Ordered community structure in networks

Community structure in networks is often a consequence of homophily, or assortative mixing, based on some attribute of the vertices. For example, researchers may be grouped into communities corresponding to their research topic. This is possible if vertex attributes have unordered discrete values, but many networks exhibit assortative mixing by some ordered (discrete or continuous) attribute, such as age or geographical location. In such cases, the identification of discrete communities may be difficult or impossible. We consider how the notion of community structure can be generalized to networks that have assortative mixing by ordered attributes. We propose a method of generating synthetic networks with ordered communities and investigate the effect of ordered community structure on the spread of infectious diseases. We also show that current community detection algorithms fail to recover community structure in ordered networks, and evaluate an alternative method using a layout algorithm to recover the ordering.     

How to get the timing right. A computational model of the effects of the timing of contacts on team cohesion in demographically diverse teams

Lau and Murnighan’s faultline theory explains negative effects of demographic diversity on team performance as consequence of strong demographic faultlines. If demographic differences between group members are correlated across various dimensions, the team is likely to show a “subgroup split” that inhibits communication and effective collaboration between team members. Our paper proposes a rigorous formal and computational reconstruction of the theory. Our model integrates four elementary mechanisms of social interaction, homophily, heterophobia, social influence and rejection into a computational representation of the dynamics of both opinions and social relations in the team. Computational experiments demonstrate that the central claims of faultline theory are consistent with the model. We show furthermore that the model highlights a new structural condition that may give managers a handle to temper the negative effects of strong demographic faultlines. We call this condition the timing of contacts. Computational analyses reveal that negative effects of strong faultlines critically depend on who is when brought in contact with whom in the process of social interactions in the team. More specifically, we demonstrate that faultlines have hardly negative effects when teams are initially split into demographically homogeneous subteams that are merged only when a local consensus has developed.     

Advances in the attraction model for inter-group relations

The present paper discusses a model of attraction towards similar others as a theoretical account of the problem of homophily in social relations. The focus of the paper is on investigating the properties of the model. More specifically, upon presentation of the logic behind the model, the discussion goes on to investigate the effects of social structural conditions (i.e., margins of an association matrix) on the model parameters — especially those referring to the strength of the homophily bias. This investigation leads to a reformulation of the problem of salience of a characteristic for association, defined, after Blau (1977) and Skvoretz (1983), in terms of the difference between the actual frequency of intra-group ties and the frequency expected under randomness. The attraction model is then compared with the log-linear model of quasi-independence. The objective of this comparisons is a precise specification of the attraction model’s explanatory scope. The paper ends with an illustrative application of the model to GSS data on confiding relations along ethnicity and religious affiliation.