Disciplinary Sanctions

DISCIPLINARY SANCTIONS IN
EUROPEAN CUP COMPETITIONS
Dr Peter Dawson
University of Bath, UK
IASE 10th Annual Conference, Gijon, Spain. 9-10
May 2008.
INTRODUCTION
A number of studies have investigated the behaviour of agents in sporting contests.
• Managerial and coaching decision-making (Clement and McCormick, 1989)
• Strategic behaviour of penalty-takers and goalkeepers (Palacios-Huerta, 2003)
Another important relationship is between players and match officials (referees).
• In football, as indeed in most sports, referees are assigned the task of
implementing the laws of the game and ensuring that players abide by its
regulations.
• Evidence based on match analysis from the European Championship of 2000
suggests that a top official would make an average 137 observable interventions
during any one game.
INTRODUCTION
The performance of players and the actions taken by referees can have an
important bearing on the outcome of individual contests and the course of a
season.
Player transgressions that result in red cards have been found to influence
match outcomes (Ritter et al. 1994).
A yellow card, which is often issued for a lesser offence, can influence
subsequent decision-making made by that player for the reminder of the
match.
Consequently, and if we assume players rationality, players will commit
major transgressions such as professional fouls either if the probability of
detection is low and / the player believes the foul will benefit the team.
OUTLINE
• Previous Literature
• Data
• Modelling Disciplinary Sanctions
• Empirical Results
PREVIOUS LITERATURE
Recent academic literature, which has focused on domestic
leagues in a number of major European football leagues, has found
evidence to suggest match officials are inconsistent in their decision-making
both in terms of the distribution of disciplinary sanctions and time added-on
at the end of contests. Some examples:
Disciplinary Sanctions
Buraimo et al. (2007) – English Premier League and German 1st Bundesliga.
Dawson et al. (2007) – English Premier League.
Downward and Jones (2007) – English FA Cup matches.
Time Added-On
Dohmen (2005), Sutter and Kocher (2004) – German 1st Bundesliga.
Garciano et al. (2005) – Spanish league
Scoppa (2007) – Italian league
Main finding: referee decision-making is biased.
CONTRIBUTION OF PRESENT STUDY
• This study looks at disciplinary sanctions in the context of European
cup competitions at club level (UEFA Cup and Champions League).
- Comparison of elimination tournaments
- Ability to identify possible nationality effects (evidence from
subjectively judged sports indicates this is important determinant
of decision-making)
• Very much a work in progress (comments / suggestions welcome!)
DATA
• Match based data provided by UEFA
• Full dataset includes over 13,000 observations over the period 1971-2007
• Sub-set of observations used in this study: 2002-03 to 2006-07
• n = 1720
- Champions League
- UEFA Cup
DATA
Cross-Tabulation of % Yellow Cards Issued to Home and Away Teams (by Competition)
Home
Yellow
Away Yellow
5
0
4.63
(3.78)
3.59
(3.84)
2.46
(2.45)
1.23
(0.63)
0.66
(0.25)
0 (0)
6
0 (0)
1
6.52
(6.10)
9.64
(9.88)
6.43
(7.05)
2.74
(3.65)
0.76
(0.76)
0.09
(0.25)
0 (0)
7
8
0 (0)
0 (0)
0 (0)
0 (0)
0
1
2
3
4
2
5.67
(5.35)
11.15
(7.17)
8.98
(8.75)
3.12
(4.47)
0.95
(1.83)
0.38
(0.25)
0.19
(0.13)
0 (0)
0 (0)
3
2.27
(2.52)
6.52
(5.98)
5.48
(6.23)
3.40
(3.52)
0.95
(0.88)
0.09
(0.13)
0 (0)
4
0.95
(0.94)
3.31
(2.83)
1.98
(2.83)
1.13
(2.14)
0.57
(0.44)
0.19
(0.06)
0.19
Total
5
0.76
(0.25)
0.38
(0.88)
1.13
(1.13)
0.38
(0.94)
0.19
(0.38)
0.28
(0.06)
0 (0)
6
0
(0.06)
0.19
(0.13)
0.09
(0.31)
0.09
(0.19)
0
(0.19)
0 (0)
7
0 (0)
8
0 (0)
0
(0.06)
0
(0.06)
0
(0.06)
0 (0)
0
(0.06)
0
(0.06)
0
(0.06)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
0.09
0 (0)
0 (0)
0 (0)
0 (0)
0 (0)
(0)
Total
12.57
26.18
30.43
18.81
8.32
3.12
0.57
0
0
(10.95) (27.69) (27.94) (19.26) (9.25) (3.65) (0.88) (0.19) (0.19)
Note: Main figures Champions League data, figures in parentheses UEFA Cup data
20.79
(19.01)
34.78
(30.84)
26.65
(28.89)
12.10
(15.67)
4.16
(4.72)
1.04
(0.76)
0.38
(0.13)
0 (0)
0.09
(0)
100.00
DATA
Cross-Tabulation of Red Cards Issued to Home and Away Teams (by Competition)
Champions League
Red Card Away
Red
Card
Home
0
1
2
3
Total
0
82.89
5.10
0.19
0
88.19
1
9.74
0.85
0.28
0
10.87
2
0.95
0
0
0
0.95
UEFA Cup
Red Card Away
Total
93.57
5.95
0.47
0
100.00
0
81.31
6.54
0.13
0.06
88.04
1
9.13
1.57
0.25
0
10.95
2
0.82
0.13
0.06
0
1.01
Total
91.25
8.24
0.44
0.06
100.00
MODELLING DISCIPLINARY SANCTIONS
A number of different modelling approaches have been used in the literature:
• Univariate OLS and Poisson (Witt, 2005)
• Bivariate Poisson and Negative Binomial Models (Dawson et al., 2007)
• Bivariate Probit Models (Buraimo et al., 2007)
It is also possible to model count data by discrete choice methods that
recognise the sequential nature of the data (Cameron and Trivedi, 2005).
One such candidate is the ordered probit model:
y h*  Xβ h  ε h
y a*  Xβ a  ε a
MODELLING DISCIPLINARY SANCTIONS
The dependent variable is constructed as the total number of disciplinary
“points” incurred by the home team and away team in each match.
One point awarded for each yellow card and two points for a red card.
The dependent variables in the home and away equations are then recoded as follows:
0 if

