soccer governing institutions) and pleasing the crowd. As the crowd has the more
immediate inuence, a home bias arises.
The problem with such motivational explanations is that they also allow the
opposite outcome. Just because referees feel bullied by the home crowd, they
might deliberately act against home teams/athletes (as predicted by reactance
theory, Miron & Brehm, 2006). Further, motivational inuences have difculties
explaining results in experimental laboratory settings. For example, in the experi-
ment by Nevill and colleagues (2002), the question arises why referees should try
to appease a crowd that is not present at all—one needs the additional assumption
that referees indeed act “as if in the game.”
In contrast, the present cue hypothesis is a purely cognitive concept that
explains biased decisions due to crowd noise as a simple case of cue learning (i.e.,
it only matters whether the cue is present or not), which allows testable predictions
independent of the specic settings (i.e., laboratory vs. real games). However,
motivational explanations are still a necessary construct because there are ndings
and results that our pure cue-learning hypothesis cannot explain; for example, the
data by Nevill et al. (2002), or effects of crowds on biased allowed extra time in
soccer (e.g., Dohmen, 2008). We will return to this issue in the general discussion.
Theoretical Background: The Case of Cue Learning
Our hypothesis is built on a Brunswikian perspective; according to Brunswik
(1957), there are many properties to which people have no direct sensory access.
Crowd Noise and Referee Decisions 487
To assess these distal properties, people rely on proximal cues. Again, the most
prominent case is depth perception. People’s visual perception does not assess depth
directly; rather, people infer depth from cues provided by the two two-dimensional
images on the retina. The most important cues to infer depth are overlap (objects
that hide other objects are closer to the perceiver), motion parallax (when the per-
ceiver moves, objects that are closer seem to move faster than faraway objects),
and texture gradients (distant objects have denser textures than close objects). And
infants learn by haptic feedback, while trying to reach for objects, such that if one
object overlaps another object, the former is closer than the latter. In other words,
people learn the correlation between a cue (here: overlap) and a criterion (here:
distance). This cue usage is often so over-learned that in most cases, people are
not even aware of using these cues at all.
This Brunswikian perspective, the need to use proximal cues to assess otherwise
inaccessible properties, is by no means restricted to basic perceptual processes. For
example, recipients of persuasive speeches are more convinced if the message is
accompanied by favorable audience reactions (Axsom, Yates, & Chaiken, 1987).
These authors also framed their results as such that people use audience reactions
as a cue in their judgments. Similarly, people laugh longer at a funny video clips
when another laughing person is present, again showing how people use an avail-
able cue to evaluate a distal criterion (Devereux & Ginsburg, 2001).
The implications for judgment and decision making in sports are immediately
obvious: If referees have to judge aesthetic performance in gure skating, they
cannot access this criterion directly, but must infer it from multiple cues that are
integrated into a judgment. If referees have to judge the severity of a challenge in
soccer, they also have to integrate multiple cues into a nal judgment (Mascarenhas
et al., 2006; Plessner, Schweizer, Brand, & O’Hare, 2009). And because they learn
that crowd noise correlates with the severity of fouls (or the excellence of perfor-
mance), they use crowd noise as an additional cue in their judgment processes,
similar to cue learning in perception (e.g., Jacobs, 2002), memory judgments
(Unkelbach, 2006), or decision making (e.g., Evans, Clibbens, Cattani, Harris, &
This Brunswikian cue-learning approach shares some properties with the idea
that referees use crowd noise as a “judgmental heuristic.” However, the notion of
a heuristic carries three problems: First, as dened in the judgment and decision
literature, heuristics are normally employed when people are not motivated or
accountable for their decision, as a mental effort–saving device (e.g., Chaiken, 1980;
Chaiken & Trope, 1999). This does not t with most referees’ situations, although
the time pressures that referees experience are often used as manipulations to force
people into a heuristic judgment mode. Second, heuristics are conceptualized as
available “rules of thumb”; in contrast, many authors attribute the inuence on
referees to processes that are not accessible to awareness (e.g., Nevill and Holder,
1999, p. 232; Sutter & Kocher, 2004, p. 468). And third, heuristics do not explain
how they are acquired, whereas the cue-learning approach species a perception-like
process, based on feedback learning (e.g., Jacobs, 2002). Thus, we believe the idea
of a cue-learning process is superior to the idea of a heuristic judgment process; it
builds on well-established models in perception and psychophysics, and is more
parsimonious in its assumptions. However, more recently, the use of cues has been
called a “heuristic” as well (e.g., Hertwig, Herzog, Schooler, & Reimer, 2008). In
488 Unkelbach and Memmert
that sense, our idea that referees use crowd noise as a cue is indeed a judgmental
heuristic, but then the label no longer has implications for the underlying processes.
