POWER, RELATIONSHIPS, AND SOCIAL NETWORKS

As classically formulated by Dahl (Reference Dahl, Bell, Edwards and Wagner1969), power is when person A gets person B to do something that he or she would not otherwise do on his or her own. It is clear from this definition that power is relational. That is, power only exists when considering interactions between and among individuals and groups. Returning to the seminal concept of power, as Crozier and Friedberg suggest, “Power can develop only through exchange among the actors in a given relation” (Reference Crozier and Friedberg1980, 20).

Even as political scientists have challenged Dahl's simple definition and formulation of power, the concept has always remained relational. For Bachrach and Baratz (Reference Bachrach and Baratz1962), power is not exercised solely in determining which decisions are made, but also by determining which issues are allowed into the public domain.Footnote ¹ They ask,

As this quote suggests, issues are overlooked in public discourse because of the situation of a particular actor or group in a larger web of interactions. An even more radical view of power, as espoused by Lukes (Reference Lukes1974) and adopted by scholars such as Gaventa (Reference Gaventa1980), emphasizes a relational view of power. While this view of power examines how power is exercised on action, it also applies to inaction. As Lukes notes, this third view of power suggests that actor A exercises power over actor B “by influencing, shaping or determining his very wants” (Reference Lukes1974, 23). Although some scholars reject this view of power as too manipulative or insidious, a social network perspective adopts a more realistic view that focuses on how relationships with others affect preferences and choices.

As Hafner-Burton, Kahler, and Montgomery (Reference Hafner-Burton, Kahler and Montgomery2009) note, traditional operationalizations of power are somewhat at odds with these conceptualizations, because they focus on the possession of resources. Such an approach unintentionally limits the concept of power in ways that miss the role that connections play between actors; for example, the presence of a large military but the absence of an actor to threaten with these capabilities is a fairly empty understanding of power. Instead, Hafner-Burton, Kahler, and Montgomery (Reference Hafner-Burton, Kahler and Montgomery2009) claim that resources only make sense in the context of an actor's web of connections, which can affect both the resources that he or she has and his or her ability to use them effectively. These scholars specifically argue that SNA provides appropriate tools and concepts for bringing relationships back into our operationalizations of power.

We agree. A network approach to understanding politics explicitly adopts a relational perspective on any political process and implicitly adopts a view of power that is different from standard operationalizations. So while many in our discipline conceive of their work in terms of power—whether social status for individuals or natural resources for a state—very rarely do they explicitly consider how that power derives from or is influential because of relational context. Once we begin to consider these questions, it becomes clear that a failure to investigate these relations and their import impedes our ability to understand and explain important political and social problems.

RELATIONAL ANALYSIS AT THE CORE OF POLITICAL SCIENCE

It is one thing to argue that power is inherently relational but quite another to suggest that this kind of work ought to lie at the core of the discipline. Scholars with a critical eye can reasonably suggest that outcomes that appear to be driven by relationships among actors can, in fact, be explained by other theories. Alternatively, others can argue with cause that overemphasizing relationships and networks discounts the importance of agency and self-determination—in short, they can object to the absence of self-determination. Still another objection to our argument is that the inherent need of relational approaches to study problems holistically represents a step backwards for a discipline that has recently embraced the scientific criteria of causality and experimentation. Each critique has merit. But rather than seeing them as reasons to ignore networked views of political power in political science, we see them as opportunities for political scientists to add their own insights to the existing understandings of social networks concepts and methods.

The first objection to relational analysis derives from the straightforward principle that correlation is not causation. A scientist steeped in the network tradition sees correlation between attributes of the network and individual behavior as evidence of causality; scientists from other traditions see the same evidence as indicative of other processes, such as self-selection driven by conflict avoidance.Footnote ² We agree with skeptics of social influence theories that unambiguous evidence of network effects is difficult to attain, but we disagree that this challenge should dissuade us from adopting a relational approach to political science. Indeed, to the extent that we want to understand power in politics, an important part of the discipline is trying to understand which relationships are influential, which ones are merely supportive, and which ones are simply irrelevant. Only the serious pursuit of strong evidence of social influence in the face of omitted variable bias, homophily, and a host of other methodological challenges to the establishment of causality allows us to make such distinctions.

