%20-%20https://sketch.com%20--%3e%3ctitle%3e3ECBBF50-FD08-412A-943D-2CFC5E025A16%3c/title%3e%3cdesc%3eCreated%20with%20sketchtool.%3c/desc%3e%3cg%20id='Stylesheet'%20stroke='none'%20stroke-width='1'%20fill='none'%20fill-rule='evenodd'%3e%3cg%20transform='translate(-926.000000,%20-668.000000)'%20id='BUTTONS'%3e%3cg%20transform='translate(100.000000,%20597.000000)'%3e%3cg%20id='BTN-/-RIGHTS'%20transform='translate(816.000000,%2065.000000)'%3e%3cg%20id='ICON-/-Rights'%20transform='translate(10.000000,%206.000000)'%3e%3cpath%20d='M13.5752841,0%20L1.34943182,0%20C0.59375,0%200,0.586666667%200,1.33333333%20L0,22.6666667%20C0,23.4133333%200.620738636,24%201.34943182,24%20L17.6505682,24%20C18.40625,24%2019,23.3866667%2019,22.6666667%20L19,5.36%20L13.5752841,0%20Z%20M13.5752841,1.89333333%20L17.0838068,5.36%20L13.5752841,5.36%20L13.5752841,1.89333333%20Z%20M1.34943182,22.6666667%20L1.34943182,1.33333333%20L12.1988636,1.33333333%20L12.1988636,5.36%20C12.1988636,6.10666667%2012.7926136,6.69333333%2013.5482955,6.69333333%20L17.6235795,6.69333333%20L17.6505682,22.6666667%20L1.34943182,22.6666667%20Z'%20id='Shape'%20fill='%23595959'%20fill-rule='nonzero'%3e%3c/path%3e%3ccircle%20id='Oval'%20stroke='%23595959'%20stroke-width='1.5'%20fill='%23595959'%20cx='9.5'%20cy='13.5355339'%20r='5.5'%3e%3c/circle%3e%3cpolyline%20id='Path-2'%20stroke='%23FFFFFF'%20stroke-width='1.5'%20transform='translate(9.500000,%2013.035534)%20rotate(45.000000)%20translate(-9.500000,%20-13.035534)%20'%20points='8%2016.0355339%2011%2016.0355339%2011%2010.0355339'%3e%3c/polyline%3e%3c/g%3e%3c/g%3e%3c/g%3e%3c/g%3e%3c/g%3e%3c/svg%3e)
The academic community studying terrorism has changed dramatically in the past decade. From a research area that was investigated by a small number of political scientists and sociologists and employed mainly descriptive and qualitative studies that resulted in limited theoretical progress (Schmid and Jongman Reference Schmid and Jongman1988; Crenshaw Reference Crenshaw2000), it has in a short time become one of the more vibrant and rapidly developing academic realms in the scholarly world today. Scholars from different branches of the social sciences have engaged in an effort to unravel this phenomenon, introducing new theoretical outlooks, conceptualizations, and methods.
The descriptive and explanatory potential of social network analysis (SNA) in the study of violent political groups attracted a number of new students in the wake of the September 11 attacks (Van Meter Reference Van Meter2001; Carley, Ju-Sung, and Krackhardt Reference Carley, Ju-Sung and Krackhardt2002; Krebs Reference Krebs2002). Yet, even though their studies showed strong potential in providing significant insights about the structures and internal processes of terrorists groups, the use of SNA in the study of political violence has remained quite limited and still amounts to only a small fraction of the research in the field. Our experience in presenting SNA of violent groups in various platforms and events has led us to conclude that this narrow scope is a result of two factors that sustain each other. The majority of political violence students have very limited acquaintance with the rationale, main concepts, and methodological tools of SNA; hence, many are still reluctant to exercise SNA in their studies and consequently tend to express doubt regarding its efficiency and relevance for the study of complex social phenomena.
