A century ago, soccer games were plenty of violence and it was typical that players came to blows and the crowd booed the referee. George Orwell, famous English writer, stated in 1945 regarding this behavior that “Serious sport has nothing to do with fair play. It is bound up with hatred, jealousy, boastfulness, disregard of all rules, and sadistic pleasure in witnessing violence: in other words, it is war without the shooting.”1.

Nowadays the rules have changed and referees are respected and empowered to penalize this conduct. Misconduct, infringement, offence, commonly called soccer fouls are violation of the conduct of the game and there are ways to punish offenders.2

Violent actions could evolve into a disadvantage that impacts on the result, and be aware of their repercussion could be important to soccer coaches, when deciding their tactics to win, choosing the proper players and training them for a better performance.

This conception of violence conforms with what Bredemeier described as aggressive behavior in sport: “The intentional initiation of violent and or injurious behavior. ‘Violent’ means any physical, verbal or nonverbal offense, while ‘injurious behaviors’ stand for any harmful intentions or actions.”3

Designed by skadyfernix / Freepik

With the aim of illustrating the effect of violence on this sport, this report relies on historical statistics and results from the first-tier leagues in the four most relevant countries in Europe: Spain, Germany, England and Italy, according to UEFA associations’ club coefficients rankings4, with data from seasons 05-06 to 15-16 using the datasets provided by the site Football-Data.co.uk.

Outlined as a research based on archived records, this is an observational study, from which we will merely observe the available information without choice to decide about any experimental designing, and all the cases are retrospective, since they have been gathered through compiling recorded files which do not directly interfere with how data arise.

Since the severity and punishment are less for a simple foul than for a yellow card, by which a player is cautioned, and the same occurs for a yellow card versus a red card, which entails a player dismissal from the game, a misconduct will be weighted depending on its nature.

Hence the amount of violence for a team during a soccer game is estimated as the weighted sum of the violent actions for a team during a soccer game where yellow cards count as 5 fouls and red cards as 20 fouls. During a single match, this represents a quantity between 0 and 125, being the typical value 26.

As we can see in the box plot (Figure 1) that displays the effect of violence amount on the result for a single game, there is a general similarity and only a slight variation when the match was won and the amount of violence lower. Considering the possible points (0 for losing, 1 for drawing and 3 for winning) instead of the categorical result, the correlation between the amount of violence and points is -0.1389015, confirming the presence of a very weak negative linear relationship between them.

Figure 1. Effect of the amount of violence on the result of a single game

Figure 1. Effect of the amount of violence on the result of a single game

However, the behavior of a team in the course of a season is not consigned to a separate match, but to a set of matches comprising the whole season, and the overall degree of misconduct is a consequence of the tactics applied by the coach and the players. In order to acquire a perspective from that point of view, the data need to be aggregated at a different level.

We can aggregate the data and look at the average amount of violence and the average of points achieved by a team for a given season. Similarly, we can group the data by country in order to be able to distinguish the effect for each competition. Arranged this way, the data make more sense for the sake of this analysis.

On the other hand, as reported by some players who have participated in multiple leagues as well as other sources, there is a potential bias on the country because referees do not follow the same criteria to penalize infringements, and they are more permissive in some countries while more severe in others. To deal with this divergence, the amount of violence will be standardized with respect to the country.

Chelsea striker Diego Costa is one of these players. He played in three different competitions: Portugal, Spain and England, and he recently complained about the differences in England saying: “They go in strongly but the referees won’t give it as a foul - in Spain, they would.”5 Accordingly, the book Science and Football6 suggests that, probably suited to the player’s acceptance level, English referees seem to allow more physical play than their Italian colleagues, and therefore professional soccer in England has fewer fouls than in Italy.

Certainly, we can see in the Figure 2 where the data are grouped but still biased, that the amount of violence appears to be inferior in England and Germany as well as superior in Spain and Italy but, after normalizing the data, the Figure 3 shows that there is not a significant difference among the countries. Both scatter plots depict a point for each team throughout a season, while the color denotes the country.

Figure 2. Biased effect of violence on the average of points awarded to a team during a season

Figure 2. Biased effect of violence on the average of points awarded to a team during a season

Figure 3. Unbiased effect of violence on the average of points awarded to a team during a season

Figure 3. Unbiased effect of violence on the average of points awarded to a team during a season

Regardless the country, the general trend in the Figure 3 is that the more violent teams for a season appear with a lower average of points.

Once the data have been grouped by country, team and season and the independent variable has been normalized with respect to the country, the resulting linear correlation is -0.3925907, about -0.4, stronger than the -0.1492187 estimated before aggregating and normalizing.

As the standardized average amount of violence is centered at 0, we can classify the teams according to their tactics into two categories: violent if their standardized amount of violence is greater than 0, and otherwise moderate. The box plot in Figure 4 emphasizes that the average of points seems to be generally higher for moderate tactics and lower for a violent tactics.

