Overview
Beginnings
When the project started (when Davenport was at the University of Maryland before moving to the Kroc Institute at the University of Notre Dame and later still the University of Michigan), it was in need of some assistance with wading through a thousand or so documents provided by the organization Relatives for Justice. These documents provided in amazing detail specific activities undertaken by the British army and Royal Ulster Constabulary (RUC) against individuals within the catholic community in Belfast as well as other parts of Northern Ireland.
Four courageous and energetic youth (affiliated with two different institutions: the University of Florida and Dartmouth College) jumped at the chance to assist - Amanda Stone (Florida State), Tami Martin (Dartmouth), John Tepperman (Dartmouth) and Chelsea Moore (Florida State). This group put the initial group of material in chronological order, working frequently 8-12 hour days. Although they were compensated for the trip, this was clearly above the call of duty and they achieved this with a degree of seriousness and concentration that was inspiring.
Four courageous and energetic youth (affiliated with two different institutions: the University of Florida and Dartmouth College) jumped at the chance to assist - Amanda Stone (Florida State), Tami Martin (Dartmouth), John Tepperman (Dartmouth) and Chelsea Moore (Florida State). This group put the initial group of material in chronological order, working frequently 8-12 hour days. Although they were compensated for the trip, this was clearly above the call of duty and they achieved this with a degree of seriousness and concentration that was inspiring.
Following the archives
While searching for information on the Troubles/Conflict that has not yet been brought to light, we have been coming across disparate collections of documents throughout Northern Ireland.
In January 2009, NIRI sent a team of 4 researchers to Armagh, Northern Ireland under the supervision of Cyanne Loyle in order to index and archive the writings of Father Raymond Murray, Father Denis Faul and Sister Sarah Clark. These three collections are housed at the Tomas O’Fiaich Library at Saint Patrick’s Cathedral, Armagh.
Ft. Murray’s archive details his work as the prison chaplain for Armagh Prison in the 1970s (when it served as a women’s prison) and following this when he remained posted there through Internment (as it became an overflow facility for male and female prisoners from Belfast and surrounding areas were transferred to Armagh). Fr. Murray formally documented complaints, abuse and torture throughout the Troubles. In addition to his personal notes on various abuses, including prison torture and the degrading sexual abuse of women prisoners, Fr. Murray collected many of the personal statements taken by the Associate for Legal Justice. This work was indexed and digitized as part of our ALJ statement collection.
Fr. Murray also worked with a fellow priest, Fr. Denis Faul to document patterns of abuse throughout Northern Ireland. Fr. Murray and Fr. Faul published a series of pamphlets and reports on systematic abuses during the Troubles. This research included victim and witness statements as well as historic information collected by the priests across time. Digital copies of these pamphlets were collected for NIRI.
In addition to the joint work of Fr. Murray and Fr. Faul, pieces of Fr. Faul’s personal writings were also indexed and digitized. The majority of Fr. Faul’s writing are at the National Library in Dublin, Ireland, however, much of his personal collection that was still in his possession at the time of his death is being housed in Armagh. Of this collection, works pertaining to the Troubles, including arrest lists and abuse statements were indexed and digitized by NIRI.
Finally, the works of Sister Sarah Clark were also included in this data collection effort. Unlike Frs. Murray and Faul, Sr. Clark worked primarily with Republican prisoners in Britain. Her personal collection mostly contains the personal correspondence between her and the families of prisoners. Where this information was relevant, it was digitized and included with the other NIRI records.
This was only the beginning.
In January 2009, NIRI sent a team of 4 researchers to Armagh, Northern Ireland under the supervision of Cyanne Loyle in order to index and archive the writings of Father Raymond Murray, Father Denis Faul and Sister Sarah Clark. These three collections are housed at the Tomas O’Fiaich Library at Saint Patrick’s Cathedral, Armagh.
Ft. Murray’s archive details his work as the prison chaplain for Armagh Prison in the 1970s (when it served as a women’s prison) and following this when he remained posted there through Internment (as it became an overflow facility for male and female prisoners from Belfast and surrounding areas were transferred to Armagh). Fr. Murray formally documented complaints, abuse and torture throughout the Troubles. In addition to his personal notes on various abuses, including prison torture and the degrading sexual abuse of women prisoners, Fr. Murray collected many of the personal statements taken by the Associate for Legal Justice. This work was indexed and digitized as part of our ALJ statement collection.
Fr. Murray also worked with a fellow priest, Fr. Denis Faul to document patterns of abuse throughout Northern Ireland. Fr. Murray and Fr. Faul published a series of pamphlets and reports on systematic abuses during the Troubles. This research included victim and witness statements as well as historic information collected by the priests across time. Digital copies of these pamphlets were collected for NIRI.
In addition to the joint work of Fr. Murray and Fr. Faul, pieces of Fr. Faul’s personal writings were also indexed and digitized. The majority of Fr. Faul’s writing are at the National Library in Dublin, Ireland, however, much of his personal collection that was still in his possession at the time of his death is being housed in Armagh. Of this collection, works pertaining to the Troubles, including arrest lists and abuse statements were indexed and digitized by NIRI.
Finally, the works of Sister Sarah Clark were also included in this data collection effort. Unlike Frs. Murray and Faul, Sr. Clark worked primarily with Republican prisoners in Britain. Her personal collection mostly contains the personal correspondence between her and the families of prisoners. Where this information was relevant, it was digitized and included with the other NIRI records.
This was only the beginning.
The National Science Foundation Proposal & An Organizational approach to state repression
1. Research Questions and Project Objectives
For nearly 50 years, research in political science and sociology has systematically attempted to understand why governments repress their citizens. To date the bulk of attention has been given to exploring the influence of diverse political, economic and demographic characteristics of nation-states as these are believed to influence the decision-making processes employed by those choosing to use repression as well as the specific activities involved by those who implement it. Employing diverse approaches, datasets and methods, this work has revealed a significant number of insights. For example, we know that (on average) repressive behavior escalates in response to behavioral threats from challengers and diminishes in the presence of democracy, economic development and, to some extent, human rights treaties (e.g., Hibbs 1973; Poe & Tate 1994; Davenport 1995; 1999; 2007a;b; Keith et al. 2009; Carey 2010; Cingranelli & Richards 1999; Moore 2000; Franklin 1997; Wood 2008; Krain 1997; Hafner-Burton 2005; Hill 2010; Conrad 2011; Lupu 2013). More recently, viewing two consistently influential variables interactively, we have come to know that even democracies repress when confronted with behavioral threats (e.g., Davenport 2007a; Conrad & Moore 2010).