1 if
2 if

yh  
3 if
4 if

5 if
y h*  0
y h*  1
y h*  2
y h*  3
y h*  4
y h*  5
0 if

1 if
2 if

ya  
3 if
4 if

5 if
y a*  0
y a*  1
y a*  2
y a*  3
y a*  4
y a*  5
MODELLING DISCIPLINARY SANCTIONS
Variables
Difference in Team
Coefficient Index
Stage of
Competition
Champions League
Attendance / 1000
Attendance to
Capacity Ratio
Track
Fence
Home Cards
Away Cards
Referee Nationality
Effects
Home Team
Nationality Effects
Away Team
Nationality Effects
Big Five Home
Big Five Away
Elite Referees
Definition
Team Coefficient Index of Home Team Minus Team
Coefficient Index of the Away Team.
= 1 if Round of 32 onwards (Round of 16 in the case of the
Champions League), 0 otherwise
= 1 if Champions League Match, 0 otherwise (i.e. UEFA Cup
match)
Attendance scaled by 1000
Attendance divided stadium capacity
= 1 if stadium has a running track, 0 otherwise
= 1 if stadium has fencing, 0 otherwise
Total disciplinary “points” issued to the home team
Total disciplinary “points” issued to the away team
Dummy Variables which represent referee nationalities
(minimum of 70 observations required for inclusion).
Dummy Variables which represent home team nationality
(minimum of 70 observations required for inclusion).
Dummy Variables which represent the away team nationality
(minimum of 70 observations required for inclusion).
=1 if home team from one of the “big five” leaguesa, 0
otherwise
=1 if away team from one of the “big five” leagues a, 0
otherwise.
=1 if match official officiated at the 2004 European
Championship, 0 otherwise.
Notes: a Team from England, France, Germany, Italy or Spain.
EMPIRICAL RESULTS
Variables
Difference in
Team Coefficient
Index
Stage of
Competition
Champions
League
Attendance / 1000
Attendance to
Capacity Ratio
Track
Fence
Referee
Nationality Fixed
Effects
Rho ()
Thresholds
Cut1
Cut2
Cut3
Cut4
Cut5
Pseudo-R2
LR test (joint
significance of
covariates)
LR test
(independent
equations)
N
Univariate Ordered Probit
Home Team
Away Team
Bivariate Ordered Probit
Home Team
Away Team
-0.008** (0.003)
0.002 (0.003)
-0.008** (0.003)
0.002 (0.003)
0.113* (0.060)
0.150** (0.060)
0.114* (0.060)
0.149** (0.060)
-0.199*** (0.069)
-0.187*** (0.068)
-0.199*** (0.069)
-0.188*** (0.068)
0.0003 (0.002)
0.072 (0.093)
0.002 (0.002)
0.232** (0.093)
0.0003 (0.002)
0.073 (0.093)
0.002 (0.002)
0.231** (0.093)
0.185*** (0.059)
0.101 (0.066)
INCLUDED
0.019 (0.059)
-0.013 (0.065)
INCLUDED
0.185*** (0.059)
0.101 (0.066)
INCLUDED
0.019 (0.059)
-0.012 (0.065)
INCLUDED
0.241*** (0.025)
-0.924*** (0.077)
-0.031 (0.075)
0.727*** (0.076)
1.366*** (0.081)
1.847*** (0.090)
0.013
69.68
-1.174*** (0.079)
-0.316*** (0.075)
0.384*** (0.075)
0.963*** (0.077)
1.484*** (0.081)
0.013
75.63
-0.924*** (0.077)
-0.028 (0.075)
0.728*** (0.076)
1.363*** (0.081)
1.840*** (0.090)
-1.176*** (0.079)
-0.314*** (0.075)
0.388*** (0.075)
0.964*** (0.077)
1.481*** (0.081)
69.57
87.34
1720
1720
1720
EMPIRICAL RESULTS
Variables
Home Team
Difference in Team
Coefficient Index
Stage of
Competition
Champions League
Attendance / 1000
Attendance to
Capacity Ratio
Track
Fence
Big Five Home
Big Five Away
Big Five Home x
Big Five Away
Elite Referee
Away Team
Difference in Team
Coefficient Index
Stage of
Competition
Champions League
Attendance / 1000
Attendance to
Capacity Ratio
Track
Fence
Big Five Home
Big Five Away
Big Five Home x
Big Five Away
Elite Referee
Referee Nationality
Fixed Effects
