The economic valuation of soccer players: what role does
behavior play?
Abstract:
In this paper, the relationship between soccer players’ behavior and their economic valuation is studied. Here, it is assumed that a player’s valuation depends on both on- and off-pitch performance. Using a sample of players playing in the Dutch Eredivisie and Jupiler League, I find leadership behavior to positively influence a player’s value, whereas being ‘too loyal to and dependent of the team’ may harm individual performance and corresponding value. I also find players to generally place high emphasis on performance oriented striving, working hard, group atmosphere and being emotionally stable. These findings do not enable to distinguish between high-/low-valued players, however. Furthermore, the results also show that it remains difficult to obtain a complete overview of all factors influencing a soccer player’s value. Further research is still needed to improve the knowledge in this field of interest.
Name: Thomas van Brunschot Student number: 10590803
University: University of Amsterdam
Faculty: Faculty of Economics and Business Track: Managerial Economics and Strategy Supervisor: Sander Onderstal
Statement of Originality
This document is written by Student Thomas van Brunschot who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text
and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of contents
1. Introduction……….. 4
2. Literature review……….. 7
2.1. Theories of superstar formation……….. 7
2.2. Superstar effects in soccer……… 8
2.3. Personality traits athletes………. 9
2.4. Behavior athletes……….. 13
2.5. Contribution to existing studies……….. 15
3. Methodology………. 15
3.1. Data set……….. 16
3.2. Method………...18
4. Results………... 19
5. Discussion and conclusion……… 28
6. Appendix……… 30
1. Introduction
What factors determine the economic value of professional soccer players? And what distinguishes the most successful players from the large pool of competition below them? Is it simply the exceptional talent, or are factors like popularity and personality also decisive? Soccer analysts as well as academic researchers have tried to find answers to these questions and with the still ever increasing popularity of soccer and the astronomical amounts of money that go along with it, it is therefore interesting to further deepen the knowledge concerning the valuation of its main assets: the players. In 1981, Rosen proposes a theory of superstar formation. Herein, he states that talent, which translates into higher so-called on-pitch performance for professional athletes, is the main driver for the value and salary differences. On the other hand, Adler (1985) argues that superstars might just as well appear among equally talented performers due to the positive network externalities of popularity. Simply put (and further elaborated upon in the next section), Adler meant that besides on-pitch performance, popularity also differentiates players from their competitors. Over the years, both theories have been tested for soccer players (Brandes, Franck, & Nüesch, 2008; Franck, & Nüesch, 2008; Franck, & Nüesch, 2012; Lehmann, & Schulze, 2008). In line with Adler’s theory, Franck and Nüesch (2008) find that both performance as well as popularity indicators have a positive and significant effect on players’ market values. However, Lehmann and Schulze (2008) find that neither performance indicators nor popularity translate into disproportionate salary differences. At this stage, the overall results thus remained ambiguous.
In 2016, Korzynski and Paniagua stress the increasingly important role of social media and thus online popularity for athletes. By reaching an online audience, individuals cannot only build a personal brand, they also create more value for their clubs through marketing and advertisements (p. 186), and also through higher gate receipts, sponsors’ contributions and merchandise sales (Lehmann, & Schulze, 2008, p. 9). Complementary to the previously mentioned on-pitch performance, this so-called off-pitch performance has gained an increasing influence over the last decade and therefore needs to be examined more closely.
Further research also shows that, besides physical and demographic characteristics, an athlete’s personality traits may also influence his/her on-pitch performance, influencing his/her value altogether. ‘’For two athletes with a similar exercise capability, personality is likely to make a difference in their long-term achievements’’ (Malinauskas et al., 2014, p. 145). For example, in a personality study of high/low performing soccer players, Panda and Bisivas (1989) showed that high performing players had significantly higher levels of extraversion, confidence, anxiety, emotionality and aggression as compared to low performing players. In line with results found by Panda and Bisivas, Williams and Parkin (1980), Kirkcaldy (1982) and Egloff and Gruhn (1996) also found high-level athletes to be more extraverted. However, they also show higher levels of emotional stability among high-level athletes, thereby contradicting the findings of Panda and Bisivas. Allen et al. (2011) also found high-level athletes to have lower levels of neuroticism (high emotional stability), but also higher levels of agreeableness and conscientiousness than lower-level athletes. In conclusion, current studies have not yet led to a clear framework of the relationship between personality, performance and corresponding value. Possible explanations for these inconsistent results are the absence of solid theoretical backgrounds to support investigations (Beauchamp, 2007, p.28), inconsistency in terms of control groups used (e.g. athletes vs. non-athletes, high-level athletes vs. low-level athletes) and the failure to distinguish between different sports (Morris, 200, p. 719).
Question is whether personality as such should be decisive in a clear framework. As such, personality does not necessarily show. In his book Performance: the secrets of successful behavior (2006), Robin Stuart-Kotze questions this ‘’validity of personality traits in terms of measuring performance’’ in a convincing, clarifying way. ‘’Ability and capability are not about traits, personality or genes - they are about behavior. Unlike genetics or personality, behavior can be described, observed, measured and changed. As a result, both ability and capability can be increased. It’s what you do that matters, not what you are or who you are.’’
Especially nowadays with social media being of such influence, athletes need to be careful how they profile themselves both on- and off-pitch. Earlier mentioned studies state the drivers for both value and salary differences among soccer players. Since
player salaries need not necessarily equal their value –that depends on the way their salary is being negotiated ultimately- the scope of this thesis is not focused on player salaries. Furthermore, as past studies in the field of personality-performance comparisons did not lead to consistent findings, I ultimately decided to look at behavior rather than at personality traits in order to try and explain soccer players’ performance. In this thesis, I study the effect of behavior on performance (on- and off-pitch) and corresponding value, leading to the following research question:
How does soccer players’ behavior relate to their on-pitch performance and corresponding value and can this behavior also explain their off-pitch popularity?
As the measurement indices used for the analysis have not, or to a limited extent, been used in scientific studies, this study serves the purpose of a fishing expedition. Without setting hypotheses or expectations upfront, the aim of this study is to find valuable information that adds to the existing knowledge in this field of research. Based on a sample of 45 players playing in the highest Dutch soccer leagues, I find leadership behavior (both faith/confidence in own leadership capabilities as well as the need to lead other players) to have the strongest impact on one’s on-pitch performance and corresponding value. I also find all players to generally have a high focus on performance oriented striving, cooperation and being emotionally stable. However, these findings are not capable of distinguishing between high/low performing players, but can be considered as general behavior aspects that professional soccer players focus on. Unfortunately, an in-depth analysis trying to link behavior to specific on-pitch factors rather than overall performance did not enable me to draw valid conclusions due to limited data and a small sample size. The same holds for the off-pitch analysis, as crucial parts of the data needed proved too difficult to gather, due to privacy regulations and cost concerns. Although a concise version of the analysis is still taken into account in the results section, I do not draw any conclusions based on the available data.