Comparatively, we know much less about other factors potentially related to repressive behavior. For example, one lucrative area concerns the organizational characteristics of repressive institutions – particularly, the relationship between “principals” (those who order repressive acts) and “agents” (the individuals physically committing such activities) and ultimately how these characteristics influence how each behave. When considering this relationship, the literature currently offers three very different approaches: 1) the principal-agent (PA) argument where private/opportunistic motives of repressive agents drive repressive action (e.g., Mitchell 2004; Butler et al. 2007; Demeritt 2009); 2) the organizational pathology argument where the objectives of principals drive repression (e.g., Cunningham 2003; 2004) and, 3) the “Blue” approach where the perceptions of behavioral threats by repressive agents drives state repressive action (e.g., Earl & Soule 2006; Earl et al. 2003). Thus far, these arguments have been viewed separately and each has received support.
While an important direction, we maintain that the different approaches identified above each capture only part of the relevant process and that the best way to examine as well as understand repressive behavior is to bring the diverse approaches together. Specifically, we argue that the nature and scale of the challenger’s threat influences the motives/objectives of the principal and the agent, but to different ends. Eventreporting is an essential component of understanding how principals receive information from agents and how that information is converted into threats. And finally, how the repressive apparatus is organized will impact how orders from the principal are given and ultimately received by various agents, including mechanisms for oversight and sanction. To empirically evaluate our argument, we seek support from the National Science Foundation to identify the repressive activity of the United Kingdom (UK) government that took place during the “Troubles”/'Conflict" in Northern Ireland from 1968-1998 as well as assess the role of principals and agents in this conflict across time (by day), space (by neighborhood), and actor (by battalion). Using diverse data sources (i.e., government reports, NGO statements, and newspapers) as well as an extensive set of interviews, the project will catalogue the activities of the UK government (both principals and agents) as they interacted with Republican/Catholic as well as Loyalist/Protestant challenging organizations and behaviors. Our focus in this proposal is on the UK government and on the British Army because they had direct oversight on security matters throughout the Troubles (Dewar 1985). For example, Northern Ireland’s police force, the RUC, was instructed not to directly engage with challengers unless they were accompanying an Army patrol. Subsequent research will incorporate data (previously collected from the Public Records Office of Northern Ireland [Loyle et al. 2014]) on the behavior and organization of the police. Given past work on the topic (e.g., Ellison & Smyth 2000), we also acknowledge the potential collusion between Loyalist violence and the security forces. Within this proposal we focus specifically on security force violence, while in subsequent research we aim to empirically examine arguments related to collusion between these two forces.
2. Three Arguments Concerning Repression and Organizations
Acknowledging that prior research was exploring variables that maintained very little connection with the coercive behavior being used by political authorities, in the mid-1990s a new approach emerged focused on institutions. This literature has put forward three competing approaches to explain the occurrence of repressive action: the Principal-Agent (P-A) model, the “organizational pathology” approach, and the “blue” framework. The differences in approach are important to identify for they lead us to focus on distinct causal factors as well as trace evidence. Each of these arguments as well as their limitations are described in detail below.
In the first approach – the Principal-Agent model – advocated by individuals such as Mitchell (2004), Brehm & Gates (1993), Butler et al. (2007) and Demeritt (2009), it is presumed that while principals set forth their objectives as well as the accepted repertoire of tactics to be used, agents enact those objectives as they see fit. In this context, agents have their own (hidden) objectives (e.g., wealth, survival and sexual predation) and they employ repressive behavior in accordance to what they see as necessary given these private motives. Within this approach, agents maintain informational advantages because they operate in the field (closer to the behavior of interest that they are delegated to enact) whereas principals do not. This results in agents holding back certain information that they would prefer the principals not to know while reporting information that they are comfortable passing on as it most likely fits with the principal’s objectives as well as the accepted tactical repertoire of engagement. Accordingly, repressive behavior is expected when circumstances arise where agents believe it to be lucrative such as when there is opportunity for physical humiliation of a challenger, loot is readily available or women are left alone. The more isolated individual agents are from the rest of their organization/unit, the greater the likelihood that relevant behavior would be enacted. In other words, repressive activity is the product of individual actors’ motivation and opportunity. It is further facilitated as well as covered by biased reporting from agents back to principals.
In the second approach – the “Organizational Pathology” model – advocated by individuals such as Gurr (1986) and Cunningham (2003; 2004), principals set forth their objectives as well as the accepted repertoire of tactics and agents enact them dutifully, to the letter. In fact, within this approach, agent responsiveness to principal directives is so complete that the former tend to neglect all other factors such as local context; not completely, but significantly enough that it is clear where their primary focus lay. While there is no clear argument about what objectives ultimately guide principals, researchers have maintained that initial challenger threat as perceived by the principal is key. Differentiated from the P-A argument, the pathology model maintains that information relevant to the behavior of interest moves one-way, downward and that any information moving upward would 1) exclusively consider/respond to information sent down and 2) generally be complete (i.e., nothing would be left out).
The third and final approach is the "Blue" framework advocated by Earl & Soule (2006). In this work, principals set forth their objectives as well as acceptable tactical repertoires and agents enact them as they see fit. Here, different from the Pathology model, agents are not following the directives given to them without modification and different from the P-A approach, private motives are not what drive repression. Rather, in this approach, agents respond to the behavioral threats they encounter in the field of competition/battle (e.g., being surrounded, engaging in shootouts or losing a comrade). These are very different from the more abstract threats that principals concern themselves with such as declarations of a political challenge or the release of a manifesto. Here, repressive behavior is expected when circumstances arise and agents believe certain behaviors to be necessary.