In the remainder of this section, we investigate the Brunswikian cue approach in
German soccer, but the present hypothesis applies to all sports in which subjective
referee judgments and decisions inuence competition outcomes. Our dependent
variable of interest is awarding a yellow card for committed fouls. It is important
to keep in mind that the case of yellow cards is only one example of a way to learn
how crowd noise contributes to the home advantage via referee decisions—the
same logic is applicable to virtually all subjective judgments and decisions in the
Yellow cards have been implemented in soccer since the 1970 World Cup
as an ofcial warning sign for rough and dangerous fouls and unsportsmanlike
behavior; as such, they have become one of the most important instruments to
regulate soccer games. They are ideally suited to study judgment and decision
processes because referees have a great deal of freedom with when to award such
warnings or not (e.g., Unkelbach & Memmert, 2008). The probability to award
a yellow card should increase with the roughness of a given foul. And as crowd
noise should also be a direct function of how rough a foul was, referees should
learn the correlation between crowd noise and foul roughness. Hence, referees
should award yellow cards with higher probability in the given high crowd noise
compared with low crowd noise.
In Study 1, we investigate the impact of crowd noise on yellow card decisions
in soccer, using data from the Bundesliga, the highest soccer league in Germany. We
will show that referees award more yellow cards to the away team, an effect that is
amplied when crowd density is high (see above; Schwartz & Barsky, 1977), and
when games are played in “pure” soccer stadiums compared with stadiums with a
running track (i.e., when the crowd is closer to the eld, and thus, the referee). Yet,
as we have argued, archival data does not allow causal conclusions. Thus, Study 2
will present an experiment that avoids some of the design problems of the experi-
mental studies so far and corroborate the effects we found in the databank analysis.
Study 1: A Databank Analysis Testing the Inuence
of Crowd Noise on Yellow Cards in Soccer
Our hypothesis predicts that more crowd noise should lead to more yellow cards
against the away team, and consequently, a home advantage. To test this, we ana-
lyzed ve seasons of the German Bundesliga, the highest soccer league in Germany
(1997/98 through 2001/2002).
Materials and Indices
The German Bundesliga has 18 teams, which play a rst and second leg per season;
thus, the data set included 1530 games. We dened the home advantage as the dif-
ferences in goals scored by the home and away team (i.e., goals scored by the home
minus goals scored by the away team; for example, as done by Boyko, Boyko, and
Boyko, 2007). Similarly, we dened a yellow card effect as the difference between
yellow cards awarded against the home and away team (i.e., yellow cards awarded
against the away minus yellow cards awarded against the home team). We included
Crowd Noise and Referee Decisions 489
only rst yellow cards in the analysis; second yellow cards against a player (i.e.,
yellow-red cards and thereby send-offs) were not included (cf. Downward & Jones,
2007). When these indices are positive, they testify to a home advantage. For
crowd noise, we used crowd density as an index, that is, the percentage used of a
stadium’s absolute visitor capacity.1 Additionally, the architecture of the different
venues in which all these games took place allows for a more rened test of the
hypothesis that larger crowds lead to more yellow cards against the away team. In
Germany, there are two kinds of stadiums: First, there are all-purpose venues that
are characterized by a track and eld lane that separates the pitch from the audi-
ence. Second, there are “pure” stadiums, in which the crowd is not separated from
the pitch. Consequently, games in pure stadiums should show amplied crowd
effects because crowd noise is more directly transmitted to referees and players.
Study 1 Results
First, we checked whether a home advantage exists at all in this data set. On aver-
age, home teams scored 1.72 (SD = 1.36) goals and away teams only 1.17 (SD
= 1.13) goals, t(1529) = 11.98, p < .001.2 This result is consistent with the data
reported by Clarke and Norman (1995), who also found a home advantage of about
half a goal in English soccer. However, this home advantage did not correlate with
crowd density (again, visitors number relative to stadium size), r(1530) = –.006,
ns, whereas other authors found correlations of crowd density with outcomes (e.g.,
Schwartz & Barsky, 1977; Agnew & Carron, 1994).