This article is not methodological per se in the sense that our goal is not to provide a methodological introduction to SNA. We do strive, however, to clearly present the advantages of this approach for the study of terrorism and related fields, as well as its main, relevant methodological tools and concepts, by using pertinent and intelligible examples. These examples illustrate how network analysis complements conventional approaches to the study of political violence and how it can provide important information about the characteristics of the group structure (and how this structure influences members' motives, behaviors, and the outcome of their actions), its recruitment processes, its evolution, and the division of political and social power among its members. We hope that this exploration will encourage more scholars to incorporate SNA into their studies and, consequently, will further our understanding of the processes, causes, and implications of political violence.
USING SNA IN THE STUDY OF POLITICAL VIOLENCE AND TERRORISM—WHY AND WHEN?
The use of SNA in the study of political violence in recent years can be attributed to several conceptual and methodological developments. The first is the growing acknowledgment within the academic community of the important association between the group's dynamic and social structure, and its members' motivations and behaviors (Sageman Reference Sageman2004; Sageman Reference Sageman2008). Sociological and organizational texts began to incorporate network perspectives in the early 1970s, emphasizing this approach's importance in understanding collective action (Granovetter Reference Granovetter1973; Snow, Zurcher, and Ekland-Olson Reference Snow, Zurcher and Ekland-Olson1980; Perrow Reference Perrow1986). Until recently, the study of political violence tended to neglect this dimension of violent groups, since it focused on two major elements: the individual and the organization. Scholars using the first approach investigated psychological and sociological tendencies to decipher the profile of terrorist group members and understand their motives (Victoroff Reference Victoroff2005). Scholars applying the second approach used organizational theories to explain the strategic and tactical decisions of terrorist organization elites (Crenshaw Reference Crenshaw1981; Moghadam Reference Moghadam2003; Pape Reference Pape2003; Bloom Reference Bloom2004; Sandler and Enders Reference Sandler and Enders2005). Yet, three decades of research yielded inconclusive answers regarding terrorists' profiles, and a mounting number of empirical observations revealed that a growing proportion of violent attacks by political groups in recent years were not perpetrated by hierarchical paramilitary organizations, but rather by small, informal, and non-hierarchical social frameworks, which tend to be more adaptive and dynamic (Raufer Reference Raufer2003; McAllister Reference McAllister2004; Sageman Reference Sageman2004; Sageman Reference Sageman2008).
The growing prominence of these empirical observations led an increasing number of scholars to shift their attention from the individual or organizational levels of analysis to the social dynamic within the groups and impelled them to investigate how the type and intensity of ties among the group's members and their multifunctional nature are associated with the group's radicalization, the emergence of common identity and ideological commitment, and engagement in violent activities (Rodriquez Reference Rodriquez2005; Koschade Reference Koschade2006; Pedahzur and Perliger Reference Pedahzur and Perliger2006). As we further demonstrate in this article, SNA is well suited for such investigations.
A related development was the growing popularity of revisionist views regarding the timetable of the process that leads to acts of political violence. Until recently, conventional wisdom in the field assumed that the cause of the violence preceded the formation of the violent group, which was established in response to an external trigger and aimed to obtain identifiable political goals (Crenshaw Reference Crenshaw1992; Pape Reference Pape2003; Victoroff Reference Victoroff2005). Nevertheless, in the face of new evidence garnered in recent years, especially from studies focusing on violent religious groups in Western countries (Hoffman Reference Hoffman and Lesser1999; Juergensmeyer Reference Juergensmeyer2003; Stern Reference Stern2004), scholars have found that it is not uncommon for the formation of the social framework responsible for the violence to precede the cause of the violence (Sageman Reference Sageman2004; Sageman Reference Sageman2008; Rodriquez Reference Rodriquez2005; Pedahzur and Perliger Reference Pedahzur and Perliger2009).
In other words, many of the social networks engaging in terrorism existed long before they became involved in terrorist activities. Hence, understanding the motives and processes that led the group to engage in political violence requires a look beyond the apparently causal relationship between the causes of the violence and the violent activities. Consolidation of a cause is just one stage in the group's long slide toward violence. This connection again reflects the idea that social processes within social networks are highly relevant in understanding political violence.