For nearly 50 years, research in political science and sociology has systematically attempted to understand why governments repress their citizens. To date the bulk of attention has been given to exploring the influence of diverse political, economic and demographic characteristics of nation-states as these are believed to influence the decision-making processes employed by those choosing to use repression as well as the specific activities involved by those who implement it. Employing diverse approaches, datasets and methods, this work has revealed a significant number of insights. For example, we know that (on average) repressive behavior escalates in response to behavioral threats from challengers and diminishes in the presence of democracy, economic development and, to some extent, human rights treaties (e.g., Hibbs 1973; Poe & Tate 1994; Davenport 1995; 1999; 2007a;b; Keith et al. 2009; Carey 2010; Cingranelli & Richards 1999; Moore 2000; Franklin 1997; Wood 2008; Krain 1997; Hafner-Burton 2005; Hill 2010; Conrad 2011; Lupu 2013). More recently, viewing two consistently influential variables interactively, we have come to know that even democracies repress when confronted with behavioral threats (e.g., Davenport 2007a; Conrad & Moore 2010).
Comparatively, we know much less about other factors potentially related to repressive behavior. For example, one lucrative area concerns the organizational characteristics of repressive institutions – particularly, the relationship between “principals” (those who order repressive acts) and “agents” (the individuals physically committing such activities) and ultimately how these characteristics influence how each behave. When considering this relationship, the literature currently offers three very different approaches: 1) the principal-agent (PA) argument where private/opportunistic motives of repressive agents drive repressive action (e.g., Mitchell 2004; Butler et al. 2007; Demeritt 2009); 2) the organizational pathology argument where the objectives of principals drive repression (e.g., Cunningham 2003; 2004) and, 3) the “Blue” approach where the perceptions of behavioral threats by repressive agents drives state repressive action (e.g., Earl & Soule 2006; Earl et al. 2003). Thus far, these arguments have been viewed separately and each has received support.
While an important direction, we maintain that the different approaches identified above each capture only part of the relevant process and that the best way to examine as well as understand repressive behavior is to bring the diverse approaches together. Specifically, we argue that the nature and scale of the challenger’s threat influences the motives/objectives of the principal and the agent, but to different ends. Eventreporting is an essential component of understanding how principals receive information from agents and how that information is converted into threats. And finally, how the repressive apparatus is organized will impact how orders from the principal are given and ultimately received by various agents, including mechanisms for oversight and sanction. To empirically evaluate our argument, we seek support from the National Science Foundation to identify the repressive activity of the United Kingdom (UK) government that took place during the “Troubles”/'Conflict" in Northern Ireland from 1968-1998 as well as assess the role of principals and agents in this conflict across time (by day), space (by neighborhood), and actor (by battalion). Using diverse data sources (i.e., government reports, NGO statements, and newspapers) as well as an extensive set of interviews, the project will catalogue the activities of the UK government (both principals and agents) as they interacted with Republican/Catholic as well as Loyalist/Protestant challenging organizations and behaviors. Our focus in this proposal is on the UK government and on the British Army because they had direct oversight on security matters throughout the Troubles (Dewar 1985). For example, Northern Ireland’s police force, the RUC, was instructed not to directly engage with challengers unless they were accompanying an Army patrol. Subsequent research will incorporate data (previously collected from the Public Records Office of Northern Ireland [Loyle et al. 2014]) on the behavior and organization of the police. Given past work on the topic (e.g., Ellison & Smyth 2000), we also acknowledge the potential collusion between Loyalist violence and the security forces. Within this proposal we focus specifically on security force violence, while in subsequent research we aim to empirically examine arguments related to collusion between these two forces.
2. Three Arguments Concerning Repression and Organizations
Acknowledging that prior research was exploring variables that maintained very little connection with the coercive behavior being used by political authorities, in the mid-1990s a new approach emerged focused on institutions. This literature has put forward three competing approaches to explain the occurrence of repressive action: the Principal-Agent (P-A) model, the “organizational pathology” approach, and the “blue” framework. The differences in approach are important to identify for they lead us to focus on distinct causal factors as well as trace evidence. Each of these arguments as well as their limitations are described in detail below.
In the first approach – the Principal-Agent model – advocated by individuals such as Mitchell (2004), Brehm & Gates (1993), Butler et al. (2007) and Demeritt (2009), it is presumed that while principals set forth their objectives as well as the accepted repertoire of tactics to be used, agents enact those objectives as they see fit. In this context, agents have their own (hidden) objectives (e.g., wealth, survival and sexual predation) and they employ repressive behavior in accordance to what they see as necessary given these private motives. Within this approach, agents maintain informational advantages because they operate in the field (closer to the behavior of interest that they are delegated to enact) whereas principals do not. This results in agents holding back certain information that they would prefer the principals not to know while reporting information that they are comfortable passing on as it most likely fits with the principal’s objectives as well as the accepted tactical repertoire of engagement. Accordingly, repressive behavior is expected when circumstances arise where agents believe it to be lucrative such as when there is opportunity for physical humiliation of a challenger, loot is readily available or women are left alone. The more isolated individual agents are from the rest of their organization/unit, the greater the likelihood that relevant behavior would be enacted. In other words, repressive activity is the product of individual actors’ motivation and opportunity. It is further facilitated as well as covered by biased reporting from agents back to principals.
In the second approach – the “Organizational Pathology” model – advocated by individuals such as Gurr (1986) and Cunningham (2003; 2004), principals set forth their objectives as well as the accepted repertoire of tactics and agents enact them dutifully, to the letter. In fact, within this approach, agent responsiveness to principal directives is so complete that the former tend to neglect all other factors such as local context; not completely, but significantly enough that it is clear where their primary focus lay. While there is no clear argument about what objectives ultimately guide principals, researchers have maintained that initial challenger threat as perceived by the principal is key. Differentiated from the P-A argument, the pathology model maintains that information relevant to the behavior of interest moves one-way, downward and that any information moving upward would 1) exclusively consider/respond to information sent down and 2) generally be complete (i.e., nothing would be left out).
