Thomas Grund

Network structure and team performance: The case of English Premier League soccer teams

Abstract A defining feature of a work group is how its individual members interact. Building on a dataset of 283,259 passes between professional soccer players, this study applies mixed-effects modeling to 76 repeated observations of the interaction networks and performance of 23 soccer teams. Controlling for unobserved characteristics, such as the quality of the teams, the study confirms previous findings with panel data: networks characterized by high intensity (controlling for interaction opportunities) and low centralization are indeed associated with better team performance.

How Natural Selection Can Create Both Self- and Other-Regarding Preferences, and Networked Minds

Biological competition is widely believed to result in the evolution of selfish preferences. The related concept of the ‘homo economicus’ is at the core of mainstream economics. However, there is also experimental and empirical evidence for other-regarding preferences. Here we present a theory that explains both, self-regarding and other-regarding preferences. Assuming conditions promoting non-cooperative behaviour, we demonstrate that intergenerational migration determines whether evolutionary competition results in a ‘homo economicus’ (showing self-regarding preferences) or a ‘homo socialis’ (having other-regarding preferences). Our model assumes spatially interacting agents playing prisoners dilemmas, who inherit a trait determining ‘friendliness’, but mutations tend to undermine it. Reproduction is ruled by fitness-based selection without a cultural modification of reproduction rates. Our model calls for a complementary economic theory for ‘networked minds’ (the ‘homo socialis’) and lays the foundations for an evolutionarily grounded theory of other-regarding agents, explaining individually different utility functions as well as conditional cooperation.

Multidimensional homophily in friendship networks

Homophily - the tendency for individuals to associate with similar others - is one of the most persistent findings in social network analysis. Its importance is established along the lines of a multitude of sociologically relevant dimensions, e.g. sex, ethnicity and social class. Existing research, however, mostly focuses on one dimension at a time. But people are inherently multidimensional, have many attributes and are members of multiple groups. In this article, we explore such multidimensionality further in the context of network dynamics. Are friendship ties increasingly likely to emerge and persist when individuals have an increasing number of attributes in common? We analyze eleven friendship networks of adolescents, draw on stochastic actor-oriented network models and focus on the interaction of established homophily effects. Our results indicate that main effects for homophily on various dimensions are positive. At the same time, the interaction of these homophily effects is negative. There seems to be a diminishing effect for having more than one attribute in common. We conclude that studies of homophily and friendship formation need to address such multidimensionality further.

Ethnic heterogeneity in the activity and structure of a Black street gang

Despite acknowledgment of ‘hybrid’ street gangs in the literature, there is little systematic research into ethnic heterogeneity within gangs. This research aims at moving beyond the broad categorization of the Black street gang. For this purpose, we examine an all-Black London-based gang in detail, using fieldwork and police arrest data, and investigate the role of ethnic heterogeneity for the workings of the gang. Our findings suggest that ethnic heterogeneity within this gang is crucial for its criminal operation. Although there is no evidence for ethnicity-related specialization of crime, the structural co-offending pattern of the gang’s activities is dominated by ethnicity. Ethnicity matters for who offends with whom.

Ethnic Homophily and Triad Closure Mapping Internal Gang Structure Using Exponential Random Graph Models

Previous studies indicate the importance of similarities between street gang members in terms of ethnicity for mapping the patterns of co-offending relationships within gangs. Individual members are more likely to co-offend with other members who are from the same ethnicity. Past studies, however, do not appropriately account for the fact that correlation between attributes of co-offending gang members may be driven by alternative mechanisms. Most importantly, the presence of clustering in networks can dramatically affect the assessment and make us believe that homophily—the deliberate choice to co-offend with others from the same ethnic group—is important while in fact it is not. In this article, we recreate the internal structure of a London-based street gang with exponential random graph (ERG) models. Our results refine the role of ethnicity for co-offending within gangs. While homophily is still prevalent, the effect diminishes when triad closure—the tendency for two individuals to offend with each other when they also offend with a common third person—is considered. Furthermore, we extend existing ERG specifications and investigate the interaction between ethnic homophily and triad closure. Findings indicate that ethnic homophily is even stronger when it is embedded in co-offending triads.