From a methodological perspective, contemporary students of political violence enjoy superior access to information and data about violent groups. Since September 11, growing numbers of media outlets have increased their coverage of terrorist incidents and groups. This rise in attention, combined with a striking increase in efforts and resources invested in data collection on these groups by academic and governmental agencies in recent years (e.g., START at the University of Maryland and TIGER at the University of Texas at Austin), have simplified the adaptation of research methods—such as SNA—that demand high-resolution information about terrorists and their groups. Last, the growing popularity of terrorism as a research area, which has attracted social networks experts, has also contributed to the growing use of SNA in the study of political violence (Krebs Reference Krebs2002).
The previously mentioned factors responsible for the growing use of SNA also hint at the characteristic manifestations of political violence that are particularly suited for investigation using this approach. The first is almost trivial: violence or political action is a result of collective action—an output of a process—which is an action of a group of actors who interact with each other on some level. Lone wolves like Theodore Kaczynski or groups like the one behind the Oklahoma City bombing that consist of very few members (two to four) are naturally less suitable for SNA. On the other hand, groups that consist of more than dozens of members, such as Hamas or Aum Shinrikyo, are also problematic to study from this perspective, since it is difficult to accurately map the relations among all the actors of the clandestine group, as well as recruitment paths and changes in types of relations. In the next section, we elaborate on the methodological facets of mapping and defining the boundaries of the terrorist network.
Another element influencing the network's suitability for investigation using SNA refers to its level of institutionalization. While hierarchical paramilitary organizations such as the IRA, the Red Brigades, Hezbollah, or the Liberation Tigers of Tamil Eelam usually include social networks, the social processes within them are not the most relevant factor for explaining group activities, target selection, and motives. In these cases, the relevant processes occur within the higher echelon of group leaders and can be better understood through the use of theories that focus on organizational processes, leadership behavior, and rational decision making. By contrast, when group interactions are less formal, as in the cases of the Red Army Faction, the Order, and the Hamburg cell; when the division between the different echelons of the group is less strict; and, in general, when there are more interactions and transmission of information among group members and echelons, SNA becomes more efficient in uncovering the group's dynamic, structure, activities, and motives.
HOW TO STUDY VIOLENT SOCIAL NETWORKS
When practicing SNA, we evaluate the targeted social phenomenon as a product of a (social) framework that includes actors (in the case of terrorist groups, these actors are group members, who are represented by nodes in a graphical representation of the social network) and ties (relations or interactions between different actors). Hence, in contrast to other quantitative methods that tend to focus on the description and aggregate analysis of the attributes of those actors who make up the research population, SNA assumes that to comprehend the social phenomenon, it is more conducive to map out and analyze the system of ties among the various actors and the ways in which these relational patterns shape actors' activity, decision making, group dynamics, and, eventually, the outcome of the group's collective action (Wasserman and Faust Reference Wasserman and Faust1994; Scott Reference Scott2000). In the following paragraphs, we epitomize how this analytic approach can be applied to the study of violent political groups.
Mapping the Group
The process of mapping the networks involves two critical methodological decisions: which actors to include and how to measure or categorize the ties between them. Two basic approaches to solving the first dilemma are common in SNA studies (Wasserman and Faust Reference Wasserman and Faust1994). The first is based on the actors' self-definition of the boundaries of the group. Individuals who are considered by the members as belonging to the group are included in the group. While this approach can be efficient in cases in which the researcher has access to the group members, it can be problematic when dealing with clandestine networks. The nominalist approach, which is more practical for the study of violent groups, is based on the theoretical focus of the researcher, who shapes the boundaries of the group based on his or her interests. Naturally, in the case of terrorist groups, researchers strive to include all actors who take part in either the group's activities or the social processes leading up to the violent activities. However, this approach raises the question of what constitutes significant participation in these processes.
The cases of the social networks responsible for the November 2003 attacks in Istanbul and the July 2005 attacks in London illustrate the dilemma. In the 2003 case, some group members were in contact with the top leaders of al-Qaeda during the initial planning stages of the operation. In the 2005 case, spiritual leaders were in contact with the group members, although they played a limited (if any) part in the operational execution of the attacks. What methodological criteria can we use to determine whether the al-Qaeda leaders in the first case and the clerics in the second case should be considered part of the terrorist network?
