Multilevel Network Analysis for the Social Sciences

Abstract

The objective of this paper is to illustrate some of the benefits of understanding formal organizations as multilevel network systems by examining the interdependence between formal and social interaction in organizations. We show how such an approach supports a more informative and contextually richer representation of the interdependences between formal and informal relations in organizations. We document the existence, complexity, and context-dependence of the relationships linking informal networks between lower-level actors (individuals in the case that we will be presenting) to formal networks between higher-level actors (subsidiary units in our case) in organizations. We argue that ignoring the formal relations existing between higher-level units may lead to overestimating the autonomy of social networks from the formal authority structure existing within organizations. Because authority relations cross-cut organizational levels, this issue cannot be fully addressed in studies of social networks within organizations conducted at a single level. We argue that the unique value of the most recent generation of MERGMs is to turn this problem into empirically testable hypotheses. Using field data that we have collected on communication and advice relations among the 47 members of a top management team within an international multiunit industrial group we show how this weakness may be addressed. We exploit the natural multilevel structure of social networks within organizations to specify and estimate MERGMs for different intra-organizational networks (advice and communication). We show that the effects of formal structure on social networks are contingent upon the specific kind of network that is being considered.