“Social networks,” reads a passage in the new book, Social Network Analysis: Methods and Examples, “have been a defining feature of society since the early dawn of humanity – people have always interacted with each other or have made friends of enemies.” But the widespread use of the term “social network” is a creature of the internet, even if the academic analysis of social networks – think Simmel and Durkheim — predates its arrival in the vernacular. The first edition of SAGE’s own ‘little green book’ on social network analysis, for example, came out in 1982.
Regardless of its roots, the analysis of these social networks — how these human connections affect the perceptions, beliefs, and behaviors of individuals, groups and organizations — has found wide applicability across a range of endeavor, in the social sciences and in non-academic milieu such as national security, finance and management.
Song Yang, the lead author of that new book on social network analysis, here answers some questions about social network analysis, its applications and how to teach its use. Yang is a sociology professor in the Department of Sociology and Criminal Justice at the University of Arkansas. His research interests have long included a focus on social network analysis, and in 2007 he and David Knoke co-authored a second edition of that ‘little green book’ — the colloquial name for the short works in the Quantitative Applications in the Social Sciences series – on social network analysis. Yang, also holder of the Qianren Professorship with School of Sociology and Political Science at Shanghai University, answered our questions while on sabbatical in China.
What are the strengths of social network analysis? What are its weaknesses?
Strength: People make decisions, find best deals, conduct daily business, and even maintain mental health using their social network contacts. Traditional social science models do not account for social network context in explaining people’s behaviors, missing out an important explanatory framework. Social network analysis (SNA) offers compelling arguments to shed light on people’s choices and actions. The development of SNA is fueled by explosive growths in methodology, which is largely due to 1) interdisciplinary collaborations, 2) powerful and inexpensive PCs, and 3) computer supported social network developments.
Weakness: SNA is used indiscriminately from metaphorical references to formal mathematical modeling. To the general public, it can be confusing, as many assume social networks are the same as Facebook contacts, while others relate it to people they have dinner with. The mixed use of SNA by academics, while strengthening academic development of SNA, does not help to present a clarified version of SNA to the general public.
Another issue confronting SNA scholars is causality ambiguity – does one have friends who smoke first, then form networks with its smoker friends (behavior causes network formation)? Or does one have social network of friends who are smokers, then he/she become a smoker (network causes behavior)? To the extent causality ambiguity is an issue for social science research in general, it is a particular challenging issue confronting social network scholars. I would call this not only challenge, but also an opportunity to develop solutions, thus improving SNA framework.
Social networks, and even the academic study of them, existed long before the rise of what the public might regard as social networks, i.e. Facebook or Twitter. But how have those entities affected social network analysis? How has big data itself affected SNA?
We call those networks such as Facebook and Twitter ‘computer supported social network’ (CSSN) or ‘social media network’ (SMN). The development of CSSN/SMN helped SNA, as the general public started to pay close attention to SNA. However, as soon as they look at the SNA, they realize that is not what they expected SNA to be. Certainly the situation changes, as many SNA scholars attempt to analyze CSSN or SMN data, where they encounter challenges of data collection, data mining, data storage and data analytics. Here, a cross-disciplinary effort involving social scientists and hard scientists (computer science/EE) is needed to facilitate capabilities of data mining and data analytics.
The same can be said about big data, a concept started mostly in computer science. Its growth overlaps with SNA, as CSSN and SMN commonly generates zillions of messages connecting billions of users. Big data is both a challenge and an opportunity to interdisciplinary collaborations to delve into; I think it offers a bright path for SNA to continue to grow and to be at the center of social science paradigm.
I am fascinated with the concept leadership. Of course being a leader gives one authoritative power to command their subordinates to do their jobs. The underlying logic is threat – if the subordinate refuses, he/she can be fired. But how often do leaders invoke such logic to exercise leadership? In fact, when leaders invoke threats to make sure their commands are being obeyed, the leadership nears its end.
Here, social network analysis offers compelling alternative explanation to the exercise of leadership. For example, another way to influence people’s behaviors is through consultative relations, as opposed to authoritative threats. A leader needs to connect with subordinates with not the authoritative (vertical) relations, but peer advice-giving ties. Many people would respond very positively when they perceive that the leader is offering them best advice in their situations, and they will react pretty negatively when they think the leader is commanding them with authoritative sticks.
There are several books on social network analysis, including your own ‘little green book’ on the subject from 2007. What does Social Network Analysis: Methods and Examples offer that extends or improves the existing scholarship?
The “little green book” of our SNA book offers a quick succinct overview on SNA. It gets to the points very quickly, but many topics are mentioned without much elaboration. For SNA scholars, our little green book of SNA is sufficient for design, implementation, and analysis of social networks. However, for many starters, they might want to see something that covers SNA much fuller in each of the topic mentioned. Our book, Social Network Analysis: Methods and Examples, should reach broader range of general public than does the little green book.
In addition, a unique feature of our SNA Social Network Analysis: Methods and Examples is that it offers four substantive chapters illustrating how scholars from diverse fields (management, criminal justice, public health, and political science) can fruitfully apply SNA to extend their respective analyses. As far as I know, this is the very first book that discusses the extensive applications of SNA in varied academic areas.
Looking at the methodology of SNA, is there something that students often struggle with? How do you address that?
Two things that students struggle with a lot: 1) data analysis, and 2) connecting the dots.
1) Data analysis has been a tough topic for social science students in general, and of course SNA is no exception. While the descriptive SNA methods (density, centrality, etc) are still posing some challenges to students, new one such as exponential random graph modeling (ERGM) require advanced training in statistics and computer science, presenting great challenges to students and teachers alike. It seems that the decoupling between methodology advancement in SNA research and classroom teaching persists — or even expands — presenting great challenges to teaching methodologies.
2) In the second point, many of my students are stunned when I move from metaphorical use of social network to concrete design, collection, and analysis of social network data. Then the class jumps to application of SNA in various research areas. Although in curriculum design, those are integral ingredients of the SNA topic, in classroom teaching, students may have hard time seeing the connections between those elements. It is extremely important for teachers to present the “big picture” to students all the time.
Your own education includes a master’s in computer engineering. How much technical ability do I need to engage in social network analysis? What tools do I need?
Computer science has a chapter on “graph theory,” which addresses network optimization (in their fields, computers are nodes, networks are ties connecting computers). However, the basic idea is somewhat similar between the two fields. In terms of technical preparation to engage in SNA studies, it requires more conceptual clarification than technical preparation. It certainly needs technical stuff, but knowing what to do each step of the way in SNA process is key to seeing it through. Presently, software that facilitates SNA are UCINET and SIENA etc., which makes conceptual clarification ever more important.
A lot of disciplines and practitioners use social network analysis, both in social science and in government and industry. Does SNA differ in the way a sociologist uses it compared to how a national security figure (or other non-academic) might use it?
I do not know much about SNA in government or industry use. One key difference between academic and government in SNA is that while academics encourage public forum of free exchanges of ideas and open discussions of various methods, government often conducts covert operation using SNA. I guess it is because they have different missions – academics emphasize educational function, government wants its utility.
What do you predict for the future of SNA?
I know it has become a buzzword that everybody is talking about. But big data have significant overlap with SNA, especially in the domains of CSSN or SMN. People spend more and more time on their computers or smart phones for various functions, social networking is a significant part of such technology use by general population. It is too big a hole to be overlooked by social scientists. But again, data mining, storage, and analysis present great challenge and opportunities for social scientists that shall work with computer/electrical engineering scientists to explore this huge area.