The third and final approach is the "Blue" framework advocated by Earl & Soule (2006). In this work, principals set forth their objectives as well as acceptable tactical repertoires and agents enact them as they see fit. Here, different from the Pathology model, agents are not following the directives given to them without modification and different from the P-A approach, private motives are not what drive repression. Rather, in this approach, agents respond to the behavioral threats they encounter in the field of competition/battle (e.g., being surrounded, engaging in shootouts or losing a comrade). These are very different from the more abstract threats that principals concern themselves with such as declarations of a political challenge or the release of a manifesto. Here, repressive behavior is expected when circumstances arise and agents believe certain behaviors to be necessary.
3. A Theory of Security Force Organization and Repression in Democracies
Guided by previous work, our organizational approach to understanding political repression represents a theory that is at once a synthesis of earlier research, but also a significant extension. In line with the extant literature, we believe that it is essential to consider a principal as well as their agents in understanding repressive action (e.g., Brehm & Gates 1993; Mitchell 2004; Butler et al. 2007). Differing from this work, however, we believe it is vital to understand the nature and the scale of the challenger threat as perceived by both principals and agents, the relationship between ‘actual’ compared with reportedevents, and the organization of the repressive apparatus. Below we discuss the role of each of these factors and how they interact in detail. In brief, however, behavioral threats determine both the principal’s allocation of repressive agents and the responses of agents on the ground. Reporting and organization matter because these are the tactics that principals and agents employ against one another as they attempt to determine the apportionment of repressive behavior. Examining the interaction of these factors allows us to better examine as well as differentiate between the diverse explanations put forward on this topic and facilitates the examination of important intersections across existing approaches that have not been previously addressed.
Challenger Threat and Repression
For government principals, challenger threats form the primary motivation underlying the deployment of repressive agents and their assignments to commit coercion. As argued, the objective of a principal is to pacify a perceived challenge and remain in office. Within existing work, the “Organizational Pathology” approach argues that given a perceived/potential threat, the government selects the appropriate security force organization to respond and an allocation of repressive behavior that it believes will generate the “correct level” of repression. By the latter, we mean that repression A) reduces public disturbances and B) does not generate “excessive” violations of human rights, where civilian bystanders become aggrieved with the government because of what is taking place. Here, focus is on the beliefs and objectives of the principals who conceive of the relevant coercive policies and there is little to no concern with the agents who enact repressive behavior.
At the same time, threats determine the underlying motivations of repressive agents, though often to different ends. As an example, the “Seeing Blue” framework argues that once repressive agents engage in relevant action, their efforts are influenced by threat dynamics in the localenvironment (i.e., on the street or in the mountains where people are questioned, beaten, arrested, shot and killed). At this level, the situation is reversed and the lives and welfare of agents become the focus of attention as agents apply repression at levels that maximize their own individual security.
As a result of this, the principal is uncertain for any given situation as to whether the agents will apply the amount of repression dictated in their commands or “oversupply” repression potentially resulting in criticism as well as their potential removal from office. Consistent with “Principal Agent” arguments, the principal weighs the costs and benefits that may occur if repression were oversupplied against the risk of allowing challengers to proceed without repression, and assigns a security force deployment. We contend that the level of perceived behavioral threats operating during the initial troop deployment sets these objectives in motion. Specifically, at relatively low levels of perceived threats, principals would prefer the risks associated with more limited repression, rather than oversupply repression, and as a result the allocation of repression becomes more targeted. Specifically,
On the other hand, when behavioral threats from challengers are believed to operate on a larger scale (across similar dimensions), we expect that this sets in motion different deployment objectives for the principal, which results in the formulation of new organizational dynamics allowing agents greater leeway to engage in repressive activity in surrounding areas.[2]Here, the principal is seeking to target a perceived challenge, but is willing to risk the possibility that agents will oversupply repression. At higher threat levels, principals are more likely influenced by what transpires and thus there is more interest in creating opportunities for agents to crack down on challengers. Consequently,
Reporting and Repression
When considering the importance of threat dynamics following the initial deployment of security agents, the reporting of conflict behaviors between repressive organizations and political authorities becomes critically important. With troops on the ground, the principal now has access to new sources of localized information on challenger behavior. Consequently, how the principal receives, interprets, and translates a challenger’s threat into agent orders (e.g., commands dictating the assignment of troops and the tactics to be employed) becomes a product of the reporting practices of its agents. The same can be said about the awareness principals have regarding agent deviations from principal directives. The process is complex. As conceived, principals receive information from agents on behavioral threats and are able to use that information to determine appropriate repressive responses (e.g., making adjustments to expectations, perceptions and directives). However, there may be variation in which events and challenges agents choose to report to principals. In line with this, we believe that agents are strategic in their reporting. As agents respond to local threats with repression, we anticipate that the most coercive agents will be more inclined to pass information about challenger activities up the chain of command in order to validate their behaviors. Consequently,
This process produces variation in the awareness of principals with regards to threat levels on the ground, as agents strategically manipulate the flow of information upwards in order to justify their repressive practices. Areas where agents apply heightened levels of repression are precisely the same locales where principals are most likely to become aware of potential threats. As this occurs, principals can be expected to deploy greater levels of force, leading to an escalation of coercion and conflict.
The correspondence between reporting practices and the application of repression limits the capacity of principals to see through the fog of war while simultaneously enabling agents to commit opportunistic violence. As a consequence, we should expect rapid escalations in repressive behavior occurring in response to small fluctuations in local level conditions. Security forces engage in repression, which facilitates reporting that promotes further repression. The effect complicates the threat arguments specified above. Specifically,
Institutional Organization and Repression
In addition to the expected effect of threat levels and reporting on the use of repression, our argument posits that the application of repressive behavior is directly related to the structure of the organization involved and the location of individual agents distributed across space and time. Indeed, if reporting is the manner in which agents attempt to subvert or alter the principal’s directives, organizational structure is one crucial tool principals possess to retain control over their agents. In this manner, principals attempt to devise organizational structures to reduce the freedom agents possess to deviate from the level of repression preferred by the principal. For example, when we consider the principal-agent argument, chains of communication and command appear crucial for understanding when agents are likely to engage in the over-supply or under-supply of repression. Thus, we draw the proposition that
We further posit that organizational cohesion breaks down when the relevant institution is taxed in some manner, such as when it is under placed in remote locales or when it is placed under extreme duress.