Note, however, that we did not predict a direct inuence of crowd noise on
competition outcomes. To support our hypothesis, there should be a difference in
yellow card frequency for the home and the away team. Overall, 6489 yellow cards
were awarded. On average, home teams were awarded 1.89 (SD = 1.19) yellow
cards, while the away teams were awarded 2.35 (SD = 1.27) yellow cards, t(1529)
= 11.10, p < .001. The effect size is astonishing given the mean of yellow cards
awarded per game (i.e., 4.24). Most importantly, this difference in yellow cards
correlated signicantly with crowd density, r(1530) = .134, p < .001. And nally,
the difference in yellow cards correlated signicantly with the difference in goals,
our index for the home advantage, r(1530) = .095, p < .001.
Pure Soccer vs. Track and Field Stadiums
As discussed, the effects of crowd density and therefore crowd noise should be
amplied in pure soccer stadiums due to the closer proximity of crowd and refer-
ees. Our sample included 543 games in pure and 987 games in nonpure stadiums,
which allowed testing this hypothesis. Indeed, in pure stadiums, the difference
in yellow cards awarded in favor of the home time was Mpure = 0.66, whereas in
stadiums with a track-and-eld lane, the difference was reduced Mnonpure = 0.359,
t(1084) = 3.42, p < .001 (degrees of freedom are corrected for unequal variances).
The correlation between crowd density and yellow cards also shows the expected
effect. In pure stadiums, the correlation is larger, r(543) = .140, p < .001, than in
nonpure stadiums, r(987) = .093, p < .005; the difference in correlations, however,
is not signicant. Further, the distinction between pure and nonpure stadiums also
explains the lack of a direct relation between crowd density and goal differences
490 Unkelbach and Memmert
for the home and away teams. For games in pure stadiums, crowd density cor-
related signicantly with goal difference, r(543) = .120, p < .005. For games in
stadiums with a track-and-eld lane, no such correlation existed, r(987) = –.025,
ns. This leads to the nonsignicant overall correlation of crowd density and home
advantage reported above.
Differential Effects for Home and Away Teams
Nevill and colleagues (2002) argued, based on their experimental results, that the
dominant effect of crowd noise is to reduce the number of fouls called against the
home team. This contradicts the cue-learning approach and the data presented by
Nevill and colleagues (1999). The present archival data analysis allows testing these
competing predictions; if Nevill and colleagues (2002) are correct, then crowd
density should reduce the number of yellow cards awarded to the home team (i.e.,
a negative correlation). The cue-learning approach predicts that crowd density
should only increase the number of yellow cards awarded to the away team (i.e.,
a positive correlation). In the present data, the answer is clear cut: Across games,
crowd density did not correlate with the number of home team yellow cards, r(1530)
= –.009, ns, but did correlate with the number of away team yellow cards, r(1530)
= .164, p < .001.
Study 1 Discussion
Based on our hypothesis, we found that crowd density predicts the amount of
yellow cards awarded against the away team, which in turn correlates with game
outcome. Further, this relationship was strengthened when referees and crowds
were in close proximity, which supposedly amplies the impact of crowd noise
as a cue. This is an important insight because some authors failed to nd crowd
effects on outcomes (e.g., Strauss, 2002); if referees and venues were considered,
this inconsistency disappears and allows new and testable hypotheses (at least in
our data set). Crowd noise does not directly inuence outcomes, but inuences
referee decisions; additionally, this effect partially depends on the strength of the
cue (i.e., proximity of referees and crowd). In turn, only if these referee decisions
or judgments are inuential enough for the outcome should crowd size, and in the
present case, the kind of stadium, inuence competition outcomes.
In addition, the separate analysis of yellow cards awarded to the home and away
teams clearly supported the Brunswikian approach; higher home crowd density
leads to more cards against the away team, but does not inuence card decisions
toward the home team. While this is highly consistent with a cue-learning approach,
it contradicts the idea that referees want to appease the home crowd by calling
fewer fouls and awarding fewer cards toward the home team.