Several approaches could be suggested. The first includes all actors who were necessary to perpetrate the attack. A more inclusive perspective would involve all actors who assisted in the execution of the attacks, including actors who were part of the group for a short time, individuals who were present in the formation stages of the group but did not participate in the decision-making processes, or spiritual leaders who simply provided moral support for the actual perpetrators. By contrast, an exclusive outlook would include only actors who were long-time members of the group, participating consistently in both its activities and decision-making processes and maintaining continuous relations with other members. While these different approaches for drawing the boundaries of terrorist networks are all valid, we think that considering the dynamic nature of a member's status within such groups (in which peripheral members can become important actors in a short period of time), the methodological edge lends itself to the inclusive approach. Another option is to create progressive representations of the networks that refer to different time frames—in which case, the inclusive approach would better fit mapping the network in its initial stages. The number of members in the early stages is relatively small, most are significant, and any actor has the potential to become a prominent figure within the network. In the operational stages of the group's activities, a more exclusive approach would provide more valid representation of the network. In these stages, the chances are high that the group will include many passive supporters and passersby whose inclusion could disrupt the real social network, which is at play and is primarily responsible for the execution of the violence.
After deciding which actors belong to the network, a decision must be made regarding the categorizations of tie types. While ties may have different characteristics—binary or nonbinary, symmetrical or asymmetrical (i.e., equal strength in both directions or stronger in one direction), negative or positive—the measurement of tie strength seems most relevant for understanding the social dynamic within violent networks. Granovetter's (Reference Granovetter1973) classic and groundbreaking study essentially differentiates between weak ties (relationships with acquaintances) and strong ties (family and close friends), but other, more recent studies have tried to develop more sensitive categorizations. Krebs (Reference Krebs2002), one of the first scholars to analyze the September 11 network, bases his categorization on the amount of time spent together by a pair of members (living together, traveling together, short meetings) to generate three levels of tie strength. Similarly, Pedahzur and Perliger (Reference Pedahzur and Perliger2006) use three categories to distinguish between family relations, long-term friendships, and acquaintances. Brams, Hande, and Ramirez (Reference Brams, Hande and Ramirez2006) offer a more sophisticated approach using directional ties to detect the flow of information within the network and the level of influence of each actor on other actors. While different researchers have adopted different approaches—mainly based on a combination of the magnitude and frequency of interactions among actors according to their theoretical tendencies and the level of resolution of the data they were able to obtain—there is little doubt that tie strength is an important factor in explaining different processes within the network. Rodriquez (Reference Rodriquez2005), for example, illustrates the important strategic advantages of weak ties among cells of terrorist groups, given that this structure provides members with more flexibility, enhances their survivability when one of the cells is exposed, and makes them less visible. Alternatively, he and other researchers have illustrated the importance of strong ties within network subgroups, which facilitate solidarity, commitment, and cooperation, as well as reduce disagreements (Sageman Reference Sageman2004). Finally, some studies have drawn connections between the type and number of ties a member has and his or her role within the group (Rodriquez Reference Rodriquez2005; Pedahzur and Perliger Reference Pedahzur and Perliger2006).
Division of Power within the Group
One of the main advantages that SNA provides for students of terrorist groups is the capability to uncover informal divisions of influence and social capital within the group, which, in turn, influence the group's internal political and social processes and the outcome of its activities.