Finally, it is important to consider the manner in which threat, reporting, and organization interact. Related to these dynamics we also maintain that
The correspondence between directives and behaviors largely reflects the relevant agents perception that they are under greater scrutiny as well as the fact that additional repression would potentially met with a negative response from principals. While the area around the discovered repressive action is under higher scrutiny and thus compliance with principal directives, drawing on P-A dynamics, we further anticipate
Guided by previous work, our organizational approach to understanding political repression represents a theory that is at once a synthesis of earlier research, but also a significant extension. In line with the extant literature, we believe that it is essential to consider a principal as well as their agents in understanding repressive action (e.g., Brehm & Gates 1993; Mitchell 2004; Butler et al. 2007). Differing from this work, however, we believe it is vital to understand the nature and the scale of the challenger threat as perceived by both principals and agents, the relationship between ‘actual’ compared with reportedevents, and the organization of the repressive apparatus. Below we discuss the role of each of these factors and how they interact in detail. In brief, however, behavioral threats determine both the principal’s allocation of repressive agents and the responses of agents on the ground. Reporting and organization matter because these are the tactics that principals and agents employ against one another as they attempt to determine the apportionment of repressive behavior. Examining the interaction of these factors allows us to better examine as well as differentiate between the diverse explanations put forward on this topic and facilitates the examination of important intersections across existing approaches that have not been previously addressed.
Challenger Threat and Repression
For government principals, challenger threats form the primary motivation underlying the deployment of repressive agents and their assignments to commit coercion. As argued, the objective of a principal is to pacify a perceived challenge and remain in office. Within existing work, the “Organizational Pathology” approach argues that given a perceived/potential threat, the government selects the appropriate security force organization to respond and an allocation of repressive behavior that it believes will generate the “correct level” of repression. By the latter, we mean that repression A) reduces public disturbances and B) does not generate “excessive” violations of human rights, where civilian bystanders become aggrieved with the government because of what is taking place. Here, focus is on the beliefs and objectives of the principals who conceive of the relevant coercive policies and there is little to no concern with the agents who enact repressive behavior.
At the same time, threats determine the underlying motivations of repressive agents, though often to different ends. As an example, the “Seeing Blue” framework argues that once repressive agents engage in relevant action, their efforts are influenced by threat dynamics in the localenvironment (i.e., on the street or in the mountains where people are questioned, beaten, arrested, shot and killed). At this level, the situation is reversed and the lives and welfare of agents become the focus of attention as agents apply repression at levels that maximize their own individual security.
As a result of this, the principal is uncertain for any given situation as to whether the agents will apply the amount of repression dictated in their commands or “oversupply” repression potentially resulting in criticism as well as their potential removal from office. Consistent with “Principal Agent” arguments, the principal weighs the costs and benefits that may occur if repression were oversupplied against the risk of allowing challengers to proceed without repression, and assigns a security force deployment. We contend that the level of perceived behavioral threats operating during the initial troop deployment sets these objectives in motion. Specifically, at relatively low levels of perceived threats, principals would prefer the risks associated with more limited repression, rather than oversupply repression, and as a result the allocation of repression becomes more targeted. Specifically,
- H1a: Relatively low levels of threats (i.e., limited casualty counts, frequency/intensity, localized violence) result in troop deployments and repressive activity in line with the threat presented at the specific locale where the behavioral challenge manifests.
On the other hand, when behavioral threats from challengers are believed to operate on a larger scale (across similar dimensions), we expect that this sets in motion different deployment objectives for the principal, which results in the formulation of new organizational dynamics allowing agents greater leeway to engage in repressive activity in surrounding areas.[2]Here, the principal is seeking to target a perceived challenge, but is willing to risk the possibility that agents will oversupply repression. At higher threat levels, principals are more likely influenced by what transpires and thus there is more interest in creating opportunities for agents to crack down on challengers. Consequently,
- H1b: Relatively larger threats (i.e., deaths and extensive property damage) result in troop deployments and repressive behavior within locales including as well as surrounding the one where the threat manifests.
Reporting and Repression
When considering the importance of threat dynamics following the initial deployment of security agents, the reporting of conflict behaviors between repressive organizations and political authorities becomes critically important. With troops on the ground, the principal now has access to new sources of localized information on challenger behavior. Consequently, how the principal receives, interprets, and translates a challenger’s threat into agent orders (e.g., commands dictating the assignment of troops and the tactics to be employed) becomes a product of the reporting practices of its agents. The same can be said about the awareness principals have regarding agent deviations from principal directives. The process is complex. As conceived, principals receive information from agents on behavioral threats and are able to use that information to determine appropriate repressive responses (e.g., making adjustments to expectations, perceptions and directives). However, there may be variation in which events and challenges agents choose to report to principals. In line with this, we believe that agents are strategic in their reporting. As agents respond to local threats with repression, we anticipate that the most coercive agents will be more inclined to pass information about challenger activities up the chain of command in order to validate their behaviors. Consequently,
- H2a: At heightened levels of repression, information on challenger behavior is more likely to be passed up through official channels.
This process produces variation in the awareness of principals with regards to threat levels on the ground, as agents strategically manipulate the flow of information upwards in order to justify their repressive practices. Areas where agents apply heightened levels of repression are precisely the same locales where principals are most likely to become aware of potential threats. As this occurs, principals can be expected to deploy greater levels of force, leading to an escalation of coercion and conflict.
The correspondence between reporting practices and the application of repression limits the capacity of principals to see through the fog of war while simultaneously enabling agents to commit opportunistic violence. As a consequence, we should expect rapid escalations in repressive behavior occurring in response to small fluctuations in local level conditions. Security forces engage in repression, which facilitates reporting that promotes further repression. The effect complicates the threat arguments specified above. Specifically,
- H2b: Following the reporting of challenger threats, rapid increases in repressive behavior (in frequency and lethality) are likely within locales including as well as surrounding the one where the threat manifests.