The structure of this thesis is as follows: the next section gives an overview of the most important related literature as well as my own contribution compared to the existing studies. The third section covers the methodology. Section 4 covers the
results. Finally, section 5 gives a brief summary of the main findings and conclusions derived from the study, a discussion of possible limitations and some possible steps for future studies.
2. Literature review
This section provides an overview of the existing literature on the valuation of soccer players. In sections 2.1 and 2.2, I describe the two theories of superstar formation and the attempts to put these theories to practice, respectively. Section 2.3 covers the existing literature on athletes’ personality traits and how this relates to their performance. In section 2.4, I give the theoretical approach for this thesis and section 2.5 concludes with my contribution to the existing studies.
2.1 Theories of superstar formation
The existing literature covers basically two competing theories of superstar formation, as proposed by Rosen (1981) and Adler (1985). These theories state that, for the emergence of superstars, markets with large economies of scale on the supply side and high demand are needed (Franck, & Nüesch, 2008, p. 146). In soccer economies of scale arise, as the cost of production is mostly independent of the size of the audience. Most costs concerning soccer matches are made up-front and thus decrease with the number of consumers. Also, the consumption of one does not come at the cost of the other (Franck, & Nüesch, 2012, p. 203) and there are no issues of free riding due to the possibility of excluding non-paying customers (Franck, & Nüesch, 2008, p. 146), both leading to larger economies of scale. The two theories agree on these supply side assumptions. However, they differ in their explanation of the demand for superstar services. Sherwin Rosen argues that lesser talent often is an imperfect substitute for greater talent (1981, p. 846): ‘’hearing a succession of mediocre singers does not add up to a single outstanding performance’’. According to Rosen, it is thus the exceptional talent and outstanding performance of a professional sportsman that makes him valuable and attracts fans. Moshe Adler adds that besides talent, network externalities of popularity might form a second reason for the value and earnings differences among individuals of various professions (1985, p. 208). These externalities increase with the amount of star-specific knowledge consumers
acquire about an individual, also known as ‘consumption capital’, as proposed by Stigler & Becker (1977). This knowledge comes from consumption in the past (for example watching a match), discussions about a player with other knowledgeable individuals and reading about a player in newspapers, magazines and on the Internet (Franck, & Nüesch, 2012, p. 204); the more popular the player, the easier to gain knowledge.
Whereas Adler (1985) first considered luck to be the cause of superstar emergence among equally talented artists, he later takes back this consideration and acknowledges that superstars consciously use publicity (e.g. appearances on talk shows, but perhaps nowadays to a larger extent social media platforms as well) to increase their popularity (2006). Adler thus adds to Rosen’s theory the possibility of popularity influencing the probability of becoming a superstar in case there is no clear superiority of one artist over another. The next subsection focuses on currently existing studies concerning superstar formation in different top soccer leagues in Europe.
2.2 Superstar effects in soccer
Different researchers have tried to put the theories as proposed by Rosen and Adler to practice. For example, Franck and Nüesch investigate the emergence of superstar formation using data from the German Bundesliga during the 2004-2005 season (2008, p. 150). They use data on market values, individual player performance indicators and publicity (popularity indicator) to differentiate between Rosen’s and Adler’s theory. By running quantile regressions they find that both individual performance as well as popularity indicators have a positive and significant effect on players’ market values (pp. 156-158), which is in line with Adler’s Theory.
On the other hand, Lehmann and Schulze (2008) find contrary evidence in their article. They seek to explain differences in salaries among players, again using a dataset of players active in the German Bundesliga, during the seasons 1998-1999 and 1999-2000. They also run a quantile regression, claiming it to be the appropriate technique because ‘a strongly convex gradient may show up only at the top end of the income distribution and not for average players’ (p. 6). Using performance indicators
(e.g. goals, assists, shots and successful tackles) and media presence (number of ‘hits’ of a player’s name in the online version of the Kicker sport magazine during the season), they find that neither differences in on-pitch performance nor in media presence translate into disproportionate salary differences (pp. 9-10). A possible limitation is the relatively scarce number of performance indicators. Furthermore, using ‘hits’ of only one magazine to determine popularity also raises questions in terms of validity of the results. Finally, as mentioned previously, salaries need not necessarily equal a player’s value, therefore explaining the contradictory results. This study could add value to the existing literature through improvement on these parts. Finally, Korzynski and Paniagua (2016, pp. 186-188) stress the importance of the three powers of social media as an explanation for differences in value; the power of informing (through sharing private life details), the power of interacting (engaging in online discussions) and the power of inspiring (through merchandising advertisements or sharing photos from soccer matches or events in order to motivate fans to attend). As they say: ‘’Some gifted players are undermined by weak media exposure while some less talented players who actively engage in social media and attract fans in millions benefit from exorbitant contracts’’. By use of a qualitative comparative analysis on a sample of 95 top players in Europe, Korzynski and Paniagua find that young players (aged below 25) may obtain valuable contracts due to high social media activity even in the absence of outstanding performance (p. 190). Toward the end of their careers, however, players have to perform well both on- and off-pitch in order to maximize market values and acquire better contracts.
Existing research thus provides insights that both support and contradict the theories of superstar formation. Most studies conclude that in soccer it remains difficult to give a clear and full overview of what determines a player’s value or salary. On the other hand, this leaves room for improvement and perhaps new insights into this topic, which this study also seeks for.
2.3 Personality traits athletes
The previous subsection demonstrated a body of research trying to explain the effects of talent and popularity on value generation. A second body of research tries to
examine the role of personality traits in the process of performance and value generation. Researchers have been interested in finding relationships between personality and sport performance for a long time. Many of the existing studies using general personality tests were inspired by a classic study by Kane (1966). He studied the personality of a sample of English amateur and professional soccer players. Using the 16 Personality Factor questionnaire, he found that professional players were more urgent and controlled, but less dominant and adventurous compared to the general population. However, the low significance levels let him to conclude that there was little to distinguish between professional soccer players and the general population in terms of personality.
Meuris et al. (1993) also used the 16 Personality Factor questionnaire to study the personality of 1140 Belgian athletes from 11 different sports, among which 225 male indoor soccer players. In contrary to results found by Kane, the soccer players were found to be more aloof, emotionally stable, tough and realistic. These dissimilarities might be explained through cultural differences and/or differences between field versus indoor soccer.