Whereas highly institutionalized and hierarchical organizations concentrate political and social power in the top echelon of identifiable and established leaders, the situation is substantially different in less formal and hierarchical groups. In these cases, instead of discovering a formal leader, we try to detect those actors who possess unique characteristics that make them significant in the continuous operation of the group. The first role is the classic hub, or an actor that has a significantly higher number of ties than other members (i.e., high level of centrality). Such a member is often responsible for coordinating the group's activities and recruiting members, and has the ability to manipulate the flow of information within the group more easily than others (Koschade Reference Koschade2006). As an example, in the case of the Israeli Jewish Underground terrorist group (see figure 1), the hubs (Livni, Natanzon, Etzion) intentionally did not reveal the long-term, highly extreme goals of the group to some members in order to not alienate them from participating in short-term, less extreme operations. The hubs were able to conceal this information not just because of their high level of centrality, but also because entire sections of the network were dependent on them for connection with other sections (high level of betweenness). Since a relatively high number of members were tied to the network via a specific hub, this orientation also implies that the hubs were responsible for recruitment processes. Finally, detection of the hubs is important because it allows us to uncover the power divisions within the network, determine who possesses more social capital (Bourdieu Reference Bourdieu and Richardson1986), and better understand the motives behind the group's actions. As demonstrated by Pedahzur and Perliger (Reference Pedahzur and Perliger2006), by looking at the characteristics and history of the hubs, the researchers were able to detect the motivations behind studied networks of Palestinian suicide bombers, epitomizing that the hubs will, in many cases, use these networks to promote personal or local political interests.
The “Jewish Underground” Terrorist Group
Source: Ami Pedahzur and Arie Perliger, Jewish Terrorism in Israel (New York: Columbia University Press, Reference Pedahzur and Perliger2009)
The case of the Jewish Underground also illustrates the existence of other variables beyond a high number of ties in determining the criteria for detecting informal leaders. While some groups present a dense structure, others are constructed as a bundle of connected subgroups. In the latter, actors in strategic locations who serve as connectors between different subgroups possess significant power and are crucial for the survival of the network. Although these actors do not have to be connected to a large number of members, they can veto almost any operation that requires the cooperation of the different subgroups.
Finally, terrorist groups generally include a number of actors who neither have a particularly large number of ties, nor are connectors between different parts of the network, but who are situated in a strategic location in terms of their proximity to hubs or large numbers of members within the network. Hence, these members have a high level of access to information and resources.
How can these three key kinds of actors be detected? How can we compare the influence of different actors in the same or different networks? The most common techniques for evaluating the actor's role and power within the network are measures of status, sometimes referred to as centrality measures. Several prominent, relevant measures that can be effective in the study of violent groups are degree of centrality, closeness, and betweenness. A similar measure was introduced by Brams, Hande, and Ramirez (Reference Brams, Hande and Ramirez2006) that developed the concept of influence as a function of the actor's importance within the network, which is, in turn, determined by the quantity of his or her ties and their direction relative to other actors.
Understanding the division of power within the network can help us test theories about the decision-making processes within terrorist groups, assess the validity of top-down or bottom-up theories, and understand how the internal socialization operates or is activated. Finally, it is important to not overlook the significance of such analysis for developing counterterrorism models (see, e.g., Carley, Ju-Sung, and Krackhardt Reference Carley, Ju-Sung and Krackhardt2002), as these type of analysis enable understanding of how such networks could be destabilized or dismantled. For example, the case of the Jewish Underground illustrates that operational hubs—who retain operational knowledge concerning weapons, explosives, and tactical planning—are more crucial to the survival of the network than ideological ones. In this case, the retirement of the operational hub led to the collapse of the network, although the ideological leaders wanted to keep the network active. The network attacks resumed only when the operational hub returned to cooperate with one of the subgroups.
Structure and Subgroups
Structural characteristics of the network are associated with group outcomes, internal social processes, and members' behavior, such as members' decision of whether or not to participate in the group's activities, as illustrated recently by Siegel (Reference Siegel2009). The SNA literature presents a variety of tools that enable us to uncover these characteristics. First are measurements of the level of cohesion and degree hierarchy within the network, such as degree of centrality, density, and inclusiveness (see, e.g, Koschade Reference Koschade2006 for an analysis of the Bali 2004 network). These measures are important for understanding the internal dynamic within the network, patterns of socialization, and the nature of the decision-making process within the group. They are also important for testing some basic hypotheses about the characteristics of clandestine networks. Most networks need to strike a balance between efficiency (low number of redundant links, high level of group centrality) and robustness or survivability (high density, high number of redundant ties). Somewhat paradoxically, however, in the case of clandestine networks, high numbers of redundant ties seem to lessen the chances of network survival. Clandestine networks are interested in secrecy, and hence, high levels of density and group centrality increase the chances of group exposure. Then again, high density facilitates effective indoctrination, a crucial element in the radicalization process of the network. As Krebs (Reference Krebs2002) shows, one way to bypass the contradicting needs of the clandestine network is to deactivate strong ties when the network operates in hostile environments. Thus, the density of the network is reduced when the network becomes active and prepares to act.
