CITATION: Marsden, Peter V. 2005. "Network Analysis." Pp. 819-825 in The Encyclopedia of Social Measurement, Vol. 2 (Kimberly Kempf-Leonard, ed.). San Diego, CA: Academic Press.
AUTHOR_1: Peter V. Marsden
TITLE: Network Analysis
SOURCE_TITLE: Encyclopedia of Social Measurement
SUMMARY: A concise, general overview of the state of social network analysis, with specific attention paid to the measurement and basic utility of social networks and social network analysis. There is a glossary of selected common terms in network analysis at the outset.
The section on social network data and research designs suggests two dimensions for classifying network data: 1.) whole or partial network data (as opposed to the end-use of performing egocentric, dyadic, complete, etc. network analyses), and 2.) one- two- or multi-mode network data.
The section on graph theory and connectedness explains the fundamental distinction between nondirected graphs and directed graphs (digraphs), as well as the relationship between the adjacency matrix and its isomorphically related digraph. The section also solidifies the usage and relevance of key concepts like path, geodesic distance, symmetry, nodes, and edges. Bipartite graphs are defined and explained.
The section on visualization lays out basic possibilities for the graphic display of network data, including sociograms, three-dimensional plots, multidimensional scaling, and the use of spring embedding and other algorithms for standardized graph presentation. The utility of correspondence analysis for visualizing affiliation network data is noted.
The section on centrality and centralization introduces several basic measures of centrality (degree, betweenness, closeness, and prominence) and one of centralization (normalized degree centralization), and explains their relevance to social network analysis.
The section on range and composition is framed in terms of the significance of these two concepts for the measurement of social capital. Measures of each concept are listed, but not presented formally, as was done for simple measures of centrality and centralization.
The section on social differentiation and network subgroups introduces two approaches to the assignation of actors to subgroups: cohesive subgroups including cliques, and positional analysis, including structural equivalence and blockmodeling. Each of these approaches is described in brief, laying out their utility for the study of various aspects of social structure.
The chapter concludes with a section on statistical network analysis, with special attention paid to the p* family of random graph models, along with a consideration of some of the underlying assumptions of such models compared to other models of network structure.
REVIEWED BY: jrh
LABELS_FULL: reviewed, jrh, 2005, chapter, introduction, overview, history, theory, methodology, data structure, graph theory, connectivity, visualization, centrality, centralization, range, composition, social capital, subgroups, cohesive subgroups, structural equivalence, blockmodels, p* model