Allen et al. (2011) tried to explain personality differences among athletes through the NEO-Five Factor Inventory (NEO-FFI; Costa & McCrae, 1992). The traits described by this test are extraversion, agreeableness, conscientiousness, neuroticism and openness to experience. Extraversion is characterized by being sociable, action oriented, talkative and outgoing (Costa & McCrae, 1992). Agreeableness means being cooperative and collegial rather than suspicious and antagonistic towards others. Conscientiousness involves being organized, dependable, and self-disciplined as well as having a preference for planned rather than spontaneous behavior. It can also be described as an achievement orientation (Barrick & Mount, 1991). Neuroticism refers to the degree of emotional stability. High levels of neuroticism associate with being easily triggered by emotions such as insecurity, anxiety, depression and anger. At last, openness to experience reflects the degree of creativity, curiosity and preference for novelty and variety of an individual. Allen et al. (2011) used the NEO-Five Factor Inventory to assess the personality traits of 253 athletes competing in 34 different sports (p. 843). They find that team sport athletes show higher levels of extraversion and neuroticism, but lower levels of conscientiousness and openness to new
experiences (pp. 844-845). Results regarding extraversion among team sport athletes were also found by Eagleton, McKelvie and De Man (2007) and Malinauskas et al. (2014). Although the study by Allen et al. lacked theoretical underpinnings, Eagleton, McKelvie and De Man referred to Eysenck’s biological theory to support the influence of ‘the two basic biological traits of personality’ (neuroticism and extraversion) on athletic performance (p. 266). Eysenck stated that the extent to which an individual seeks sensation depends on one’s level of cortical arousal (1982, p. 3). Low levels of cortical arousal translate to high sensation seeking, which are found in sport and exercise activity. As Extraverts have low levels of cortical arousal, Eysenck concluded that athletes are more likely to be extraverted than introverted (pp. 4-5). Furthermore, he stated that extraverted and socially stable individuals seek out social situations involving intimacy and competition. Introverts and neurotics on the other hand avoid these situations making it more likely that team sport athletes have higher levels of extraversion and lower levels of neuroticism compared to individual sports (p. 6).
Inspired by Eysenck, Panda and Bisivas (1989) used an early version of the Eysenck Personality Inventory to compare personality traits of 50 ‘high-achieving’ and 50 ‘low-achieving’ Indian soccer players. They find the ‘high-achieving’ soccer players to be significantly more extraverted, confident, anxious, emotional and aggressive than the ‘low-achieving’ group. A possible limitation of the study by Panda and Bisivas and perhaps an explanation for differences found in levels of neuroticism compared to other articles is the standard of soccer in India. The level differences between Indian and European professional soccer could create difficulties in terms of generalizing results and comparisons to studies of elite European soccer players. Allen et al. (2011) also distinguished between ‘high-level and low-level’ athletes. High-level athletes were described as ‘competing at national and international level’, whereas low-level athletes ‘compete at university and club level’ (p. 845). They found that high-level athletes had lower levels of neuroticism and higher levels of conscientiousness and agreeableness than lower-level athletes. This is partly in line with results found by Williams and Parkin (1980), Kirkcaldy (1982) and Egloff and Gruhn (1996). They also demonstrated high-level athletes to be more emotionally stable (low neuroticism) but also more extraverted compared to lower-level athletes.
A possible explanation for these differences and perhaps also a limitation of the study by Allen et al. is the wide range of sports included (p. 849) resulting in a heterogeneous sample. As this study only consists of soccer players playing in a single national league, these problems are limited.
A final extension to the current literature relates to the composition and compatibility of individual personality traits and their relationship to group cohesion (Beauchamp et al., 2007, p. 28). Group cohesion, consisting of social cohesion (orientation toward creating and maintaining social relationships within a group) and task cohesion (understanding of the group’s instrumental objectives and how to achieve them), is found to be influential for a sport team’s functioning and success (Tziner et al., 2003). Barrick et al. (1998) found that higher levels of extraversion and emotional stability (low neuroticism) translated to higher social cohesion. This makes sense, as more outgoing and emotionally stable individuals are more likely to interact frequently (Beauchamp et al., 2007, p. 29). Barrick et al. did not address task cohesion; van Vianen and de Dreu (2001) however did in their study. Besides finding similar results regarding social cohesion, they find that higher levels of agreeableness and conscientiousness could explain greater task cohesion.
As to date, many studies have been conducted in order to try and find patterns in athletes’ personality profiles that add to their performance. High levels of extraversion and low levels of neuroticism are the most consistently recurring traits among soccer players in the current literature. However, various personality tests resulting in very mixed and sometimes contradictory outcomes have led researchers to conclude that this topic still needs more research in order to improve current knowledge. One common limitation, as described by Beauchamp et al. (2007), is the fact that some researchers have tried to predict individual athletic performance through personality traits despite the absence of a solid theoretical background to support their investigations (p. 28). Also, comparing athletes to non-athletes might raise questions in terms of validity of the results, as it does not answer the question ‘what makes a great athlete’? Furthermore, Morris adds that current research is still too fragmented in terms of methods used and that small sample sizes, variable definitions of ‘top-class’ athletes and the failure to distinguish between different sports could explain at least part of the inconsistent results (2000, p. 719).
2.4 Behavior athletes
As shown in the previous subsection, past methods and personality tests used did not always result in consistent findings. Therefore, this study focuses on a method that gained more attention over the last decade, namely the more tangible, measurable behavior of soccer players, rather than looking at personality traits as such. Whereas traits generally put an individual into a pigeonhole, behavior only describes how one prefers to behave according to a specific situation. With proper coaching, behavior may change over time, whereas traits are rigid. As the earlier quoted Stuart-kotze stresses, ‘’a critical difference between behavior and personality is that your personality is essentially fixed at an early age and after that you can’t really change it (2006, p. 4).’’ In an environment inherent to continuous and rapid changes like that of top sport, analyzing behavior instead of personality traits is likely to be more valuable and therefore used as a guideline in this study.
In this subsection, I discuss the Personal Identity Test (PIT) approach, which is used to assess the behavior of the players included in this study. Inspired by ‘the big five’ and Eysenck’s personality theory, this approach gives insights into the behavior of a soccer player through seven scales each consisting of two subscales. The seven scales are leadership, conflict management, cooperation, flexibility, emotional stability, performance oriented striving and precision. Below, a brief explanation of all subscales is given. More specific details regarding high/low scores on all subscales can be found in table 6 in the appendix.
The first scale, leadership, is divided into the subscales belief in leadership capabilities and the need to lead others. The first subscale reflects the extent to which someone has faith in his own leadership capabilities (note: this does not mean that someone actually possesses the skills necessary to fulfill a leading role). The second subscale measures to what extent someone feels the need to lead others and wants to influence them.
Conflict management consists of the subscales assertiveness and emotion control. Assertiveness describes the extent to which someone will stand for his opinion and will argue for his point of view. Furthermore it measures how one reacts to a conflict
(enter or avoid it). The subscale emotion control measures if someone is capable of keeping his emotions under control or even able to hide them from the public.