One might argue that the observed correlations, albeit highly signicant, are
not impressive. Yet, in soccer, there are many other decisions that correlate with
crowd noise that heavily inuence a game’s outcome (e.g., free kicks, red cards,
and penalties; see Boyko, Boyko, & Boyko, 2007). We did not consider them here
because the following experiment used yellow cards as the main dependent vari-
able. As argued above, without experimental evidence to back up archival data,
there are many simple counter-arguments: For example, losing teams could start
Crowd Noise and Referee Decisions 491
to play more aggressive and the causality is actually reversed; because away teams
lose more frequently, they receive more yellow cards. Similarly, better teams could
simply have larger or better attended venues and consequently, this third variable
could create the correlation. A crowd density index partly avoids this problem (see
Note 2), but one would need the additional control that better clubs do not simply
have higher crowd density. In addition, crowd density could also cause more rev-
enue for teams and clubs, allowing them to attract better players, leading to better
performances. Again, the causality would be reversed.
Given this caveats, the data are highly consistent with earlier reports on the
relation between crowd size and referee decisions in soccer. For example, Downward
and Jones (2007) reported a similar trend for crowd size and number of yellow
cards awarded against the away team. They analyzed 857 games of the Football
Association Cup in England and found that 1.71 rst yellow cards were awarded
toward away teams, whereas only 1.35 cards were awarded toward home teams;
again, a highly signicant difference (but also see Note 3).
However, our analysis did not include many other variables authors have
suggested that might moderate the relation between crowd noise and the home
advantage, for example, nonlinear relationships between crowd size and home bias
(e.g., Downward & Jones, 2007),3 crowd composition (i.e., the ratio of home and
away spectators; Garicano, Palacios-Huerta, & Prendergast, 2005), or personality
variables of the referees (e.g., Page & Page, in press). The point of the present data-
bank analysis was to demonstrate that there is a systematic home bias in awarding
yellow cards and this bias is related to crowd density, our proxy for crowd noise.
Having established the basic phenomenon, we can now turn to a direct experimental
test of the hypothesis that referees use crowd noise as a cue.
Study 2: An Experiment Testing the Inuence
of Crowd Noise on Yellow Cards in Soccer
The cue-learning hypothesis predicts that louder crow noise should inuence referee
decisions in the direction of the cue’s correlation with the criterion (e.g., noise and
foul severity). Hence, they should award yellow cards with higher probability in the
same scene given high crowd noise compared with low crowd noise. The following
experiment presents a direct test of this prediction.
Study 2 Method
Participants, Design, and Materials
Twenty male referees of the German Football Association (DFB) participated
(Mage = 22.5; SD = 8.04), and they were recruited during an educational workshop
at a DFB training center. They had refereed across various levels of DFB leagues
(depending on their age), with a minimum experience of two years. The main
independent variable, whether a scene was presented with high or low crowd noise
volume, was manipulated within participants.
We used 56 digitally available foul scenes, from 56 different soccer games, that
were successfully employed in previous experiments on referee decision making
492 Unkelbach and Memmert
(Memmert, Unkelbach, Ertmer, & Rechner, 2008; Unkelbach & Memmert, 2008).
Twenty-eight of these scenes led to a factual yellow card on the pitch and 28 did
not. Naturally, the scenes did not include any hint to the actual decisions and they
were rated to be of equal roughness (see Unkelbach & Memmert, 2008). For crowd
noise, we used four different sound les containing crowd noise from other soccer
stadiums when a foul just happened (i.e., the wavelike increase and decrease when
something of importance happens in a stadium). We randomly combined these
sound les with the video clips of the fouls scenes (i.e., each sound le was used
14 times). The volume for each scene was set to high and low randomly during
the actual presentation (see below). This procedure avoids confounding crowd
noise with other factors inherent in the scenes, and crowd noise was thus fully
independent of the home and away status of a team. This design also ameliorates
the problem of stimulus sampling, as we used scenes from 56 different games. In
addition, each scene would be played with high and low volume instead of sound
vs. no sound, which avoids the problem that watching a scene without sound is
a somewhat unnatural condition for referees. Thus, in this design, any difference
between a scene presented with low volume and high volume can only be due to
the volume level of crowd noise.
The experimenter informed the referees that they would participate in an experi-
ment to determine how and when referees award yellow cards. All referees agreed
to participate. Then, the experimenter led a rst group of ten referees into a seminar
room of the training center, which was equipped with a video projector and a
sound system. Referees sat around the projector screen. They received a 56-page
questionnaire with two boxes on each page to indicate whether they would award a
yellow card for a presented foul or not. The experimenter instructed them that they
should make their decision individually and “as if in an actual game.” Then the
foul scenes presentation started. A Microsoft Visual Basic computer program con-
trolled this presentation and the computer’s output volume level, but the volume
level of the room’s sound system was xed. The program randomly selected half
of the scenes from the factual yellow card and no yellow card categories as high
volume scenes (presented at 90% sound output) and the remaining as low volume
scenes (presented at 10% sound output), resulting in an approximately 50 dB dif-
ference in the volume in terms of the physically equivalence. Each scene lasted
4–7 s, and participants made their decisions immediately afterward. After each
decision round, the experimenter prompted the program to continue. Following
the 56 scenes, the referees answered demographic questions and were informed
that the study was now over. Then, the second group was called into the room.
