Using Multiple Membership Multilevel Models to Examine Multilevel Networks in Networked Organizations

As the network structures of work and community have grown more complex, multilevel networks have emerged as the main structural feature in organizational settings. Stressing the importance of the affiliation ties of the meso-level network, we propose a conceptualization of multilevel networks within networked organizations. To examine such networks, researchers have used both hierarchical linear models(HLMs, and exponential random graph models(ERGMs)and both show strengths and weaknesses. HLMs have focused on the effects of group characteristics on individual level nodes, and assumed that each node is affiliated with only one group. Thus they are unable to analyze the complexity of the cross-cutting ties in multievel network data from networked organizations. ERGMs, on the other hand, have been used for analyzing such networks and are able to show if the presence of certain ties shapes the development of others. However, these models assume that networks are self-organizing systems of endogenous ties and, as a result, exogenous factors are excluded from them. In this paper, we propose a new method of multiple membership multilevel models that reveal the complexity of the network at the meso-level, i.e., multiple ties between one individual-level node and multiple group-level nodes. To accomplish this we offer an examination of the Canadian research organization, GRAND NCE (Graphics, Animation, New Media and Design Network of Centers of Excellence)