Theory and methods -- Social Network Analysis: Method and Applications by Stanley Wasserman and Katherine Faust

Abstract

Review.

Social Network Analysis: Methods and Applications, by Stanley Wasserman and Katherine Faust. New York: Cambridge University Press, 1994. 825 pp. $64.95 cloth. ISBN: 0-521-38269-6. $29.95 paper. ISBN: 0-521-38707-8.

DAVID KNOKE University of Minnesota

The long-awaited publication of this volume marks a half-century maturation of social network analysis into a multidisciplinary research specialty with distinctive vocabulary, theoretical principles, and data-analytic techniques. Although Stanley Wasserman and Katherine Faust modestly describe their goal as a methodological reference work, a perusal of their 1,000+ references and the numerous data illustrations reveals a wide range of substantive applications by sociologists, social psychologists, anthroplogists, political scientists, social workers, communications, business and public health researchers, and even the odd economist. By no coincidence, this stock-taking effort appeared on the eve of the consolidation by U.S. and European branches into a single international social network conference. As a road map of the terrain, Social Network Analysis is assured of frequent consultation by seasoned researchers and novices alike. (The Book Reviewer's Full Disclosure Act requires revealing my professional connections to the authors and the fact that Cambridge's structural analysis series includes my own network book: another instance of "small world" phenomena found in every specialty.)

With their roots in Simmel's web of affiliations and Moreno's sociograms, contemporary social network analysts emphasize the relational connections among social actors rather than the standard focus on individuals' attributes, attitudes, and actions. The contrast is epitomized by alternative survey items asking, "How often do you have sexual intercourse?" and "Who are your sexual partners?" Using actor-to-actor interactions to map ties between all participants in a social collectivity enables network investigation to proceed at several levels of analysis, from single ego-actors, to small groups, to the complete systems as large as the planetary world economy. The potential for bridging the micro-macro chasm is an enduring attraction for many scholars who are disenchanted with atomistic images of human behavior and variable-centric measures. Network approaches conceptualize such basic social units as roles, positions, statuses, groups, cliques, and circles as fundamentally structural in nature. The ultimate aim is to extract enduring patterns from the superficial welter and to explain their origins and consequences.

Although social network methodologies are strongly grounded in finite mathematics, algebra, and graph theory, Wasserman and Faust's exceptionally lucid treatment demonstrates that many core ideas are surprisingly simple to grasp and to apply. Clearly marked sections alert apprehensive readers to particularly difficult and tangential topics. Still, some statistical and mathematical background, particularly in matrix manipulations, will enable users to optimize their understanding of procedures. The authors build in much good redundancy across the chapters, allowing many to be approached as stand-alone treatments of a particular topic. Although periodic recaps of basic concepts, notations, and principles necessarily expand the volume, these occurrences offer numerous entry ports, since few people will be disposed to read it from cover to cover.

Despite attention to many topics, ranging from Galois lattices to automorphic equivalence, the "encyclopedic" adjective should not be attached to this 800-page tome. Rather, it offers a solid foundation, a careful codification of extant approaches, and in-depth expositions of a dozen fundamental properties: centrality, balance, cohesion, affiliation, structural equivalence, blockmodels, relational algebras, dyads and triads, statistical fits. But many derivations and proofs are not presented, and esoteric advanced topics are not covered, although numerous citations to the original sources abound. Wasserman and Faust deliberately restrict their attention to full networks, with scant attention to ego-centered nets and network samples. They draw a useful distinction between "one-mode" networks (N actors linked to the same N actors) and "two-mode" networks (N actors connected to M events), pointing out the relevant options for analysis and interpretation. Many methods are first described for commonly collected binary networks (where only the presence or absence of a tie is known), then generalized to less-prevalent signed and valued networks (where positive-negative direction and the strength of a tie is measured). Fundamental formulas and notations are always followed by clear verbal restatements (e.g., "the symmetry of X means ...") and often by graphic illustrations (although I eventually grew annoyed at the perky antics of six imaginary children, who each receive preface acknowledgments]). When several alternatives exist-for example, the bewildering varieties of centralization measures-Wasserman and Faust make informed recommendations and suggest appropriate choices for various conditions. I found very few typos in this well-designed and systematically structured project. In short, Social Network Analysis is an exceptionally user-friendly guide through the tangled thickets in the network forest.

Among the book's many outstanding features is frequent exemplification of abstract techniques with several small, real datasets, ranging from medieval Florentine families, to high-tech managers, to international trade patterns. By exposing these simple binary matrices to repeated deconstruction, the authors offer readers valuable insights into the rich information lodes that networks miners can unearth. Another asset is their periodic referral to computer programs, particularly to UNICET, which has emerged as a vehicle of choice for many networkers. Unfortunately, no problem sets are included, presumably to avoid the look and feel of a classroom text, which means that network instructors still lack an effective didactic tool. While on the matter of limitations, I must reproach the authors for compiling a particularly skimpy index to such a long and complex volume. For example, the "graph" entry directs readers to a mere three locations, and neither it nor the five cross-references enable readers to find a useful discussion on page 163 linking graphs and incidence matrices. Ironically, given that an index is a linguistic network, it falls short on the reliability standard the authors advocate.

Network analysis came of age during the past generation. Today it is fully institutionalized in the academy and is steadily winning converts in numerous disciplines. With Social Network Analysis, Wasserman and Faust provide a compass by which to steer our path into the next century.