Punishing the Foreigner: Implicit Discrimination in the Premier League Based on Oppositional Identity*

We present the first empirical study to reveal the presence of implicit discrimination in a non-experimental setting. By using a large dataset of in-match data in the English Premier League, we show that white referees award significantly more yellow cards against non-white players of oppositional identity. We argue that this is the result of implicit discrimination by showing that this discriminatory behaviour: (i) increases in how rushed the referee is before making a decision, and (ii) it increases in the level of ambiguity of the decision. The variation in (i) and (ii) cannot be explained by any form of conscious discrimination such as taste-based or statistical discrimination. Moreover, we show that oppositional identity players do not differ in their behaviour from other players along several dimensions related to aggressiveness and style of play providing further evidence that this is not statistical discrimination.

Network Dynamics and Market Structure: The Case of Art Fairs

Annual art fairs are a crucial element of the contemporary art market. Art galleries represent artists and exhibit their work at fairs all over the world. These fairs expose the gallery and its artists and create and maintain relationships with key actors in the global art market. Using data recording artists and galleries presenting at art fairs alongside qualitative interviews with different actors operating in the art market, we study the development of the art fair network over three years, 2005–2007. Our findings suggest significant network and homophily effects regarding the status and age of artists at fairs. However, the internationality of fairs and geographic distances between them do not seem to have any effect.

The Relational Value of Network Experience in Teams: Evidence From the English Premier League

Previous studies indicate the importance of experience for the performance of teams. Theory suggests that working together allows individuals to (a) improve their knowledge about who knows (and can do) what and (b) facilitates learning to combine individual resources efficiently. Yet it remains elusive how experience translates into interaction patterns in teams. Drawing on unique data of career histories of 800 players and 283,259 passes between these players in 760 English Premier League soccer matches, I propose a new measure for network experience and demonstrate how it relates to network intensity and decentralization in teams. Soccer teams exhibit a higher passing rate when players know each other from before. At the same time, network experience has no effect on the decentralization of team play. Further dyadic analyses confirm these findings. Controlling for selection effects, two players are more likely to pass the ball between each other when they know each other more. Network experience affects the way team members interact and ultimately leads to performance outcomes.

Overlapping crime: Stability and specialization of co-offending relationships

Abstract Dyadic analyses of relationships between criminals have mostly ignored the multiplex nature of criminal ties. This study attempts to provide a more complete assessment of co-offending networks by incorporating the different types of crime that relate individuals with each other. Drawing on a large dataset of arrests in Quebec between 2003 and 2009, we focus on co-offending stability and specialization and illustrate how co-offending networks based on different types of criminal activities overlap. We portray a pattern of co-offending, which extends debate of criminal specialization/versatility to the dyadic level. Our study illustrates the ways in which the frequency and spectrum of crime include a relational component. More generally, the article emphasizes the need to consider the semantics of network ties, and further, the association between different types of networks, which ultimately offers a reassessment of social structure.

EDITORIAL: AGENT-BASED MODELING AND TECHNO-SOCIAL SYSTEMS

In the past, sociologists have predominantly drawn on empirical research methods, while, for example, economists have mostly used formal and analytical techniques. While there are still opportunities to fruitfully combine these approaches, one further field where cross-fertilization can take place is the study of complex systems with their counterintuitive interaction patterns, self-organization phenomena and emergent system properties. So far, only a few tools were available to study these kinds of phenomena. A promising avenue to advance such research is the use of computer simulations and agent-based models (ABMs). ABMs can simulate and help to understand macro-level phenomena from bottom-up principles. The agents, as “virtual” representations of individual actors (be it individuals, companies or institutions), are situated within social, structural and normative environments, which are reflected by the ABMs [1]. The behavior of agents, in combination with the interactions between agents, can be shown to produce interesting and often unexpected macro-level effects. Computer simulations have been used as a standard tool in engineering and the natural sciences for many decades. In the social sciences, however, the use of ABMs is still at the beginning, although the notions of agents and individuals quite naturally overlap. The rule-based rather than the equation-based approach of agentbased modeling is simple, flexible and intuitive. In particular, ABMs are suited to reflect the inherent heterogeneity of decision-making agents. Furthermore, the important role of randomness or noise for the emergence of macro-level phenomena can be easily addressed [2]. Computer simulations offer new possibilities for studying