Everything was identical to the rst group, with one exception: The presenta-
tion order of the scenes was not randomized, but yoked with the rst group, and
each scene that was selected as a high volume scene was now presented as a low
volume scene (and vice versa). After the second group nished, the experimenter
debriefed all participants about the hypothesized effects and thanked them for
Crowd Noise and Referee Decisions 493
Study 2 Results
In postexperimental funneled questioning, no referee reported suspicion about the
sound volume and only one mentioned the variations in sound. We rst analyzed
participants’ mean probability to award yellow cards as a function of high and low
volume and actual decision (yellow card or not). As predicted, referees awarded
yellow cards with higher probability when scenes were presented with high volume
(M = .589, SD = .088) compared with when the same scenes were presented with
low volume (M = .486, SD = .126), t(19) = 3.88, p < .001, d = 1.78. As could be
expected, referees also awarded yellow cards with higher probability when an
actual yellow card had been awarded (M = .723, SD = .164) compared with when
no actual yellow card had been awarded (M = .351, SD = .138), t(19) = 14.01, p <
.001, d = 4.27. However, there was no interaction of these two variables, F(1, 19)
= 0.06, ns, indicating that the inuence of crowd noise was independent from the
factual decision on the pitch.
The previous analysis is collapsed across groups; Table 1 presents the full
design (actual decision on the pitch: yellow cards vs. no yellow cards × volume:
high vs. low × group: rst vs. second group), corrected for referees’ mean propensity
to award a yellow card, making the comparison of cells equivalent to an overall
repeated measurement analysis. This table allows the direct comparison of the same
scenes under high and low volume conditions. As can be seen, for all comparisons
involving the same scenes, high volume led to higher probabilities of a yellow card.
A mixed ANOVA4 with all three factors (group, factual decisions, and volume, with
repeated measures on the latter two factors) delivers the same results as the simple t
tests, demonstrating a highly signicant effect of volume F(1, 18) = 14.96, p < .001,
d = 1.82. When scenes were presented with high volume referees were more likely
to award yellow cards compared with when the same scenes were presented with
low volume. More importantly, this analysis also shows that the effect of volume
did not interact with group or factual decision on the pitch, Fs < 1, ns.
Study 2 Discussion
High volume crowd noise led to substantially more yellow cards than low volume
crowd noise. Presented with high volume crowd noise, referees had an approxi-
mately .10 higher probability to award a yellow card than when the identical scene
Table 1 Mean Decision Proportions to Award a Yellow Card as a
Function of Factual Decision, Group, and High vs. Low Volume (SD
in Parentheses). Comparisons of High and Low Volume Are Based
on Identical Scenes.
Factual Decision High Volume
Yellow Card .826 (.084) .684 (.151) .720 (.105) .669 (.171)
No Yellow Card .332 (.124) .261 (.097) .485 (.078) .332 (.102)
494 Unkelbach and Memmert
was presented with low volume crowd noise. As argued, we believe this effect is
due to referees learning the correlation between foul severity and the crowd noise.
In this vein, the present results differ from the data by Nevill, Balmer, and
Williams (2002), who found fewer challenges awarded for the home team when
crowd noise was present. Yet, the data converge with the results from Nevill,
Balmer, and Williams (1999). Given this convergence, together with Sutter and
Kocher’s critique (2004) and the data presented in our archival analysis, it seems
possible that there is something specic about the game used in Nevill and col-
leagues’ 2002 experiment.