Institutional Organization and Repression
In addition to the expected effect of threat levels and reporting on the use of repression, our argument posits that the application of repressive behavior is directly related to the structure of the organization involved and the location of individual agents distributed across space and time. Indeed, if reporting is the manner in which agents attempt to subvert or alter the principal’s directives, organizational structure is one crucial tool principals possess to retain control over their agents. In this manner, principals attempt to devise organizational structures to reduce the freedom agents possess to deviate from the level of repression preferred by the principal. For example, when we consider the principal-agent argument, chains of communication and command appear crucial for understanding when agents are likely to engage in the over-supply or under-supply of repression. Thus, we draw the proposition that
- H3a: Organizational cohesion (i.e., the degree to which security force institutions are made coherent in terms of chains of command) can lead to repressive policies that are less divergent from the directives of the principal.
We further posit that organizational cohesion breaks down when the relevant institution is taxed in some manner, such as when it is under placed in remote locales or when it is placed under extreme duress.
- H3b: Agents who are more physically proximate to command headquarters will be more likely to deliver repression at levels mandated by their principals than those who are more isolated. In contrast, locations that are geographically remote from command headquarters should be more likely to experience repressive activity that is not in line with levels mandated by principal directives.
- H3c: At low levels of threat where scrutiny and oversight are believed to matter more for principals than the production of repression, repressive behavior will be more closely tied to levels mandated by principals. Conversely, as perceived threat levels increase, we expect to see principals more willing to allow their agents discretion to deal with perceived challenges. Consequently, there will be greater variation in individual agent behavior away from directives from principals.
Finally, it is important to consider the manner in which threat, reporting, and organization interact. Related to these dynamics we also maintain that
- H3d: Where information about previously undisclosed repressive behavior gets discovered, there is short-term behavioral similarity between principal directives and agent activity surrounding the area where the relevant events took place.
The correspondence between directives and behaviors largely reflects the relevant agents perception that they are under greater scrutiny as well as the fact that additional repression would potentially met with a negative response from principals. While the area around the discovered repressive action is under higher scrutiny and thus compliance with principal directives, drawing on P-A dynamics, we further anticipate
- H3e: Where information about previously undisclosed repressive behavior gets discovered, there is short-term behavioral dissimilarity between principal directives and agent activity as well as egregious activity in areas away from the one(s) where the relevant events took place.
Guided by previous work, our organizational approach to state repression represents something that is at once a synthesis of earlier research but also a significant extension. In line with extant literature, we believe that it is essential to consider a principal as well as their agents in understanding repressive action. Differing from this work, however, we believe that it is vital to understand the nature and the scale of the challenger threat as perceived by both principals and agents, the organization of the repressive apparatus, and the relationship between ‘actual’ compared with reported events. Such an approach not only allows us to better examine as well as differentiate between the diverse explanations put forward above concerning how organizations influence repressive behavior but it also facilitates the examination of important intersections across existing approaches that have not been previously addressed. This is diagrammed below.
4. Research Design and Data Compilation
To empirically access our theoretical claims, we propose to analyze data on the Troubles/Conflict in Northern Ireland between 1968 and 1998. There has been a great deal written about this period and a large number of insights have been provided by this work; indeed, there are very few conflagrations in world history that have been studied as much (e.g., see the Conflict Archive on the Internet [CAIN] reading list). For several reasons, Northern Ireland is a particularly fruitful area to situate our analysis.
Existing data on Northern Ireland
Since the beginning of the conflagration commonly known as the Troubles/Conflict, there has been a great deal written about who has done what to whom. As a result, there have been a series of studies designed to explicitly catalog the behavior that took place. This includes compendiums detailing all deaths that occurred throughout the conflict (e.g., Sutton 2004) as well as efforts to catalog the experiences of particular communities (e.g., Ardoyne Commemoration Project 2002). These data have proven useful in generating insights on a wide variety of topics, ranging from a study of the causes of ethno-nationalist violence (e.g., Maney 2005) to a study estimating the cost of terrorism on stock prices (e.g., Frey et al. 2007). A limited number of scholars have begun to examine other forms of conflict/violence during the Troubles, such as punishment or sectarian attacks on symbolic targets (e.g., Jarman 2004; LaFree et al. 2009), but this effort has been limited to the analysis of annual level changes across the entire country allowing only macro-level independent variables and explanations to be considered. Further, such projects have been limited to the analysis of a single form of conflict/violence such as terrorism.
Efforts have been made to specifically understand repression in the Troubles (e.g., White & White 1995; Francisco 1996; White 1999). Such efforts should be credited for improving our understanding in this crucial area. At the same time, the existing work provides almost no discussion as to how decisions were made to deploy repressive violence, and who made decisions on how repressive action was deployed across space, time and context. This research relies on highly aggregated units, for example examining counts of repressive behavior across the entire province and generally relies on a single source or a limited number of related source (i.e., several different newspapers). To understand repressive organizations and their links to relevant state behavior, progress in the collection of Troubles-related data requires significantly more disaggregation across both time and space, as well as the ability to systematically gather data across battalions in the British military.
The Northern Ireland Research Initiative. In 2007, we began the Northern Ireland Research Initiative (NIRI) to collect disaggregated, multi-source information on the Troubles. This data is particularly well suited to address the hypotheses presented above and the existing limitations in our empirical understanding of agent behavior. To measure threats, reporting, organizational characteristics, and political repression, NIRI employs a broad and diverse set of high quality data sources available in the United Kingdom (including Northern Ireland).
By collecting data across the various organizational levels of the military, our project is able to document (1) what orders were being transmitted down the chain of command to direct military agents towards engaging in specific behaviors towards particular points or times, (2) what types of coercive behaviors were occurring, as well as when and where they were ongoing, and (3) which behaviors as well as which relevant contextual information was being reported upward to the principals in command. In addition to documentation originating from state security forces, NIRI has collected and coded independent data measuring the behaviors of both challengers and government forces. To gather additional data on political activity pre- and post-military deployment, the project has collected a set of events records from a comprehensive media survey cataloging all politically relevant behavior identified in the Irish and British news media beginning in 1968. The British military and media data are complemented by a set of event data drawn from victims’ statements taken by a human rights organization that was active during the relevant period. Because these statements were taken directly from the victims of repression, they are able to capture a broad set of hidden forms of activities not frequently reported in military documents or published by the media.