The third scale, cooperation, describes someone’s preferences regarding group functioning. Its first subscale, the need to be part of a group, measures to what extent one wants to be part of a group. It also measures whether someone can act independent from other group members (low score on this subscale) or not. Second is the need for a good atmosphere. This scale measures if someone wants to prevent tension or try to resolve it. It also reflects whether someone tries to create a good working atmosphere or gets distracted when the atmosphere doesn’t feel good.
Flexibility is divided into the subscales changer and finisher. The subscale changer measures to what extent someone is willing to face new challenges and likes to experience new situations. The finisher subscale measures the extent to which someone wants to finish a task once he started one.
Emotional stability consists of the subscales impulsiveness and vulnerability. Impulsiveness measures the inability to control certain thoughts, feelings, needs and impulses. Vulnerability measures if someone easily experiences pressure or stress and finds it hard to deal with difficult situations.
Performance oriented striving includes the subscales hard worker and performing. The former measures to what extent someone gets motivated by just working hard (innate propensity to put effort in his work, not the demand of others to do so or the existence of a deadline). The latter measures to what extent someone gets motivated by the outlook on promotion or success. It also shows if someone aims for progression and wants to distinguish himself from others.
Finally, structure and details are the two subscales describing the seventh main scale precision. Structure measures the extent to which someone needs clear rules and guidelines in order to be able to work. The subscale details measures if someone is detail oriented and likes to do work which has a certain degree of accuracy.
As to date, the PIT test has not been used for scientific research. However, and further elaborated upon in section 3, this gives the opportunity for new findings and insights in this field of research.
2.5 Contribution to existing studies
As to date, limited scientific studies have been conducted analyzing the behavior of soccer players. As past studies mainly focusing on personality traits did not lead to consistent results, this study seeks to add value to the currently existing literature by switching to behavior instead of personality. Besides, there are two other reasons that distinguish this study from earlier studies. First of all, this study considers a different way of measuring player performance and corresponding value as compared to earlier studies. As mentioned before, many existing studies use rather poorly defined and variable measures of on-pitch performance. This study addresses this issue by making use of the so-called Euro Player Index (EPI), an algorithm based tool designed to measure a player’s contribution to the team’s performance through data and thereby stressing his sporting and corresponding financial value. In the next section, this performance measure is elaborated upon further. Finally, to my knowledge, no earlier studies have tried to translate athletes’ behavior to their social media reach. As sections 2.1 and 2.2 show that popularity and social media activity may influence the valuation of a soccer player, it is interesting to examine whether certain behavior correlates with a higher focus on off-pitch performance. The next section, which covers the methodology, addresses the method of this study and a description of the data set (including all sources of the data).
3. Methodology
This section covers the methodology and consists of the following two subsections: In section 3.1, I address all the data included in the analysis. A brief description of the data as well as the sources used is given. In section 3.2, I describe the method of this study. Here, all possible comparisons and links are explained.
3.1 Data set
In order to answer my research question, I use data on the behavior, individual on-pitch performance, and social media activity for soccer players. All 45 players included in the sample currently play at two different clubs in the Dutch Eredivisie and Jupiler league. Due to privacy regulations, the names of the clubs and players may not be disclosed.
In order to map the behavior of all 45 soccer players, I use PIT test data received from De Talentenacademie: a Dutch organization specialized in performance coaching and helping athletes get the most out of their talents. They developed the PIT test to gain insights into the behavior of the athletes they work with. It is an ipsative test, meaning all questions contain two statements where the participant has to choose the statement that suits him/her the most. Despite the fact that the PIT test has so far not been used for scientific research, an early research by Saville and Willson (1991) supports the use of ipsative approaches in the measurement of personality/behavior. In their attempt to compare ipsative and normative approaches, they used both hypothetical ‘true’ scores through a computer simulation as well as real data on 243 subjects. Under some conditions, they find that ipsative scores actually correlate better with hypothetical ‘true’ scores than the normative form (p. 228). As for the real data, both normative and ipsative approaches showed significant correlations with external ratings (pp. 234-236). Their conclusions regarding the reliability of using ipsative tests are supported by, among others, Block (1957), Bowen et al. (2002) and Chan (2003). The PIT test contains a total of 42 questions. The outcome of the test includes a score, ranked between 0 and 6, on all 14 subscales as described in the literature review. A total of 42 points is distributed, making it impossible to get a minimum/maximum score on all subscales. In principle, all scales are independent of each other: a high score on the subscale structure does not necessarily coincide with a high score on any other subscale. However, conclusions can be drawn based on the combination of scores on different subscales.
Furthermore, I use the EPI to measure individual on-pitch performance and corresponding value of all players. This is an algorithm-based instrument created in 2007 by Hypercube and Remiqz; two companies specialized in predictive intelligence
services for soccer clubs. Through the EPI, they can determine the current and future added sporting and corresponding financial value (expressed in ‘EPI points’, not in euros) of soccer players to their current clubs as well as potential future clubs. The EPI measures not only a player’s basic individual statistics (matches played, goals, cards etc.), but also and perhaps more importantly his specific contribution to the results of his team (‘’what difference does it make for match outcomes if he plays or not?’’). This is done by taking into account the strength of both his own as well as the opponents’ team (through the Euro Club Index, which is similar to the EPI, but at the club level, displaying the relative playing strength of a club at a given point in time), the competition, whether it is a home or away fixture (due to home advantage) and the timing of an occurrence during a game. This means a winning goal during extra time yields more additional EPI points than the fifth goal of a 6-1 victory, something that is not taken into account by just simply looking at ‘number of goals scored during the season’. It also means a difficult away win yields more points than a pre-expected home win. Although the EPI and ECI have, to my knowledge, not been used in scientific research or only to a limited extent, the two indices have proven to be consistent measurements of individual performance and match outcome predictions (‘’Remiqz wins group stage’’, 2018; ‘’RPS 17/18’’, 2018). Therefore, through its use, this study seeks to add value to the existing literature. Besides using the Euro Player Index, data retrieved from https://wyscout.com/ concerning more specific on-pitch factors (average amount of dribbles/shots/key passes etc. per 90 minutes) is used for the analysis.
Unfortunately, due to privacy regulations and cost concerns, some crucial parts for this off-pitch analysis could not be provided. Therefore, only a concise version of the analysis is included in the results section. This is solely done for indication purposes, I do not draw any conclusions based on the available data. For this part, the number of Instagram followers is used as an index for off-pitch popularity. Of the big social media platforms (e.g. Facebook, Twitter, Instagram), Instagram is considered the most used platform in terms of informing, interacting with and inspiring the fans. As social media also influence the value of a player, data regarding their use are also taken into consideration. Not as a main driver, but nowadays the role of social media cannot be neglected.
3.2 Method
As made clear in section 3.1, both measurement indices of behavior and on-pitch performance have not been used in scientific research or only to a limited extent. Therefore, no hypotheses or expectations are made up front. This study serves the purpose of a fishing expedition, thereby hoping to find valuable information regarding the relationship between soccer players’ behavior and their on- and off-pitch performance.
