Centrality measures for disease transmission networks.推荐.pdfVIP

Social Networks 21 1999 1–21 Centrality measures for disease transmission networks David C. Bell , John S. Atkinson, Jerry W. Carlson Affiliated Systems, 3104 Edloe, Suite 330; Houston, TX 77027-6022, USA Abstract This paper addresses the issue of the applicability of various network measures provided by analysis programs such as UCINET for the analysis of a disease transmission network. A simulation was conducted to estimate HIV transmission probabilities within an empirical network of cocaine injectors. The results of the simulation were then used as a criterion to evaluate network measures of disease vulnerability and infectivity. The analysis concludes that, for the type of disease transmission network studied, relatively simple ego-net- work-based network measures can be adequate to the measurement of vulnerability to disease and infectivity. 1999 Elsevier Science B.V. All rights reserved. 1. Introduction Network analysis has become popular in a number of social science disciplines. There is, however, a sense in which network analysis has been driven much more strongly by its methods than by its theories Marsden and Laumann, 1984; Friedkin, 1991 . Some observers have cautioned that the theoretical construct of a phenomenon of interest ought to determine the method chosen to measure that phenomenon Friedkin, 1991; Borgatti, 1995 . This sound advice is supported by the disparate views on which network measures are ‘best’ Freeman, 1979; Freeman et al., 1980; Knoke and Burt, 1983; Doreian, 1986; Bolland, 1988; Stephenson and Zelen, 1989; Friedkin, 1991 . The choice of method carries with it certain implied the