Empirical Analysis
As presented above, we are interested in empirically evaluating three sets of hypotheses. The first concerns the relationship between challenger threat and repressive activity, the second addresses the reporting of repressive behavior, while the third concerns the organizational characteristics and repressive behavior. As such we will examine highly disaggregated data on threat, reporting, organization, and repressive behavior across space, time, and organizational unit. In particular, the project will examine the repressive behavior of individuals within the 58 battalions deployed across Northern Ireland during the Troubles, with each battalion referenced to 1) the area it was deployed, 2) the timing of its deployment, and 3) the brigade it served under. The units of analysis will be the battalion-week, and empirical models will estimate the relationship between different organizational characteristics and repressive behavior.
We will begin by evaluating each of the behavioral hypotheses in conjunction with one another. The models will be specified as cross-sectional time-series, and employ battalion level fixed-effects. The aim is to identify whether the different behavioral, organizational, and reporting characteristics correlate with applied levels of political repression, while controlling for rival hypotheses. From there, greater attention will be paid to identifying causality in the relationship between these related characteristics and repressive behavior. Here, we will evaluate each of the competing hypotheses individually using two approaches: a matched sample difference-in-difference (DiD) approach and an instrumental variables design. The matched sample DiD models will employ genetic matching to ensure balance across a range of relevant observable confounding factors (e.g., deployment location, regional levels of development, sectarian settlement patterns, etc.), while also controlling for any remaining unobservable differences between treatment and control groups (e.g., Lyall 2009; Sullivan 2014). The instrumental variables models will exploit the as-if random assignment in battalion deployment created by the British military’s rules on the duration of tours of duty (e.g., Crost and Johnson 2010). In addition to introducing variation in deployment, rules of tour duration also affected levels of threat (as challengers reacted to new battalions), reporting (as different battalions chose different reporting patterns), and organization (as deployment shuffled the number of battalions deployed within each brigade). When combined with the results from the cross-sectional time-series analyses, the causal inference econometrics will provide more conclusive evidence regarding the validity of the causal mechanisms articulated in our theory.
Examining the Level of Challenger Threat. Addressing the first set of hypotheses requires understanding the nature and evolution of the conflict across space and time. Using the data collected from the Associate for Legal Justice and newspapers sources (as detailed above), we will be able to create an event catalog of “actual” challenger threats which details as accurately as possible the activities of the conflict across the country and across time. This catalog will be used to evaluate the challenger threat, its intensity and its distribution to allow us to test our hypothesis about threat perception and repressive behavior across principals and agents.
Examining Organizational Structure. Addressing diverse institutional characteristics during the Troubles requires a detailed understanding of both the organizational structure of the security apparatus deployed in Northern Ireland and the systems of reporting used for communication across different levels of the organization. At the cabinet level there were a number of different committees and offices that had oversight positions above the British Army in Northern Ireland, including the Joint Intelligence Committee, the Prime Minister’s office and Cabinet 9B, the cabinet committee responsible for governing Northern Ireland during the period of direct rule (1972-1998). In Northern Ireland, the Army was lead by HQNI, their command headquarters in Belfast. From there, the Army was typically divided into three brigades, which were assigned spatially to different areas Northern Ireland. Battalions were the principal units composing the brigades and rotating in and out of the province during the Troubles.
Table 1 displays the principal reporting documents produced at each of the relevant levels of the organization and the corresponding variables and hypotheses that each source will be used to test. The documents present accountability at distinct levels in that they represent reporting from one level to another. For example, log files were written by individual battalions but were ultimately delivered to brigades. Serious encounters were reported up the chain of command through the Situation Reports (at the brigade level) and Major Incident Reports (at HQNI). Organizational hypotheses will be evaluated by examining variation in the congruence between the principal directives and repressive behavior.
To empirically access our theoretical claims, we propose to analyze data on the Troubles/Conflict in Northern Ireland between 1968 and 1998. There has been a great deal written about this period and a large number of insights have been provided by this work; indeed, there are very few conflagrations in world history that have been studied as much (e.g., see the Conflict Archive on the Internet [CAIN] reading list). For several reasons, Northern Ireland is a particularly fruitful area to situate our analysis.
- First, while conflict and repression are often measured using national averages, our data identify how Northern Ireland was exposed to a range of conflict activities and repression, which had widely varying intensities across time and space. A detailed within-case investigation of repression can account for this variation, while also enabling the study to disaggregate different components of the repressive bureaucracy.
- Second, discerning the connection between repression and the specific causal mechanisms articulated in our argument as well as its principal alternatives necessitates an analysis capable of measuring each proposition’s concepts and identifying how they relate to liberalization. Whereas a broad, cross-national study could face significant difficulties identifying important theoretical concepts and ruling out observationally equivalent explanations, the availability of diverse sources of data from multiple actors across the entire span of the conflict makes the case an ideal setting to analyze hard to measure concepts such as threat, reporting, and organization.
- Finally, the particularities of the case also reveal a number of important facets relevant to the study of liberalization. Throughout the course of the conflict the UK remained a functioning democracy adhering to many of the world’s human rights treaties. And yet there remains unaccounted for variation in the use repression across different military units.
Existing data on Northern Ireland
Since the beginning of the conflagration commonly known as the Troubles/Conflict, there has been a great deal written about who has done what to whom. As a result, there have been a series of studies designed to explicitly catalog the behavior that took place. This includes compendiums detailing all deaths that occurred throughout the conflict (e.g., Sutton 2004) as well as efforts to catalog the experiences of particular communities (e.g., Ardoyne Commemoration Project 2002). These data have proven useful in generating insights on a wide variety of topics, ranging from a study of the causes of ethno-nationalist violence (e.g., Maney 2005) to a study estimating the cost of terrorism on stock prices (e.g., Frey et al. 2007). A limited number of scholars have begun to examine other forms of conflict/violence during the Troubles, such as punishment or sectarian attacks on symbolic targets (e.g., Jarman 2004; LaFree et al. 2009), but this effort has been limited to the analysis of annual level changes across the entire country allowing only macro-level independent variables and explanations to be considered. Further, such projects have been limited to the analysis of a single form of conflict/violence such as terrorism.