We argue that such a focus is preferable to other approaches that either assume independence of actors or else make the assumption that dependence only flows from rules and institutions. Although there is an old, vigorous debate about whether assumptions in models should be realistic or just useful (e.g., Friedman Reference Friedman1953), we side-step this dispute by claiming that the assumption of independence of individuals limits the study of political science to certain questions. Relaxing this assumption brings social relations to the fore and suggests a host of new directions for political science research. Whether a social network approach is more “realistic” or “better” is ultimately an empirical question that can be resolved only through a relational versus an individualist research program (see Bueno de Mesquita Reference Bueno de Mesquita2010, 389–402).Footnote ³

Such a discussion flows naturally into the expression of concerns that a relational approach leads the discipline down a path of social determinism. To state this differently, the objection is that prioritizing questions about the influence of networks and the origins of their structures significantly downplays the role of individual agency in politics. This critique is grounded on both intellectual and political grounds and is legitimate in both cases. However, as a practical matter, there is no reason to assume that a focus on the attributes of network structures is inconsistent with the principal of methodological individualism. To be sure, the tools for analyzing networks are more advanced when attempting to unpack structures and predict relationships as opposed to understanding how these structures and relationships interact with the attributes of the actors in the networks.

Yet here is a case in which political science can make a strong contribution to the study of networks. With so many of us studying political and economic systems that are best described in terms of how much agency is allowed for individuals, it would be untenable for us to overlook the hard questions about how this agency is balanced against the impact of networks. Only by focusing clearly on the degree to which relations constrain agency can we begin to fully explain our depictions of what the concepts of liberty, freedom, and agency mean in different political systems and contexts.

If we are right that political science can expand its ability to understand fundamental issues of power by confronting the issue of social influence and balancing this element against individual agency, we still leave open the question of whether we can investigate this subject rigorously. Although significant debates rage about what constitutes rigorous research, these arguments often rest on epistemological differences over the meaning of causation and interpretation. SNA and its accompanying methods should not be exempt from these same debates, but neither should they be rejected outright on the basis that fundamental assertions of interdependence complicate our studies and make causal effects difficult to establish.

To take dependence seriously often means placing questions of external validity, randomization, statistical independence, and even contextualization of data in the background. Instead, such a focus suggests a different set of criteria, such as completeness of the network, identification of the appropriate links, and connection of knowledge about relationships with information about the parties involved. Thus, the study of networks—including large scale-networks such as co-sponsorship links in a legislature or social media connections—often requires us to rethink standard assumptions about what constitutes meaningful evidence of an interesting social process. Only after this reconsideration does it make sense to bring those criteria back to the foreground. Yet again, we feel that this process presents more opportunity than opposition. Trying to challenge standard assumptions about what is a meaningful relationship for politics is a potentially fruitful way to begin expanding the methodological toolbox of our discipline.

HOW TO ADVANCE THE AGENDA

For a truly relational political science to succeed, we must recognize that this approach can provide a common thread across the subfields. We hope that the articles in this symposium—each devoted to explaining the importance of political networks in different corners of our intellectual cafeteria (Almond Reference Almond1988)—will demonstrate exactly these kinds of connections while being faithful to the questions that motivate those subfields.

But what other steps are necessary? We argue here for three pragmatic steps that could go a long way toward incorporating a relational view into our thinking about politics and political science. First, we must begin to integrate social network topics into the core of our methods training in the discipline. While a great deal of the SNA methodological core is quantitative in nature, we also believe that this advice holds for training in qualitative methods. The core methodological problems of SNA as practiced in other fields involve the question of how to model dependence among actors, the importance of relational patterns that show up regularly in social processes, and the fact that what looks chaotic and random often emerges from simple and elegant problems. These issues are not well incorporated into our research training, even though large bodies of literature in sociology, economics, medicine, computer science, biology, and even physics exist that suggest that the world is rife with such complexities.

Second, we believe that the discipline should begin to rethink the standards we use to determine what data are “good” and what data are “bad.” To be clear from the outset, we are not arguing that relational studies are lacking in rigor or that the criteria used to judge nonrelational studies are unreasonable. Rather, we suggest that the data that can be used to study relations are often of a different nature than many more standard forms of evidence. For example, to understand the role of networks in many areas of American politics, we may have to accept that external validity will be difficult if not impossible to achieve.Footnote ⁴ Important scientific criteria to which we should adhere have been established for relational analyses, set forth in a voluminous literature in other disciplines.

