Читать книгу Social Network Analysis - Song Yang - Страница 10

Social networks are as old as the human species. As small bands of hunter-gatherers spread around the globe, their survival depended on cooperative strategies for pursuing game and finding good foraging grounds. Ties of family and extended kin were crucial to raising the next generations. With increased size and density of agrarian settlements, succeeded by expanding urban civilizations, networks grew increasingly complex and indispensable for merchants involved in long-distance commerce and armies engaged in conquest. Palace and court intrigues ran on gossip, rumor, and favor-trading among political factions. Scientific and technological advances necessitated information flows through invisible colleges of experts. Social networks have a truly ancient lineage yet are seldom noted nor well understood by their participants.

People today commonly envision social networking as clusters of coworkers going for lunch or coffee, teams of dormmates playing basketball or softball, and bunches of friends chewing the fat. Yes, those small groups are all social networks. To give a formal definition, a social network is a set of actors, or other entities, and a set or sets of relations defined on them. In the three preceding examples, the first actors are coworkers and the relations are lunchmate and coffeemate; the second actors are residents of the same dorm and playing sports is the relation; the third network is friends gossiping leisurely. Applying the definition to diverse social settings, we can easily uncover numerous social networks, some more formal than the three previously described. For example, a college academic unit has a social network composed of faculty members, staff, students, and administrators. Multiple sets of relations suffuse such networks: collegial relations among faculty members, faculty advising graduate students, faculty instructing undergraduates, and administrators supervising faculty and staff. A police department is also structured as a formal social network, in which officers at the same rank are colleagues, whereas a quasimilitary chain of command establishes hierarchical authority relations. Typical order from top down would consist of chief of police, deputy chief, captain, lieutenant, sergeant, corporal, patrol officer.

Although people typically conceive the actors in social networks as human beings, they can just as well be collective entities or aggregated units, such as teams, groups, organizations, neighborhoods, political parties, and even nation-states. For example, corporations can engage in cooperative and competitive relations to pursue many outcomes, such as jointly developing new technologies and products or acquiring greater market shares (Knoke, 2001). Interorganizational relations take many governance forms, from contractual agreements to equity stakes (Child, 2005; Yang, Franziska, & Lu, 2016). Inside organizations, work groups and teams often engage in knowledge transfers or information sharing to facilitate innovation and improve task performance (Tsai, 2001). International relational networks also emerge and evolve, including military alliances and conflicts, trade partnerships and disputes, human migrations, intelligence exchanges, and technology sharing and embargoes (Yang et al., 2016, Chapter 8).

Nonsocial networks are prevalent in many domains: technology networks, computer networks and the Internet, telephone networks and electrical power grids, transportation and logistics networks, food delivery, and patent-citation networks. They share some similarities with social networks, except that instead of actors their units are physical entities, such as computers and transformers, and their relations are transmission and delivery lines such as Ethernet cables, wireless connections, airline routes, and interstate highways. We mention nonsocial networks primarily to note that networks are the subjects of studies by many disciplines besides the social sciences. Those investigations illuminate and inspire one another, engendering strong momentum to improve network knowledge, including social network analysis (Knoke & Yang, 2008). For example, after mathematicians developed graph theory, computer scientists applied it to construct optimal computer networks. Social network scholars can borrow algorithms from computer and mathematical sciences to decipher communication networks among friends, coworkers, and organizations.

Sociology built a long tradition of examining the social contexts of social networks. Founding fathers such as Georg Simmel, Émile Durkheim, and Max Weber promoted a structural perspective in the study of human behaviors. Social psychologist Jacob Moreno (1934) was directly responsible for laying the foundation of modern social network analysis. With Helen Jennings, Moreno invented sociometry to draw maps visualizing individuals and their interpersonal relations, revealing complex structural relations with simple diagrams. Moreover, Moreno and other pioneering social network scholars endeavored to explain how network structures affect human behaviors and psychological states (Freeman, 2004). On the one hand, we can better understand people’s actions and decisions by examining their social networks because networks provide participants with both opportunities and constraints. On the other hand, the formation and change of social networks themselves have been the object of many research projects. An important sociological principle is social homophily, which asserts that people tend to form positive relations with others similar to themselves. Actors could be attracted to others based on similarity of attributes—such as gender, age, race, ethnicity, or socioeconomic status—or similarity of behaviors—such as life experiences, political preferences, religious beliefs, or hobby interests. In this perspective, social relations are outcomes, or dependent variables, occurring because actors share some of the independent variables listed previously.

Social network analysis was vitally important to the inception of economic sociology, a major specialty in sociology. In his classical article applying sociology to economic actions, Mark Granovetter (1985) criticized the undersocialized view of economists in which human decision making is driven solely by subjective expected utility maximization. Surprisingly, Granovetter likewise disapproved of the oversocialized view of sociologists in which human actions are determined solely by norms and social roles. So how does one avoid both under- and oversocialized explanations of human behaviors? The answer, quite obviously, is by using social network analysis: by looking at actors’ social networks, we can better understand their decisions and actions. Social networks generate localized norms, rules, and expectations among their members, which reinforce mutual trust and sanction malfeasance. Thus, by examining how social networks actually operate as both causes and consequences of human perceptions and actions, theorists and researchers avoid accepting either oversocialized or undersocialized perspectives. More importantly, although Granovetter (1985) emphasized economic behaviors, his arguments are very relevant to many social pursuits, such as making friends, casting votes, looking for a job, seeking promotion, finding a therapist, searching for emotional support, and locating instrumental help.