The procedure of the analysis is as follows. First, I analyze the PIT test results of the 45 players included in the sample, both as a whole and per line. Based on these results, I perform a deeper analysis of the relationship between the behavior subscales and specific on-pitch details as well as EPI levels and thus overall on-pitch performance and corresponding value by means of OLS regressions. Then I run a simple regression in order to test whether certain behavior correlates with higher social media reach. Finally I make a connection between EPI levels and social media reach in order to find out whether high on-pitch performance correlates with higher online popularity or if active online participation may just as well increase off-pitch value.
4. Results
In this section, I present the results of the data analysis. The first two tables present basic summary statistics of the PIT test scores, both for the entire sample and per line. Table 1 also contains the average EPI levels of the players taken over the past season, their current age and the number of Instagram followers they have. Thereafter, an in-depth analysis follows to link behavior to specific on-pitch related factors. Finally, the off-pitch analysis demonstrates how off-pitch performance (number of followers) relates to EPI levels and whether social media reach can be explained through the PIT test results.
Table 1
Subscale Observations Mean Std. Dev. Min. Max.
Faith 45 2.022 1.305 0 5 Need to lead 45 2.044 1.261 0 6 Emotion control 45 3.422 1.438 0 6 Assertiveness 45 2.667 1.225 0 6 Atmosphere 45 3.511 1.121 1 5 Part of group 45 4.156 1.313 0 6 Changer 45 2.933 1.321 0 5 Finisher 45 2.978 1.138 1 5 Impulsiveness 45 1.956 1.692 0 6 Vulnerability 45 1.089 1.379 0 6 Working hard 45 3.911 1.593 0 6 Performing 45 4.844 1.224 0 6 Structure 45 3.267 1.214 1 6 Details 45 3.2 1.290 0 6 Age 45 24.42 4.065 16 37 EPI 42 1204.83 477.74 16 2177 Instagram 36 6654.06 5736.03 237 20300
This table reports the summary statistics of all subscales included in the PIT test as well as the age and average EPI levels of the players last season.
Figure 1
This figure reports the PIT test scores on the subscale need to lead others of all players included in the sample. The linear black line represents the trend line of all scores combined, showing whether the scores increase with age.
Table 1 displays the summary statistics of the PIT test scores for the entire sample of 45 players. Some interesting findings can be derived. First of all, the players have high scores on the subscales performing (4.844) and working hard (3.911). In general, they get motivated by the view of promotion and success and by putting a lot of effort in their tasks. Furthermore they want to distinguish themselves from others, which makes sense in a surrounding as competitive as that of top sport. Also interesting are the high mean scores on the cooperation subscales atmosphere (3.511) and need to be part of the group (4.156). In line with findings by Beauchamp et al. (2007) regarding group cohesion, these results indicate that top athletes care about group functioning and feel the need to be part of the team. Thirdly, and in line with Eysenck’s theory and findings by Williams and Parkin (1980), Kirkcaldy (1982), Meuris et al. (1993), Egloff and Gruhn (1996) and Allen et al. (2011), the players generally score low on the subscales Impulsiveness (1.956) and vulnerability (1.089), indicating they are emotionally stable. Lastly, the scores on the leadership subscales faith (2.022) and need to lead others (2.044) are remarkably low. This might make sense, as a team only needs one or a few players who feel the need to lead the group. It could also be due to the relatively low average age of the players in the sample (24.42). To
compare, the average age at this year’s world cup was 27.9 (‘’Every World Cup squad’’, 2018). Perhaps older, more experienced players feel a higher responsibility of carrying the team and thus leading others. Figure 1 displays the relationship between the score on the subscale need to lead others and age for all players included in the sample. The trend line in figure 1 shows only a slightly increasing slope (correlation coefficient of 0.0584) and the p-value of 0.703 indicates there is no significant difference in test scores as age increases. Given the small sample size, this can be partly explained by the top left outlier as can be seen in the figure. However, after removing it, the correlation coefficient only increased to 0.1349 (p-value of 0.38) and thus remained insignificant. Interestingly enough, the two highest scores belong to the captains of both teams included in the sample, indicating that coaches select a captain based on his behavior. However, more data on this part is needed in order to validate such assumptions.
Table 2
Subscale Goalkeepers Defenders Midfielders Attackers
Faith 1.714 2.429 1.667 2.083 Need to lead 1.857 2.357 2.25 1.583 Emotion control 3.143 3.071 3.75 3.667 Assertiveness 2.857 2.571 2 3.333 Atmosphere 3 3.714 3.583 3.5 Part of group 3.571 4.286 4.167 4.333 Changer 3.429 2.429 3.167 3 Finisher 3.571 2.786 2.917 2.917 Impulsiveness 1.286 1.5 1.583 3.25 Vulnerability 0.714 1.214 1.333 0.917 Working hard 3.857 4.429 4 3.25 Performing 5.286 4.857 4.25 5.167 Structure 3.286 3.286 3.833 2.667 Details 4.429 3.071 3.5 2.333
This table reports the mean scores on all 14 subscales per line.
Table 2 displays the average PIT test scores on all 14 subscales per line. Goalkeepers have the highest scores on the subscales changer, finisher, performing and details. Defenders have the highest scores on the two leadership subscales, atmosphere and working hard. Midfielders have the highest scores on emotion control, vulnerability
and structure and finally attackers outperform the other lines on the subscales assertiveness, part of group and impulsiveness. Testing for significant differences across subscales per line, I only find attackers’ scores on impulsiveness and goalkeepers’ scores on details to be significantly different from the scores of all three other lines on those scales. Most t-values are insignificant, however. Table 4 in the appendix gives a complete overview of the t-values and corresponding significance levels of all possible PIT test/on-pitch position comparisons.
But what do for example high scores on impulsiveness mean? In words, it means one acts fast and intuitively and likes taking on risks, which might form an explanation for the high average scores of attacking players. Considering the scarce moments they have during a match to make a difference, creativity and split-second decision-making may be crucial. Interesting would be to analyze if and how scores on impulsiveness translate to specific on-pitch factors. To test this, I look at three variables that require quick thinking and intuition: dribbles, key passes and shots (both on and off target). Here, key passes are either passes to the final third (also offensive third) or passes to the penalty area. The data on these variables are retrieved from https://wyscout.com/ and are measured as averages per 90 minutes of play, taken over the past season. By means of an OLS regression, I test whether the average height of these three variables increases as the score on the subscale impulsiveness increases. For this regression, goalkeepers are excluded as the variables used may only apply to them to a limited degree. Table 3 displays the results of the OLS regression. Average numbers of dribbles and shots per 90 minutes have a positive effect on impulsiveness, whereas the number of key passes actually decreases as the score on impulsiveness increases. However, only the coefficient on shots is slightly significant, at the 10% level. This makes sense, considering the relatively high average scores on impulsiveness among attackers. However, based on this sample the outcomes are insufficient to state that higher scores on impulsiveness lead to significantly higher scores on specific on-pitch factors.