However, we should not, as it were, throw the baby out with the bathwater. As techniques for gathering network data on a large scale become more available, we believe that the characteristics of good relational research and standard political science study will converge. In the meantime, we argue that good relational studies should be judged on grounds appropriate for the questions they raise. Consider, as an example, that the impact of the Columbia sociology research on networks and voting behavior first conducted in the 1950s is still felt decades later, even though that study was constrained to a representative sample of residents in unrepresentative Elimira, New York (Berelson, Lazarsfeld, and McPhee Reference Berelson, Lazarsfeld and McPhee1954). Simply by raising the question of how cross-cutting networks influence voters, this research made a valuable contribution by opening the door to new concepts and forcing us to think more critically about the conditions under which political conversations inhibit tolerance, participation, and voting (Mutz Reference Mutz2006; Huckfeldt et al. Reference Huckfeldt, Johnson and Sprague2004).

Finally, and perhaps most importantly, the pathways toward professional success in the discipline should not only encourage but also seek out the best relational work that is being done at any given time. Here we are referring, of course, to the choices that are made about what articles are accepted to journals, what kinds of jobs are listed by universities and other institutions, and similar indicators. Rather than dismissing relational studies of politics as better suited to other disciplines and journals, we should be encouraging this work within the boundaries of our own.

We have seen some encouraging steps in this area. A new organized section of the APSA devoted to political networks has been created, meaning that there is now space set aside at our national meeting for the best research in this vein. Through the generous support of the National Science Foundation (NSF), three summer conferences have now been devoted to training in social network methods and the presentation of political networks research.Footnote ⁵ American Politics Research devoted a special issue to political networks, with eight articles spanning subjects across the subfield.Footnote ⁶ Temple University Press has begun to publish a political science series entitled The Social Logic of Politics. Despite these advances, more progress along these lines is welcome.

Author Bios

Ramiro Berardo is a researcher with the National Council of Scientific and Technological Research (CONICET) at the Catholic University of Cordoba in Argentina. He is also an assistant research professor with the School of Government and Public Policy at the University of Arizona. He can be reached at berardo@arizona.edu.

Paul A. Djupe is an associate professor of political science at Denison University, specializing in American public opinion and political behavior, with particular interests in the political influence of religion and the nature of social influence. He is the co-author of three books, and his work has appeared in journals such as the American Journal of Political Science, the Journal of Politics, American Politics Research, and Political Research Quarterly. He can be reached at djupe@denison.edu.

David Lazer is an associate professor of political science and computer and information science at Northeastern University and a visiting scholar at Harvard University. He is a cofounder of the Political Networks conference and APSA section. Within political science, his research has been published in the American Political Science Review, World Politics, and Journal of Politics; outside of political science, he has been published in such journals as Science, Proceedings of the National Academy of Science, Administrative Science Quarterly, and Law and Society Review. He can be reached at d.lazer@neu.edu.

Scott D. McClurg is an associate professor in the department of political science at Southern Illinois University. He is the chair of the APSA's organized section on Political Networks. His published research on social networks and political participation appear in American Journal of Political Science, Political Research Quarterly, Political Opinion Quarterly, and American Politics Research. He can be reached at mcclurg@siu.edu.

Ami Pedahzur is a professor in the department of government at the University of Texas at Austin, where he is also head of the TIGER (Terrorism, Insurgencies, and Guerrillas in Education and Research) Lab. He can be reached at pedahzur@austin.utexas.edu.

Arie Perliger is an associate at the Combating Terrorism Center and an assistant professor in the department of social sciences at the U.S. Military Academy at West Point. In the past eight years, he has published four books and 15 articles and book chapters in the fields of terrorism, political violence and extremism, and the principal ways that democratic states respond to these challenges. He can be reached at aperliger@gmail.com.

David A. Siegel is an assistant professor of political science at Florida State University. His research focuses on collective action, political violence, and electoral dynamics. He can be reached at dsiegel@fsu.edu.

Anand E. Sokhey is an assistant professor of political science at the University of Colorado at Boulder. He specializes in American politics, with particular interests in political discussion, social networks, and contextual influences as they pertain to voting behavior, political participation, and opinion formation. His work has appeared in journals such as the American Political Science Review, the American Journal of Political Science, and American Politics Research. He can be reached at anand.sokhey@colorado.edu.

Joseph K. Young is an assistant professor in the department of political science at Southern Illinois University. His research interests include terrorism, civil war, and human rights. Recent articles appear in the Journal of Peace Research, Political Research Quarterly, and International Studies Quarterly. He can be reached at jkyoung@siu.edu.