Entropy-Based Centralization and its Sampling Distribution in Directed Communication Networks

Abstract

This article proposes a new measure of network centralization and reports the results of a simulation designed to determine the sampling distribution of the proposed measure. The measure is based on the information-theoretic concept of entropy and is appropriate for those networks in which traffic flows over paths and propagates by means of transference. Tables of critical values are constructed which allow the researcher to conduct tests of significance.

Keywords:

• Entropy
• Centralization
• Centrality
• Networks
• Statistical Inference

Acknowledgement

An earlier version of this paper was presented at the annual meeting of the International Communication Association, San Francisco, May 2007. The authors would like to thank Seth Richards for providing the data used to construct the network of Figure 2. Furthermore, the description of the events surrounding Kelly's suicide is based on Richards’ (2005) synopsis.

Additional information

Notes on contributors

Frank Tutzauer

Frank Tutzauer is at University at Buffalo

Benjamin Elbirt

Benjamin Elbirt is at University at Buffalo