Another important set of tools addresses the uncovering of internal, cohesive subgroups within the whole network. The theoretical importance of such subgroups stems from the causal relationship found between cohesiveness and the tendency for group uniformity, intensive socialization, and radicalization (Klandermans and Oegema Reference Klandermans and Oegema1987; McAdam and Paoulsen Reference McAdam and Paoulsen1993). The concept most often used in this context is the clique, or when each actor in the subgroup is tied to all other actors. However, less cohesive subgroups sometimes exist in the network that do not fit the definition of a clique. As a result, to identify these other types of subgroups, other methodological concepts are often used such as n-cliques (n geodesic distance between members) or k-core groups (every actor has ties to at least k number of actors within the subgroup). Identifying the subgroups allows us to detect different roles within the network (founders, collaborators, passersby), network recruitment paths, operational characteristics (in some networks, a clear distinction is drawn between a suicide bomber subgroup and other subgroups, while in others, the suicide bombers are isolated actors in the periphery of the network), and patterns of flow of information. Moreover, by looking at the attributes of the subgroups, we can evaluate ideological homogeneity and level of solidarity within the network and how these factors influence the activities and development of the terrorist network. In the case of the Jewish Underground, the combination of limited connections between the subgroups and strong cohesiveness within them helps to explain the rapid disintegration of the network after its first attack, why so many actors could easily detach from the whole group while continuing to play small roles in its activities as passersby, and how it was able to avoid exposure for so long (over five years).
We would like to end this section by pointing out the potential contribution of other theoretical concepts common in SNA to the investigation of violent political groups. Because we can assume that actors' behaviors are a product of their structural opportunities and constraints, we can expect actors with similar location characteristics to react in similar ways (i.e., structural equivalence). To illustrate, studies on Muslim and Palestinian networks found that peripheral actors in most cases are those who have the highest chances of becoming perpetrators of attacks (Pedahzur and Perliger Reference Pedahzur and Perliger2006; Rodriquez Reference Rodriquez2005). In essence, the network limits the tie opportunities of these members, since they have the highest chances of being caught. Moreover, in cases in which the perpetrators were not peripheral, we can see a very limited and primitive group organizational structure; in other words, these groups only anticipate executing a single attack. Basically, the structure and location of the perpetrators can be associated with both the overall operational capabilities of the group and its life span. Last, SNA can be of high value in understanding the relationships between different terrorist groups worldwide, especially through the use of theories of structural balance (which assume that actors will forge ties in cases of shared interests such as positive ties to another third actor), homophily, and hetrophily.
CONCLUDING REMARKS
This article illustrates the potential of SNA in the study of political violence and terrorism. This method could be highly useful for further developing and testing current prominent theoretical frameworks in the field. Theories focusing on the profiling of terrorists—the identification of individuals who are more likely to be engaged in terrorist activities—could use SNA to add a social dimension to the basic sociodemographic profile by looking at patterns of social interactions, as well as distinguishing between different roles within the network that, in turn, are reflected in different profiles. Psychological theories focusing on the social-learning process could also benefit from the use of SNA that focuses on ego-networks. Naturally, the growing literature emphasizing the role of social processes within small informal groups and involvement in violent activities is well suited to the use of SNA. Finally, counterterrorism theories based on SNA could determine which group structures are more vulnerable, how these networks could be destabilized, which actors are crucial for the continuing existence of the groups, how the radicalization process could be thwarted, and how the security forces could locate recruitment paths.