Table 3
Variable Coefficient Std. error T-value
Dribbles 0.03495 0.12002 0.29
Shots 0.62589 0.32227 1.94*
Key Passes -0.06679 0.09247 -0.72
This table reports results relating the effect of average dribbles/shots/key passes per 90 minutes on the score of the PIT test subscale impulsiveness.
* denotes significance at 10% level
Next, I look at the relationship between the PIT test scores and the average EPI levels of all players included in the sample. Here, EPI levels are measured as an average of all daily EPI levels taken over the past season, between 30/06/2017 and 30/05/2018. As displayed in table 1, three players drop out when analyzing the EPI levels. All three did not have any EPI data yet, as they just left the youth squad and still have to make their debut on the professional level. This leaves 42 players for the analysis.
Table 4.1
Variable Coefficient Std. error T-value
Faith 237.13 103.79 2.28**
Need to lead others -38.49 98.95 -0.39
Emotion control 104.05 77.01 1.35 Assertiveness -78.05 109.04 -0.72 Atmosphere 13.00 84.37 0.15 Part of group -152.83 80.13 -1.91* Changer 0 (omitted) Finisher 0 (omitted) Impulsiveness 132.48 74.23 1.78* Vulnerability -7.54 66.56 -0.11 Working hard 98.04 81.03 1.21 Performing -17.10 91.63 -0.19 Structure -15.36 78.99 -0.19 Details 46.04 101.29 0.45 Age 42.90 20.14 2.13**
This table reports the results of a multiple linear regression. The outcomes display the effect of all PIT test subscales and age on the EPI levels of the players.
* denotes significance at 10% level ** denotes significance at 5% level
Table 4.2
Variable Coefficient Std. error T-value
Faith 181.84 52.18 3.48***
Need to lead others 131.62 55.78 2.36**
Emotion control -41.22 51.17 -0.81 Assertiveness 53.71 59.09 0.91 Atmosphere -97.11 63.69 -1.52 Part of group -116.66 60.70 -1.92* Changer 29.67 57.45 0.52 Finisher -79.66 65.94 -1.21 Impulsiveness 35.31 43.62 0.81 Vulnerability -19.02 54.41 -0.35 Working hard 6.56 47.05 0.14 Performing -69.74 59.22 -1.18 Structure -24.89 61.63 -0.40 Details -56.39 57.48 -0.98
This table reports the results of 14 single linear regressions, all measuring the individual relationship between a subscale and the EPI levels. This time, the correlations between subscales are not taken into account.
* denotes significance at 10% level ** denotes significance at 5% level *** denotes significance at 1% level
A first OLS regression results in the outcomes as displayed in table 4.1. The large dispersion between EPI levels can explain the high coefficients of some variables. As the PIT test subscales only take on a value between 0 and 6, a one-unit change on a subscale can have large consequences for a player’s EPI level. Of the 14 subscales, only the coefficient on faith is significant at a 5% level, indicating one’s confidence in his leadership capabilities contributes to his sporting and financial value (through higher EPI levels). Furthermore, the subscale part of group has a significant negative effect on the average EPI, whereas the subscale impulsiveness has a positive effect, both at the 10% level. Although significant, the variable age is difficult to interpret when running a linear regression, as age and performance in general do not have a linear relationship in sports. Dendir studies the peak age for professional soccer players using data from the four major European leagues (2016, p. 91). He finds that, depending on playing position, professional soccer players peak between 25 and 27 years of age (pp. 101-102). As displayed in figure 2, I use a second order polynomial
trend line to show whether the players indeed have a peak age whereafter their EPI levels start to drop again. Here, the fluctuation of the data motivates the choice for a polynomial trend line. The trend line indeed indicates that age only positively correlates with EPI levels to some point. However, although the second order polynomial trend line fitted the data best, the low R-squared value does question its reliability.
Finally, the estimation routine omits the variables changer and finisher. It does so because of a dependency between these variables and at least one of the other independent variables in the model. Considering that all subscales are in principle independent, I am also interested in the relationship between individual subscale scores and the players’ EPI levels. I therefore run an extra set of single linear regressions measuring the individual relationship between a subscale and the EPI levels. Table 4.2 displays the outcomes of the 14 single linear regressions relating all subscales to the players’ EPI levels. In terms of significant coefficients, three things changed comparing the single linear regressions to the multiple linear regression. Firstly, the coefficient on faith becomes more significant, at a 1% level. Secondly, the coefficient on need to lead others becomes positive and significant, at a 5% level. Combined with the first finding this indicates that leaders, or at least players who exhibit leadership behavior, have higher EPI levels, resulting in higher economic valuations. This makes sense, considering the influence a leader may have on the group and thus the value he may create, both on an individual as well as a group level. Finally, compared to the outcomes as displayed in table 4.1, the coefficient on part of group remains negative but more significant, at a 5% level. As shown in table 5 in the appendix, high scores on part of group may indicate that one becomes dependent of and stays loyal to the group, perhaps even at the cost of his own individual performance.
Figure 2
This figure reports the relationship between the players’ age and average EPI levels. The polynomial trend line displays how the EPI levels change when age increases. The R2 expresses how well the trend
line fits the data.
Ultimately, I perform an off-pitch analysis. Firstly, figure 3 shows whether there is a smooth upward correlational trend between the number of Instagram followers and the EPI levels of the players. Thereafter, I run a set of single linear regressions just as in table 4.2, but instead using Instagram followers as the dependent variable.
Figure 3
This figure reports the relationship between the EPI levels of the players and the number of followers they have on Instagram. The polynomial trend line displays how the EPI levels change when the number of followers increases. The R2 expresses how well the trend line fits the data.
Out of the original sample of 45 players, 36 have a personal account on Instagram. Considering the three youth players drop out, as they do not have a Euro Player Index yet, only 33 players remain for the analysis. The scatterplot in figure 3 displays how the EPI levels and Instagram followers of all players relate to each other. I use a sixth order polynomial trend line in order to address how the EPI levels change when the number of followers increases, as it fitted the data best (highest R2). At first the line has a steady upward trend, indicating that a player’s off-pitch popularity increases according to his on-pitch performance. However, the trend line starts to drop eventually, due to a selection of players demonstrating excessive off-pitch popularity as compared to how they perform on-pitch. Regressing the EPI levels on the number of followers does result in a significant positive effect (t-value of 1.84, significant at a 10% level). Nevertheless, and previously mentioned in section 3, I draw no conclusions based on this data.