One limitation of the current study, and probably of most experimental studies,
is the laboratory situation. Referees who judge scenes on a video screen are hardly
in the same situation as referees on the pitch. At present, we see now better and
feasible way to investigate the effects with clear manipulations of the independent
variable. Yet, in our opinion, the combination of archival data analyses—that is,
factual judgments and decisions, with somewhat articial, but highly controlled
laboratory experiments—provides a useful compromise. For example, Hill and
Barton (2005) argued, based on archival data, that wearing red in a combat sport
such as wrestling is benecial for the athlete. However, more recently, Hagemann,
Strauss, and Leissing (2008) found in an experiment that it is indeed a refereeing
bias that creates the advantage for wearing red. Given identical performances,
judges awarded more points in Tae Kwon Do to ghters who wore red compared
with ghters who wore blue. Thus, neither pure archival data nor laboratory data
alone can answer the most interesting questions in sport psychology.
The psychological demands of refereeing and performance judgments have insti-
gated increasingly more research within the last few years (Mascarenhas, O’Hare,
& Plessner, 2006; Plessner & Haar, 2006). Building on this prior research, we
demonstrated the impact of crowd noise on referees’ decisions when they have to
judge the severity of foul scenes; thereby, we showed the possible contribution of
crowd noise to the home advantage via referee decisions. The underlying model is
the assumed correlation between crowd noise, a proximal cue, and the severity of
the foul, the distal criterion. The cue should be context specic, so that the same
loud audience reaction leads to more positive evaluations in gure skating, but to a
higher probability of a yellow card for fouls in soccer. Thus, although we have only
used soccer as an exemplar, the model is applicable to all sports in which subjec-
tive judgments inuence a competition’s outcome or the rating of a performance;
this guiding model is implicitly present in many other conceptions of crowd noise
inuence (e.g., Nevill et al., 1999), and even explicitly spelled out in models of
referee decision making (e.g., Plessner et al., 2009).
Yet, the present studies did not test the underlying cognitive process directly; at
present, the advantage of our cue-learning approach over other explanations (e.g.,
motivational approaches, and crowd noise as a heuristic) rests in its foundation in
well-established cognitive models and the clearly testable predictions. The direct
test of the cue-learning hypotheses is a line of research we are pursuing right now.
Based on other successful cases on learning and relearning cues (e.g., Unkelbach,
Crowd Noise and Referee Decisions 495
2006, 2007), we are creating situations in which crowd noise is no longer a valid
cue or even correlates inversely with the criterion to be judged; that is, referees
should learn that crowd noise is actually louder when there is no foul and judge
following incidents accordingly. If this relearning is successful, it will provide
strong evidence for the case of cue learning.
The Brunswikian approach is a pure cognitive model for the inuence of crowd
noise on home advantages. However, we do not want to preclude motivational
accounts on the home bias. For example, Dohmen (2008) as well as Garicano
and colleagues (2005) presented evidence that soccer referees allow differential
amounts of extra time, depending on the score: Referees allowed more extra time
when home teams needed one more goal to win or even the score, whereas they
allowed less extra time when home teams were leading. Such effects cannot be
explained by our cue-learning hypothesis; awarding extra time is a deliberate
decision, and motivational factors have greater explanatory power in this domain.
It will be a challenge for future research to reconcile and integrate deliberate and
motivational processes with basic cognitive effects such as cue learning in order to
achieve a more comprehensive model of the home advantage. As it is, we believe
the presented experiment clearly establishes the inuence of crowd noise on referee
decisions. In combination with the present archival data, we are condent about
the evidence that crowd noise inuences referees’ yellow cards decisions, which
in turn contributes to a home advantage.
1. The use of crowd density as a proxy for crowd noise instead of absolute visitor numbers
has an additional advantage; in the German Bundesliga, the size of the stadium correlates with
the strength of the team, as better teams have larger stadiums. This problem does not exist with
2. We do not report effect sizes for the databank analysis; all reported effects (i.e., mean differ-
ence in goals scored, mean differences in yellow cards awarded) refer to scales with an inherent
meaning, and can therefore be judged according to “practical” signicance (Kirk, 1996; Thompson,
2002). In addition, correlation coefcients are effect size indicators by themselves (Rosenthal &
3. The nonlinear trend reported by Downward and Jones (2007) ts nicely with well-established
models of social impact (e.g., Latané, 1981). However, the data reported by Downward and Jones
suggests some problems with the analysis. It is not clear whether variables were centered before
using them in the logistic regression analysis and the rather weak effects in terms of signicance
suggest multicollinearity in the regression. Thus, these results should be treated with caution.
4. Many researchers advocate the use of logistic regression for a binary dependent variable.
However, standard regression models, which factually underlie the presented ANOVA results,
have no trouble dealing with binary dependent variables, given sufcient cell frequencies and no
extreme values (<.05 and >.95; Lunney, 1970). We believe the ANOVA presentation is easier to
understand than logistic regression results.
