Gathering Social Network Data. jimi adams
decisions in the design of your study, the footnotes frame many of these debates and provide pointers to other resources intended to facilitate that further exploration.
I also want to mention here at the outset one key aspect that I will not really address in this book. This is a book on the methods of data collection, not on social network analysis (Knoke & Yang, 2007). There are literally dozens of good books available on social network analysis (SNA), and I don’t feel the need to compete with them or to replicate their work. If analytic approaches to network data are what you’re after, I’d point you to one of those books (for a few excellent examples, see Appendix A).1 The aims of this book are complementary to these others. In fact, I make a few assumptions in this book that you are familiar with many basic terms in social network analysis, to avoid unnecessarily breaking up this text with those definitions. However, for those of you who are new to the field or just want to be sure we’re working from the same assumptions,2 I provide a glossary in Appendix A that provides definitions for many key terms (appearing in bold where they first occur in the main text).
1 Currently, most treatments of gathering network data are single chapters in larger books, which in turn correspond with the single units in large SNA courses. Most people come to realize the need for gathering social network data after having some sense of the theoretical or analytic aims they have for that data. As such, I expect for most readers that this book will not be your introduction to the field of social networks, but you will come to it with some familiarity with SNA’s basics.
2 After all, this is a highly interdisciplinary field, and depending on which tradition you’re coming from, these various perspectives occasionally have multiple words that mean the same thing or use the same word to indicate different ideas.
Acknowledgments
I’ve often wanted a book focused on network data collection while teaching this material in courses and workshops over the past decade. So I’m grateful for Barbara Entwisle and Helen Salmon providing me the opportunity and guidance to write one for this series and for overseeing a review process that undoubtedly improved the book. Hopefully, others will find what’s here at least a fraction as useful as I’ve found the process of writing it.
Any project benefits from others’ feedback, and this book is no different. Throughout, I can see the fingerprints of several collaborators from over the years. Many of the ways I approach networks date to Jim Moody’s influence as my PhD advisor, when working with him provided my introduction to social networks research. In the years since, Ryan Light and David Schaefer have been regular sounding boards on basically all things networks. It’s hard to tell where much of my own perspective begins and theirs end at times. Except when it comes to any lingering misunderstandings or misrepresentation of ideas from the field; I manage those all on my own.
The organization of this material mostly stems from teaching opportunities I’ve had across a range of settings: courses at Columbia University’s Epidemiology and Population Health Summer Institute, American University, University of Colorado Denver, and the Inter-university Consortium for Political and Social Research’s Summer Program. In particular, a seminar on network data in health research at Arizona State University allowed me to take a much deeper dive into some of these topics than would otherwise have been possible. I’d also like to acknowledge the research assistance on elements of this book from Venice Ng Williams and Tatiane Santos, with whom I coauthored a chapter that provides the backbone for parts of this book (adams, Santos, & Williams, 2019).
A writing group and office space provided by the CU Population Center in the Institute of Behavioral Science at CU-Boulder were instrumental in completing this book. Additionally, with the willingness of SAGE, I posted a draft of this book for Open Review (M.Salganik & Baker, 2018), and I’d like to thank all who provided feedback on the manuscript via that site.
Finally, I’d like to thank a number of colleagues who read and provided feedback on draft sections of this book. Their feedback has substantially clarified the material presented here: Jason Boardman, Michał Bojanowski, Jill Harrison, Ryan Light, Chris Marcum, Ryan Masters, Ann McCranie, Sanyu Mojola, Stefanie Mollborn, David Schaefer, Jenny Trinitapoli, and Sara Yeatman. I hope you know I’m happy to reciprocate any time.
SAGE and the author would like to thank the following reviewers for their feedback:
Weihua An, Emory University
Bernie Hogan, University of Oxford
Bibhuti K. Sar, University of Louisville
Ashton Verdery, The Pennsylvania State University
About the Author
jimi adamsis an Associate Professor of Health and Behavioral Sciences at the University of Colorado Denver. His work focuses on examining social networks to understand how infectious diseases and novel ideas spread. He received his PhD in sociology from Ohio State University.
Chapter 1. Why Focus on Relationships?
What is a social network? The way I answer that question has changed a lot since I started out in this field. Early in grad school, I was generally met with blank stares whenever I mentioned that my research focused on social networks. In the early 2000s, most people actually didn’t have much of a reference point for what that term might mean. Since my research at the time focused largely on sexually transmitted infections (STIs), I’d usually say something about how I examined the patterns of risk behavior interactions among a group of people. This meant we were aiming to systematically capture who was potentially (or actually) exposing others to an STI. That requires finding data on who was having (what types of) sex with whom or sharing injection needles with which others, and ideally, we were able to determine when each of these behaviors was taking place.
In the years since, my own research, the field as a whole, and, perhaps more important, the public’s reference point on social networks has broadened considerably. I’ve started studying things like information flows through populations and adolescents’ friendships—a bread-and-butter topic for social network analysts. Scholars now study things ranging from kinship ties to telephone calls, face-to-face interactions of individuals to resource exchanges between states, social support provision to voting behavior influence, and innumerable other possibilities. And the ubiquity of social media sites like Facebook and Twitter often mean that instead of being confronted with blank stares when describing “social networks” as our research focus, social network scholars now are faced with explaining how that variety of topics we study differs from (or at times aligns with) what those sites entail.
Essentially, social networks are the collection of relationships or interactions between members of a population of social actors.1 And this book is about how we gather data on those networks. I’ll use the generic term ties to represent any of these relationships (or interactions) between a pair of actors. This term stems from the common use of visual representations in social network scholarship, wherein lines are used to represent how these ties link two actors together. Network scholars generally refer to these actors as nodes (or vertices), terms adopted from graph theory and the strategy of representing them within visualizations as points or dots. In many social network applications, the social actors we’re interested in are people, with relationship states (like friendships) or interaction events (like conversations) occurring between them. However, instead of individuals, other studies may focus on collective nodes like organizations (e.g., studying formal collaborations between them) or countries (e.g., trade patterns across them). Generically, just as a network tie can represent any number of tie types, a network node can represent an individual or collective actor.2
1 I’ll differentiate relational from interactional data below.