Figure 4. Effect of the type of tactics on the average of points

Figure 4. Effect of the type of tactics on the average of points

In consonance with this classification, we can consider two samples: the aggregated cases by country, team and season applying violent tactics and the cases employing moderate tactics, independent of one-another, corresponding to the aggregation of disjoint sets of soccer matches, so the data are not paired, condition under which we can conduct the following hypothesis test for the means:

\(H_0 : \mu_{violent} = \mu_{moderate}\) \(H_a : \mu_{violent} < \mu_{moderate}\)

Before proceeding, we need to check that the conditions for inference are met. We can consider our available data to be randomly sampled because there is nothing to think of a possible non-randomness at the time of the process of data collection.

As the whole population would be the aggregated data from matches by country, season and team from the four main European leagues all over the time, and our samples comprise matches since the season 05-06, we are examining two simple random samples from less than 10% of the population, each sample contains at least 30 observations, and neither distribution is strongly skewed according to their sizes and histograms (see Figure 5), the observations are independent and we can safely conclude that the difference of sample means can be modeled using a normal distribution.

Figure 5. Distribution of the average points for moderate and violent tactics

Figure 5. Distribution of the average points for moderate and violent tactics

Assuming that the null hypothesis is true, the mean of the difference is 0. Since the point estimate is nearly normal, we can find the upper tail using the Z-score and normal probability table.

The Z-score obtained is (point estimate - mean) / standard error = (0.2527515 - 0) / 0.0282981 = 8.9317401, so large and it is not even in the table, which ensures the p-value, that corresponds to one tail, will be 0.0002 or smaller7.

Because the p-value is less than a significance level of 0.05, we reject the null hypothesis in favor of the alternative hypothesis. We have found convincing evidence that, in the course of a European first-tier soccer league, the quantity of points awarded to teams with moderate tactics is, in average, greater than to teams with violent tactics.

These data cannot establish causality, but only correlation as long as this is an observational study. Nonetheless, they provide an answer to how violence effects this sport, concluding that violent tactics are generally counter-productive for a team during a season, at least in the European first-tier soccer.

There are many circumstances which involve applying a violent tactic for a team. One of them is when a coach highly values a “win at all costs” philosophy8 and players perceive it and become more agressive and willing to commit illegal actions.

Another scenario is when the coach promotes the use of tactical fouls with the mere purpose of aborting the opponent play. The third rule of the official code for soccer9 states that this kind of foul should be avoided as a principle.

The conclusion of this study is that statistics reveal that fouls and cards in European first-tier soccer leagues are commonly associated with worse results throughout a season, therefore coaches and players should avoid them and focus on further approaches to improve their team performance. Those who have studied the game10 conclude that, apart from natural aptitude and ability, the rest of the ability of soccer players and teams is down to their own hard work. These qualities include knowledge and mastery of the game, motivation, mental control and the development of skills such as shots, feints and dummies (tactics used by a player to trick an opponent) that make possible to beat a rival avoiding the use of violence.

References

Bangsbo, Jens, Thomas Reilly, and A. Mark Williams. 2014. Science and Football III. Routledge.

Bredemeier, Brenda Jo. 1983. Athletic Aggression: A Moral Concern. Springer-Verlag.

Diez, David M., Christopher D. Barr, and Mine Çetinkaya-Rundel. 2014. OpenIntro Statistics.

Football Bible. 2014. Soccer Foul and Misconduct Explained. http://www.football-bible.com/soccer-info/soccer-foul-rules-misconduct-explained.html/; Football Bible Company.

Orwell, George. 1945. “The Sporting Spirit.” Tribune Magazine, GB, London.

Ruiz, Laureano. 2003. The Spanish Soccer Coaching Bible: Youth & Club. Volume One. Reedswain Inc.

Schum, Tim. 1996. Coaching Soccer. Masters Press.

The42 Team. 2016. “Diego Costa on the Differences Between Playing in England and Spain.” http://www.the42.ie/diego-costa-differences-playing-england-spain-2013833-Mar2015/; The42 Team, Ireland.

Ueberroth, Peter V., and Anita L. DeFrantz. 2012. Soccer Coaching Manual. LA84 Foundation.

UEFA, Union of European Football Associations. 2016. “UEFA Associations’ Club Coefficients Rankings.” http://www.uefa.com/memberassociations/uefarankings/country.

  1. Orwell (1945)↩

  2. Football Bible (2014)↩

  3. Bredemeier (1983)↩

  4. UEFA, Union of European Football Associations (2016)↩

  5. The42 Team (2016)↩

  6. Bangsbo, Reilly, and Williams (2014)↩

  7. Reasoning when the value is not in the normal probability table as exposed in Diez, Barr, and Çetinkaya-Rundel (2014)↩

  8. Schum (1996)↩

  9. Ueberroth and DeFrantz (2012)↩

  10. Ruiz (2003)↩