A Computationally Efficient Approximation of Beta Centrality

Much attention has been focused on developing methods of assessing centrality and power in networks, with particular interest focused on recursive measures that view the status of a node as a function of the status of other nodes. Although such measures have been widely adopted in both academic research and commercial applications, they are computationally intensive and subject to misspecification. This paper introduces a simple measure, alter-based centrality, as a computationally efficient approximation of one commonly used recursive measure, beta centrality. Comparison of these measures in simulated networks indicates that alter-based centrality offers a robust approximation of beta centrality in non-bipartite networks that is computationally efficient and not subject to misspecification. Authors Zachary Neal is an Assistant Professor of Sociology and Global Urban Studies at Michigan State University. He serves on the editorial boards of Global Networks and City & Community, as editor of the Routledge “Metropolis and Modern Life” book series, and as Associate Director of the Globalizationa and World Cities (GaWC) research network. In addition to his work on network methods, his research examines how networks can be used to understand urban phenomena at a range of scales. Additional details may be found on his website at http://www.msu. edu/~zpneal. Please address correspondence to Zachary Neal, Sociology Department, 316 Berkey Hall, Michigan State University, East Lansing, MI 48824-1111; email: [email protected]. 331036_B.indd 18 7/26/13 10:27 AM Approximation of Beta Centrality Connections July | Issue 1 | Volume 33 | 12