Team Nationality
Fixed Effects
Rho ()
LR test (joint
significance of
covariates)
LR test
(independent
equations)
N
(1)
Bivariate Ordered Probit Estimates
(2)
(3)
(4)
0.0002 (0.004)
-0.0001 (0.003)
0.001 (0.004)
0.0005 (0.004)
0.080 (0.065)
0.098 (0.063)
0.062 (0.066)
0.047 (0.066)
-0.217*** (0.074)
0.0002 (0.002)
0.150 (0.100)
-0.213*** (0.070)
0.001 (0.002)
0.040 (0.094)
-0.259** (0.076)
-0.0002 (0.002)
0.158 (0.10)
-0.272*** (0.076)
-0.0002 (0.002)
0.166* (0.100)
0.142** (0.065)
0.022 (0.072)
0.177*** (0.059)
0.096 (0.066)
-0.322*** (0.075)
0.089 (0.074)
0.296*** (0.103)
0.145*** (0.065)
0.021 (0.072)
0.143** (0.065)
0.022 (0.072)
-0.258** (0.122)
0.314*** (0.120)
0.290*** (0.105)
0.190** (0.087)
0.173** (0.087)
0.0013 (0.004)
0.0001 (0.003)
0.002 (0.004)
0.001 (0.004)
0.121* (0.064)
0.138** (0.063)
0.110* (0.065)
0.097 (0.065)
-0.207*** (0.073)
0.003 (0.002)
0.273*** (0.099)
-0.198*** (0.069)
0.002 (0.002)
0.251*** (0.093)
-0.233*** (0.075)
0.002 (0.002)
0.278*** (0.099)
-0.244*** (0.075)
0.002 (0.002)
0.285*** (0.099)
0.064 (0.064)
0.022 (0.071)
0.017 (0.059)
-0.007 (0.066)
-0.053 (0.074)
-0.169** (0.074)
0.255** (0.102)
0.066 (0.064)
0.021 (0.071)
0.064 (0.064)
0.022 (0.071)
-0.053 (0.121)
-0.012 (0.120)
0.272*** (0.104)
0.116 (0.086)
0.100 (0.025)
INCLUDED
INCLUDED
INCLUDED
INCLUDED
INCLUDED
NOT INCLUDED
INCLUDED
INCLUDED
0.254*** (0.025)
167.30
0.244*** (0.025)
102.67
96.40
1720
89.92
1720
0.252*** (0.025)
171.88
0.249*** (0.025)
179.30
95.31
92.84
1720
1720
EMPIRICAL RESULTS
 The average number of disciplinary points awarded to both home and
away teams in the Champions League is lower compared to the UEFA Cup.
 A strong home team (relative to the away team) incurs less disciplinary points.
 Home teams from the “big five” leagues incur less disciplinary points.
 Contests which involve both teams from “big five” leagues generate more
disciplinary points.
 The presence of a running track has the effect of increasing the number of
disciplinary points issued to the home team.
 The number of disciplinary points awarded to the away team is higher the
fuller the stadium.
 Elite referees issue more disciplinary points to the home team.
EMPIRICAL RESULTS
Referee Nationality Effects
Nationality
Belgium
Danish
German
Greek
Norwegian
Dutch
Portuguese
Russian
Scottish
Spanish
Swedish
Home Equation
Negative
N/S
Negative
Positive
Negative
Negative
N/S
Negative
Negative
N/S
Negative
Away Equation
Negative
Negative
N/S
Positive
N/S
Negative
Positive
Negative
N/S
Negative
Negative
Difficult to establish the precise interplay between referee nationality, team
nationality and reputation…
CONCLUDING REMARKS
Consistent with previous studies we find that referees do issue more
cards to the away team.
This is partly a reflection of the home advantage and the difference in
quality of the two teams competing. However stadium specific
characteristics are also important. Furthermore there are significant
differences across the two competitions analysed in this study.
Given the apparent variation in referee behaviour across nationalities
further work should explore why such variation occurs. For example to
what extent does reputation of teams / nations matter? Are referees with
more experience less biased? Are there important differences between
the decision-making of referees in domestic competitions compared to
European competitions?