The present research was supported by a grant from the Deutsche Forschungsgemeinschaft
(DFG UN 273/1-1), awarded to the rst author.
496 Unkelbach and Memmert
Agnew, G., & Carron, A. (1994). Crowd effects and the home advantage. International
Journal of Sport Psychology, 25, 53–62.
Axsom, D., Yates, S.M., & Chaiken, S. (1987). Audience response as a heuristic cue in
persuasion. Journal of Personality and Social Psychology, 53, 30–40.
Balmer, N.J., Nevill, A.M., & Lane, A.M. (2005). Do judges enhance home advantage in
European championship boxing? Journal of Sports Sciences, 23, 409–416.
Balmer, N.J., Nevill, A.M., & Williams, A.M. (2001). Home advantage in the Winter Olym-
pics (1908–1998). Journal of Sports Sciences, 19, 129–139.
Balmer, N.J., Nevill, A.M., & Williams, A.M. (2003). Modelling home advantage in the
Summer Olympic Games. Journal of Sports Sciences, 21, 469–478.
Balmer, N.J., Nevill, A.M., Lane, A.M., Ward, P., Williams, A.M., & Fairclough, S. (2007).
Inuence of crowd noise on soccer refereeing consistency in soccer. Journal of Sport
Behavior, 30, 130–145.
Baumeister, R.F., & Steinhilber, A. (1984). Paradoxical effects of supportive audiences on
performance under pressure: The home eld disadvantage in sports championships.
Journal of Personality and Social Psychology, 47, 85–93.
Boyko, R.H., Boyko, A.R., & Boyko, M.G. (2007). Referee bias contributes to home advan-
tage in English Premiership football. Journal of Sports Sciences, 25(11), 1185–1194.
Brunswik, E. (1957). Scope and aspects of the cognitive problem. In H. Gruber, K.R.
Hammond, & R. Jessor (Eds.), Contemporary approaches to cognition (pp. 5–31).
Cambridge: Harvard University Press.
Carron, A.V., Loughhead, T.M., & Bray, S. (2005). The home advantage in sport competi-
tions: Courneya and Carron (1992) conceptual framework a decade later. Journal of
Sports Sciences, 23, 395–407.
Chaiken, S. (1980). Heuristic versus systematic information processing and the use of source
versus message cues in persuasion. Journal of Personality and Social Psychology, 39,
Chaiken, S., & Trope, Y. (1999). Dual-process theories in social psychology. New York,
NY: Guilford Press.
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 & Exercise Psychology, 14, 13–27.
Devereux, P., & Ginsburg, G. (2001). Sociality effects on the production of laughter. The
Journal of General Psychology, 128, 227–240.
Dohmen, T.J. (2008). The inuence of social forces: Evidence from the behavior of football
referees. Economic Inquiry, 46, 411–424.
Downward, P., & Jones, M. (2007). Effects of crowd size on referee decisions: Analysis of
the FA Cup. Journal of Sports Sciences, 25, 1541–1545.
Evans, J.S., Clibbens, J., Cattani, A., Harris, A., & Dennis, I. (2003). Explicit and implicit
processes in multicue judgment. Memory & Cognition, 31, 608–618.
Garicano, L., Palacios-Huerta, I., & Prendergast, C. (2005). Favoritism under social pressure.
The Review of Economics and Statistics, 87, 208–216.
Hagemann, N., Strauss, B., & Leissing, J. (2008). When the referee sees red.... Psychologi-
cal Science, 19, 769–771.
Hertwig, R., Herzog, S., Schooler, L., & Reimer, T. (2008). Fluency heuristic: A model of
how the mind exploits a by-product of information retrieval. Journal of Experimental
Psychology. Learning, Memory, and Cognition, 34, 1191–1206.
Hill, R., & Barton, R. (2005). Red enhances human performance in contests. Nature,
Crowd Noise and Referee Decisions 497
Holder, R.L., & Nevill, A.M. (1997). Modelling performance at international tennis and golf
tournaments: Is their a home advantage? Journal of the Royal Statistical Society. Series
A, (Statistics in Society), 46, 551–559.
Jacobs, R.A. (2002). What determines visual cue reliability? Trends in Cognitive Sciences,
Kirk, R. (1996). Practical signicance: A concept whose time has come. Educational and
Psychological Measurement, 56, 746–759.