Early sociological and anthropological research on social networks inspired other disciplines to investigate the mechanisms instigating network formation in those fields. Over the past half century, mass communication, strategic management, marketing, logistics, public administration, political science, international relations, psychology, public health, criminology, and even economics begin introducing ideas and methods of social network analysis into those disciplines. For example, Zeev Maoz (2012) analyzed international trade and military alliances as network processes. He found that international trade follows a preferential attachment or bandwagon process: all nations want a quick and short connection to a few key nations in the global trade network, resulting in a highly condensed, single-core structure. In contrast, for military alliances, nations tend to partner with countries sharing similar political ideologies and regime structures. This homophily preference produces a network configuration consisting of multiple small military alliance clusters that are only sparsely interconnected (see also Yang et al., 2016, p. 198).

We would be remiss not to mention social media as an explosively growing component of social networks. Facebook, Twitter, LinkedIn, WeChat, and other apps facilitate a massive amount of daily information exchange among billions of users. Much social networking nowadays occurs in virtual spaces as users contact one another via computers, laptops, iPad tablets, and smartphones linked together by Ethernet cables or wireless. Computer communication networks and human social networks converge, engendering innumerable research opportunities and challenges for social and computer scientists. How does one best search, capture, aggregate, store, share, process, reduce, and visualize vast volumes of complex data generated by online social networkers (Press, 2013; Lohr, 2013)? John Mashey, chief scientist at Silicon Graphics, is often credited with coining the term Big Data, which he described in a slide presentation as “storage growing bigger faster” (1998, p. 2). Exponentially bourgeoning quantities of structured and unstructured information have revolutionized businesses, nonprofits, and governments. For social network researchers, Big Data is a trove of rich relational databases and a smörgåsbord of computer tools for data mining, information fusion, computational intelligence, machine learning, and other applications (de Nooy, Wouter, Mrvar, & Batagelj, 2018). Although Big Data enhances organizational operations and outcomes, it also raises numerous ethical and privacy challenges, such as the rise of surveillance state capacities to predict and control populations (Brayne, 2017; Madden, Gilman, Levy, & Marwick, 2017). Russian manipulation of the 2016 U.S. presidential election was only the most notorious of innumerable criminal abuses of Big Data on social media platforms. Calls for governmental regulation of social media companies encounter conundrums of how to protect platforms and safeguard free speech while prohibiting dangerous content (Berman, 2019). The fate of our democracy hangs in the balance.

In sum, social network analysis is a vibrant multidisciplinary field. Peter Carrington and John Scott called it “a ‘paradigm’, rather than a theory or a method: that is, a way of conceptualizing and analyzing social life” (2011, p. 5). We believe the network paradigm has roots in and thrives on the integration of three elements: theories, methodologies, and applications. For theories, network analysis demands serious commitment that prioritizes actor interdependence and connectivity, emphasizing structured relations among social entities. For methodologies, network analysis borrows eclectically from diverse disciplines, collaborating across the aisles to create innovative procedures. For applications, people increasingly use their networking skills to navigate along complex interorganizational pathways to acquire desired goods and services, such as better healthcare, shopping bargains, and recreational experiences.

This volume updates the second edition of Social Network Analysis by Knoke and Yang (2008). In addition to providing a general overview of fundamental methodological topics, we cover new developments of the past decade. Our approach is didactic, aimed primarily at graduate students and professionals in many social science disciplines, including sociology, political science, business management, anthropology, economics, psychology, public administration, public health, and human resources. College faculty could assign it as a text in graduate-level courses, use it for workshops at professional association meetings or summer instructional institutes, or study it to learn more about networks on their own. Graduate and advanced undergraduate students interested in social network analyses can read it to get a jump-start on their social network skills and intellectual aspirations. Professionals face many challenges in developing social network research, such as how to design a social network project, details and problems that may arise during network data collection, and alternative techniques for analyzing their social network data. Social network scholars may find this volume a useful brief refresher or reference book. For more advanced texts, we suggest Easley and Kleinberg (2010); Dorogovtsev and Mendes (2014); Lazega and Snijders (2015); de Nooy, Mrvar, and Batagelj (2018); and Newman (2010).

We frequently illustrate concepts and methods by referring to substantive social network research problems, citing examples from children’s playgroups to organizations, communities, and international systems. We tried to write with a precision and freshness of presentation using concise language that minimizes technical complexities. The book consists of five substantive chapters. Chapter 2 introduces fundamental network assumptions and concepts, as applied to a variety of units of observation, levels of analysis, and types of measures. It contrasts relational contents and forms of relations and distinguishes between egocentric and whole networks. The structural approach emphasizes the value of network analysis for uncovering deeper patterns beneath the surface of empirical interactions. Chapter 3 concerns issues in collecting network data: boundary specification, data collection procedures, cognitive social structures, missing data, measurement error, and collecting online social media and Big Data. In Chapter 4, we discuss basic methods of network analysis, including graphs and matrices; centrality, prestige, and power; social distance, paths, walks, and reachability; transitivity and cliques; and size, centralization, density, and different measure of equivalence for pairs of actors or entities. Chapter 5 gives an overview of more-advanced methods of network analysis, including ego-nets; clustering, multidimensional analysis, and blockmodels; 2-mode and 3-mode networks; community detection; and exponential random graph models. The final section concludes with some speculations about future directions in social network analysis.

After years of painstaking efforts, network analysts developed several computer packages to facilitate social network data collection and analyses. Softwares vary on many dimensions, such as operating systems, affordability, learning curves, and strengths and weaknesses. We attached an Appendix that summarizes some useful packages and contrasts them on those dimensions. We remain most impressed, however, with the breadth and user-friendly qualities of UCINET (Borgatti, Everett, & Freeman, 2002) as both a teaching and a research tool for smaller-scale social network analyses. Consequently, we used it to make this edition whenever we demonstrated social network analysis methods.