Table 5
Variable Coefficient Std. error T-value
Faith -208.32 784.84 -0.27
Need to lead others -480.72 847.21 -0.57
Emotion control 939.84 720.94 1.30 Assertiveness 1334.57 773.34 1.73* Atmosphere 6.78 868.24 0.01 Part of group 285.25 960.70 0.3 Changer 19.73 901.47 0.02 Finisher -1644.7 883.63 -1.86* Impulsiveness 1119.08 570.28 1.96* Vulnerability -1585.58 878.33 -1.81* Working hard -817.69 646.63 -1.26 Performing 186.12 1094.29 0.17 Structure -0.66 809.64 -0.00 Details -395.82 752.81 -0.53
This table reports the results of 14 single linear regressions, all measuring the individual relationship between a subscale and the number of instagram followers.
* denotes significance at 10% level
Finally, table 5 reports the results of a set of single linear regressions demonstrating the relationship between an individual behavior subscale and the number of Instagram followers. Again, the large coefficients are due to the large dispersion between
Instagram followers and the small range of values a subscale can take on. The results indicate that assertive, impulsive behavior has a significant positive effect on the number of followers, whereas the subscales vulnerability and finisher have a negative effect, all at a 10% significance level.
5. Discussion and conclusion
This section covers the discussion and conclusion. In the first paragraph, I give a brief summary of all the findings and their contributions towards answering my research question as well as some overall concluding remarks. Thereafter I describe the possible limitations of this study. Finally, I give several suggestions for follow-up studies.
In this study, I tried to find answers to the research question ‘‘how does soccer players’ behavior relate to their on-pitch performance and corresponding value and can this behavior also explain their off-pitch popularity?’’ The first set of summary statistics showed that the players included in the sample had high average scores on the performance oriented striving and cooperation subscales, whereas low scores were found on the emotional stability and leadership subscales. Furthermore, the graph in figure 1 led to believe that the need to lead others (and thus a feeling of responsibility) does not come with age. A second set of statistics differentiating between lines on the pitch did not indicate many significant differences among playing positions. Only attackers’ scores on impulsiveness and goalkeepers’ scores on details were significantly different from the scores of all three other lines on those scales. An attempt to translate scores on impulsiveness to number of dribbles, shots and key passes during a match did not result in significant findings. Possible reasons for these insignificant results are discussed in the next paragraph. To finalize the on-pitch analysis, I ran OLS regressions in order to link behavior to the players’ sporting and corresponding financial value through their EPI levels. Leadership behavior among players turned out to have significant positive effects on EPI levels. On the other hand, being ‘too loyal to and dependent of the group’ may harm individual performance. The earlier mentioned high general scores on performance oriented striving and cooperation and low scores on emotional stability are thus not sufficient to distinguish between high/low performing players. However, it does indicate that
professional soccer players generally place high emphasis on these behavior aspects. Finally, age only positively correlated with EPI levels to a certain point, indicating that performance slowly stagnates and eventually worsens throughout a player’s career. Unfortunately, I have not been able to draw conclusions regarding the effect of behavior on off-pitch popularity. Due to privacy regulations and cost concerns, some crucial data needed for this part of the analysis turned out to be too difficult to collect. Although some interesting findings were derived from this study, not all results were significant. Furthermore, some but not all results were consistent with previous studies. Some limitations of this study may account for these differences and insignificances. First of all, and mentioned previously, the small sample size may question the validity of the results. Especially the analysis per line can be improved through sample size enlargement. Also, a deeper analysis in combination with a larger sample is needed in order to relate behavior to specific on-pitch factors. Secondly, although extensively used in practice, up until now both the PIT test and the EPI have not been used for scientific research purposes. However, earlier studies supporting the use of ipsative tests and studying behavior rather than personality traits motivated the choice of using the PIT test.Although the EPI is said to be an appropriate and reliable measurement index for a soccer player’s sporting and financial valuation, it remains difficult to obtain a complete overview of all factors that in the end determine his value to the club. Therefore, it cannot be excluded that some explanatory variables were unknowingly omitted from the regression analysis.
In conclusion, given the research question of this study, behavior, on-pitch performance and corresponding value effectively are related. However, and partly due to the limitations and difficulties in collecting data, this study also shows that it remains difficult to obtain a complete overview of the factors in the end determining the value of a soccer player. More specifically, it remains difficult to pin down the exact role of behavior in valuing a player. Still, a lot more research needs to be done in order to improve the knowledge in this field of interest. Based on the scope and limitations of this study, I have several suggestions for follow-up studies. A first suggestion is the replication of this study using a larger sample size and by this testing the validity of the results found. This also includes the use of a more diverse sample in terms of age. Given the surplus of young players included in the current sample and
the lack of the ability to clearly distinguish between old and young players (also due to the small sample size), generalizing the results is difficult. Furthermore, a qualitative study could be done in order to validate the use of the PIT test and the EPI for scientific research purposes. Also, during this study I tried to explain performance and value generation through behavior rather than personality traits. However, as to date no studies have been conducted testing which of these two is a better predictor for performance among soccer players and their corresponding value. A future study could address this question by using both behavior as well as personality traits on the same set of soccer players in order to test which method has a stronger correlation with performance. Finally, this study failed to translate soccer players’ behavior to their focus on off-pitch performance. Considering certain behavior of players is likely to lead to a higher emphasis on off-pitch performance, future studies could address this issue by means of a more approachable method.