Latané, B. (1981). The psychology of social impact. The American Psychologist, 36, 343–356.
Lunney, G.H. (1970). Using analysis of variance with a dichotomous dependent variable: an
empirical study. Journal of Educational Measurement, 7, 263–269.
Mascarenhas, D., O’Hare, D., & Plessner, H. (2006). The psychological and performance
demands of association football refereeing. International Journal of Sport Psychology,
Memmert, D., Unkelbach, C., Ertmer, J., & Rechner, M. (2008). Gelb oder kein Gelb?
Persönliche Verwarnungen im Fußball als Kalibrierungsproblem [Yellow card or no
yellow card? Soccer cautioning as a calibration problem]. Zeitschrift für Sportpsy-
chologie, 15, 1–11.
Miron, A.M., & Brehm, J.W. (2006). Reactance theory: 40 years later. Social Psychology,
Nevill, A.M., & Holder, R.L. (1999). Home advantage in sport: An overview of studies on
the advantage of playing at home. Sports Medicine (Auckland, N.Z.), 28, 221–236.
Nevill, A.M., Balmer, N.J., & Williams, A.M. (2002). The inuence of crowd noise and experi-
ence upon refereeing decisions in football. Psychology of Sport and Exercise, 3, 261–272.
Nevill, A.M., Balmer, N.J., & Williams, A.M. (1999). Crowd inuence on decisions in
association football. Lancet, 353(9162):1416.
Nevill, A.M., Newell, S.M., & Gale, S. (1996). Factors associated with home advantage in
English and Scottish soccer. Journal of Sports Sciences, 14, 181–186.
Page, K., & Page, L. (2010). Alone against the crowd: Individual differences in referees’
ability to cope under pressure. Journal of Economic Psychology, 31, 192–199.
Plessner, H., & Haar, T. (2006). Sports performance judgments from a social cognition
perspective. Psychology of Sport and Exercise, 7, 555–575.
Plessner, H., Schweizer, G., Brand, R., & O’Hare, D. (2009). A multiple-cue learning
approach as the basis for understanding and improving soccer referees’ decision
making. In M. Raab, J. Johnson, & H. Heekeren (Eds.), Progress in Brain Research:
Mind and Motion: The Bidirectional Link between Thought and Action (pp. 151–158).
Amsterdam: Elsevier Press.
Pollard, R. (1986). Home advantage in soccer: A retrospective analysis. Journal of Sports
Sciences, 4, 237–248.
Pollard, R., & Pollard, G. (2005). Long-term trend in home advantage in professional team
sports in North America and England (1876-2003). Journal of Sports Sciences, 23,
Rosenthal, R., & Rubin, D. (2003). r-sub(equivalent): A Simple Effect Size Indicator. Psy-
chological Methods, 8, 492–496.
Schwartz, B., & Barsky, S.F. (1977). The home advantage. Social Forces, 55, 641–661.
Strauss, B. (2002). The impact of supportive sports spectators on performance in team sports.
International Journal of Sport Psychology, 33, 372–390.
Sutter, M., & Kocher, M.G. (2004). Favoritism of agents – The case of referees’ home bias.
Journal of Economic Psychology, 25, 461–469.
Thompson, B. (2002). ‘Statistical,’ ‘practical’, and ‘clinical’: How many kinds of signicance
do counselors need to consider? Journal of Counseling and Development, 80, 64–71.
Unkelbach, C. (2006). The learned interpretation of cognitive uency. Psychological Sci-
ence, 17, 339–345.
498 Unkelbach and Memmert
Unkelbach, C. (2007). Reversing the truth effect: Learning the interpretation of processing
uency in judgments of truth. Journal of Experimental Psychology. Learning, Memory,
and Cognition, 33, 219–230.
Unkelbach, C., & Memmert, D. (2008). Game-Management, Context-Effects and Calibration:
The case of yellow cards in soccer. Journal of Sport & Exercise Psychology, 30, 95–109.
Wallace, H.M., Baumeister, R.F., & Vohs, K.D. (2005). Audience support and choking under
pressure: A home disadvantage. Journal of Sports Sciences, 23, 429–438.
Wells, G., & Windschitl, P. (1999). Stimulus sampling and social psychological experimenta-
tion. Personality and Social Psychology Bulletin, 25, 1115–1125.
Manuscript received: October 5, 2009
Revision accepted: April 1, 2010