Title: How opinion dynamics generates group hierarchies
Abstract: We recently proposed a model coupling the evolution of the opinions of the individual with the local network topology. The opinion dynamics is based on the Bounded Confidence model. The social networks is based on a group concept where each individual is totally connected to the members of its group and is linked to the individuals of the other groups with a given probability. During a time step, the individual has to decide between discussing with a member of its own network and applying the opinion dynamics, or moving groups because it has an opinion far from the average opinion of its own group. One of the main results we obtained is that the group sizes, starting from an homogeneous situation can be strongly heterogeneous at the equilibrium state. This kind of heterogeneity can be identified in many real networks. In this paper we present the complete set of behaviours that this complex model can exhibit, at group level. In particular we will focus on the mechanisms that lead to the stability of the groups and the appearance of heterogeneity in sizes.