6. Appendix
Table 6
Subscale High score Low score
Faith ‘Sees himself as a true leader’ ‘Confident in a leading role’
‘Can have difficulties accepting others’ leadership’
‘Low confidence in fulfilling a leading role’
‘Fine with delegating leadership’
Need to lead others
‘Likes controlling others’
‘Can have difficulties letting go of a leading role’
‘Can have difficulties accepting not getting a grip on the group’
‘No need to control others’
‘Does not want to be responsible for others’ behavior’
Assertiveness ‘Will not avoid a conflict’ ‘Is open and honest about what he wants’
‘Will avoid conflicts’
‘Can have difficulties speaking out his opinion’
Emotion control ‘Does not show/tell his feelings and emotions’
‘Makes a calm and relaxed impression’
‘Shows positive as well as negative emotions openly’
‘Can have both positive and negative consequences on the atmosphere’
Part of group ‘Likes being around and working with others’
‘Loyal to the group and seeks support of others’
‘Can become dependent of the group and sensitive to peer pressure’
‘Can be independent and resistant towards peer pressure’
‘Seems confident but at times also distant’
Atmosphere ‘Alert to preventing tensions’
‘Will put effort into getting to know all members of the group’
‘Has a positive view on people’
‘Can function in an environment with a grim atmosphere’
‘Does not bother others’ opinions about him’
Changer ‘Is open to changes and has no difficulties adjusting’
‘Does not like routine work and can get bored in case of low task variety’
‘May resist changes at first’
‘Likes routine work and tasks that remain the same over time’
‘Needs time to adjust to new situations’ Finisher ‘High responsibility for his tasks’
‘Wants to finish something once started and has difficulties handing over tasks’ ‘Is able to ignore setbacks’
‘Likes to focus on one or a few tasks, can have difficulties switching between multiple tasks’
‘Likes picking up new ideas but then delegating these ideas to others’ ‘Might risk picking up too many ideas, but remaining unfinished’
Impulsiveness ‘Has difficulties resisting thoughts, desires and impulses’
‘Prefers acting fast and intuitively’ ‘In general likes taking risks’
‘Acts based on well-considered thoughts’ ‘Others trust his decisions’
‘Thought-out decisions might come at the cost of the pace of work’
Vulnerability ‘Is sensitive to stress and can have difficulties handling stressful situations’
‘Competent to deal with pressure’ ‘Believes he is capable of handling difficult and stressful situations’ Working hard ‘Gets motivated by working hard and
putting a lot of effort into a task’ ‘Pitfall might be spending excessive time in a task, important to link the efforts to concrete goals’
‘Might work hard if told to, but it is never a goal in itself to work hard’
‘Might seem unmotivated and might avoid working hard’
Performing ‘Gets motivated by the view of promotion and success’
‘Wants to distinguish himself from others and cares about his work’
‘Is not motivated by the view of promotion or success’
‘Will not lose his motivation when the possibilities of improvement drop out’ ‘Might seem unambitious and
unmotivated’ Structure ‘Likes having a clear structure and will
act according to this structure’ ‘Seen as trustworthy by others’ ‘Might have difficulties working in a different way than used to’
‘Prefers creating his own flexible structure and taking own initiatives’ ‘Does not like tight supervision’
‘Pitfall might be that his actions are not in line with the organization’s rules’ Details ‘In general capable of noticing
shortcomings quickly’
‘Pitfall might be slowing down others or not moving forward as a
consequence of perfectionism’
‘Prefers speed of action over details’ ‘Might overlook important things as a consequence of his quantity over quality attitude’
This table reports the detailed descriptions that fit a high/low score on all 14 subscales of the PIT test.
Table 5. Test for significant differences per line
Subscale G>D G>M G>A D>M D>A M>A
Faith -1.21 0.08 -0.56 1.54* 0.68 -0.76 Need to lead -0.89 -0.61 0.69 0.18 1.65* 1.26 Emotion control 0.1 -0.86 -0.76 -1.18 -1.05 0.15 Assertiveness 0.6 1.71** -0.76 1.41* -1.58* -2.60*** Atmosphere -1.75** 1.06 -0.78 0.35 0.49 0.16 Part of Group -1.14 -0.77 -1.1 0.25 -0.11 -0.31 Changer 1.55* 0.37 0.76 -1.34* -1.23 0.33 Finisher 1.57* 1.1 1.25 -0.28 -0.31 0 Impulsiveness -0.33 -0.45 -2.47*** -0.15 -2.69*** -2.45*** Vulnerability -0.86 -0.81 -0.37 -0.19 0.61 0.67 Working hard -1.05 -0.23 0.65 0.87 1.71** 0.99 Performing 0.65 2.27** 0.31 1.09 -0.59 -2.36*** Structure 0 -1.15 1.34* -1.07 1.23 2.84*** Details 2.48*** 1.86** 3.54*** -1.05 1.59* 2.65***
This table reports the t-values of the differences tests of all 14 subscales per line. Here, G=goalkeeper, D=defender, M=midfielder and A=attacker. In the second column, a positive t-value means that goalkeepers have higher average scores than defenders. For a negative t-value, the opposite holds.
* denotes significance at 10% level ** denotes significance at 5% level *** denotes significance at 1% level
7. Reference list
Adler, M. (1985). Stardom and talent. The American Economic Review, 75(1), 208-212.
Adler, M. (2006). Stardom and talent. In V. Ginsburgh, & D. Throsby (Eds.), Handbook of economics of art and culture. Amsterdam: Elsevier, 895-906
Barrick, M. R., & Mount, M. K. (1991). The big five personality dimensions and job performance: A meta-analysis. Personnel Psychology, 44, 1–26.
Barrick, M. R., Stewart, G. L., Neubert, M. J. and Mount, M. K. (1998). Relating member ability and personality to work team processes and team effectiveness. Journal of Applied Psychology, 83, 377–391.
Beauchamp, M. R., Jackson, B., & Lavallee, D. (2007). Personality processes and intra-group dynamics in sport teams. In M. R. Beauchamp & M. A. Eys (Eds.), Group dynamics in exercise and sport psychology: Contemporary themes (pp. 25-41). Oxon, UK: Routledge.
Beauchamp, M., Maclachlan, A., & Lothian, A. (2005). Communication Within Sport Teams: Jungian Preferences and Group Dynamics. The Sport Psychologist, 19(2), 203-220.
Block, J. (1957). A comparison between ipsative and normative ratings of personality. The Journal of Abnormal and Social Psychology, 54(1), 50-54.
Bowen, C., Martin, B. A., & Hunt, S. T. (2002). ‘’A comparison of ipsative and normative approaches for ability to control faking in personality questionnaires’’. The International Journal of Organizational Analysis, 10(3), 240-259.
Brandes, L., Franck, E., & Nüesch, S. (2008). Local heroes and superstars an empirical analysis of star attraction in German soccer. Journal of Sports Economics, 9(3), 266-286.
Chan, W. (2003). Analyzing ipsative data in psychological research. Behaviormetrika. 30(1), 99-121.
Costa, P. T. Jr., & McCrae, R. R. (1992). Revised NeEO Personality Inventory (NEO-PI-R) and NEO Five Factor Inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment Resources.
Dendir, S. (2016). When do soccer players peak? A note. Journal of Sports Analytics, 2(2), 89-105.
Eagleton, J., McKelvie, S., & De Man, A. (2007). Extra Version and Neuroticism in Team Sport Participants, Individual Sport Participants, and Nonparticipants. Perceptual and Motor Skills, 105(1), 265-275.
Egloff, B., & Jan Gruhn, A. (1996). Personality and endurance sports. Personality and Individual Differences, 21, 223-229.
Franck, E., & Nüesch, S. (2008). Mechanisms of Superstar Formation in German Soccer: Empirical Evidence. European Sport Management Quarterly, 8(2), 145-164. Franck, E., & Nüesch, S. (2012). TALENT AND/OR POPULARITY: WHAT DOES IT TAKE TO BE A SUPERSTAR? Economic Inquiry, 50(1), 202-216.
Is experience key? Every World Cup squad ranked by average age oldest to youngest. (2018). Retrieved from https://talksport.com/football/379772/every-world-cup-squad-ranked-average-age-oldest-youngest-180605285273/
Kane, J.E. (1966). Personality description of soccer ability. Research in Physical Education, 1, 54-65.