Efforts have been made to specifically understand repression in the Troubles (e.g., White & White 1995; Francisco 1996; White 1999). Such efforts should be credited for improving our understanding in this crucial area. At the same time, the existing work provides almost no discussion as to how decisions were made to deploy repressive violence, and who made decisions on how repressive action was deployed across space, time and context. This research relies on highly aggregated units, for example examining counts of repressive behavior across the entire province and generally relies on a single source or a limited number of related source (i.e., several different newspapers). To understand repressive organizations and their links to relevant state behavior, progress in the collection of Troubles-related data requires significantly more disaggregation across both time and space, as well as the ability to systematically gather data across battalions in the British military.
The Northern Ireland Research Initiative. In 2007, we began the Northern Ireland Research Initiative (NIRI) to collect disaggregated, multi-source information on the Troubles. This data is particularly well suited to address the hypotheses presented above and the existing limitations in our empirical understanding of agent behavior. To measure threats, reporting, organizational characteristics, and political repression, NIRI employs a broad and diverse set of high quality data sources available in the United Kingdom (including Northern Ireland).
By collecting data across the various organizational levels of the military, our project is able to document (1) what orders were being transmitted down the chain of command to direct military agents towards engaging in specific behaviors towards particular points or times, (2) what types of coercive behaviors were occurring, as well as when and where they were ongoing, and (3) which behaviors as well as which relevant contextual information was being reported upward to the principals in command. In addition to documentation originating from state security forces, NIRI has collected and coded independent data measuring the behaviors of both challengers and government forces. To gather additional data on political activity pre- and post-military deployment, the project has collected a set of events records from a comprehensive media survey cataloging all politically relevant behavior identified in the Irish and British news media beginning in 1968. The British military and media data are complemented by a set of event data drawn from victims’ statements taken by a human rights organization that was active during the relevant period. Because these statements were taken directly from the victims of repression, they are able to capture a broad set of hidden forms of activities not frequently reported in military documents or published by the media.
Empirical Analysis
As presented above, we are interested in empirically evaluating three sets of hypotheses. The first concerns the relationship between challenger threat and repressive activity, the second addresses the reporting of repressive behavior, while the third concerns the organizational characteristics and repressive behavior. As such we will examine highly disaggregated data on threat, reporting, organization, and repressive behavior across space, time, and organizational unit. In particular, the project will examine the repressive behavior of individuals within the 58 battalions deployed across Northern Ireland during the Troubles, with each battalion referenced to 1) the area it was deployed, 2) the timing of its deployment, and 3) the brigade it served under. The units of analysis will be the battalion-week, and empirical models will estimate the relationship between different organizational characteristics and repressive behavior.
We will begin by evaluating each of the behavioral hypotheses in conjunction with one another. The models will be specified as cross-sectional time-series, and employ battalion level fixed-effects. The aim is to identify whether the different behavioral, organizational, and reporting characteristics correlate with applied levels of political repression, while controlling for rival hypotheses. From there, greater attention will be paid to identifying causality in the relationship between these related characteristics and repressive behavior. Here, we will evaluate each of the competing hypotheses individually using two approaches: a matched sample difference-in-difference (DiD) approach and an instrumental variables design. The matched sample DiD models will employ genetic matching to ensure balance across a range of relevant observable confounding factors (e.g., deployment location, regional levels of development, sectarian settlement patterns, etc.), while also controlling for any remaining unobservable differences between treatment and control groups (e.g., Lyall 2009; Sullivan 2014). The instrumental variables models will exploit the as-if random assignment in battalion deployment created by the British military’s rules on the duration of tours of duty (e.g., Crost and Johnson 2010). In addition to introducing variation in deployment, rules of tour duration also affected levels of threat (as challengers reacted to new battalions), reporting (as different battalions chose different reporting patterns), and organization (as deployment shuffled the number of battalions deployed within each brigade). When combined with the results from the cross-sectional time-series analyses, the causal inference econometrics will provide more conclusive evidence regarding the validity of the causal mechanisms articulated in our theory.
Examining the Level of Challenger Threat. Addressing the first set of hypotheses requires understanding the nature and evolution of the conflict across space and time. Using the data collected from the Associate for Legal Justice and newspapers sources (as detailed above), we will be able to create an event catalog of “actual” challenger threats which details as accurately as possible the activities of the conflict across the country and across time. This catalog will be used to evaluate the challenger threat, its intensity and its distribution to allow us to test our hypothesis about threat perception and repressive behavior across principals and agents.
Examining Organizational Structure. Addressing diverse institutional characteristics during the Troubles requires a detailed understanding of both the organizational structure of the security apparatus deployed in Northern Ireland and the systems of reporting used for communication across different levels of the organization. At the cabinet level there were a number of different committees and offices that had oversight positions above the British Army in Northern Ireland, including the Joint Intelligence Committee, the Prime Minister’s office and Cabinet 9B, the cabinet committee responsible for governing Northern Ireland during the period of direct rule (1972-1998). In Northern Ireland, the Army was lead by HQNI, their command headquarters in Belfast. From there, the Army was typically divided into three brigades, which were assigned spatially to different areas Northern Ireland. Battalions were the principal units composing the brigades and rotating in and out of the province during the Troubles.
Table 1 displays the principal reporting documents produced at each of the relevant levels of the organization and the corresponding variables and hypotheses that each source will be used to test. The documents present accountability at distinct levels in that they represent reporting from one level to another. For example, log files were written by individual battalions but were ultimately delivered to brigades. Serious encounters were reported up the chain of command through the Situation Reports (at the brigade level) and Major Incident Reports (at HQNI). Organizational hypotheses will be evaluated by examining variation in the congruence between the principal directives and repressive behavior.
Examining the Reporting of Repressive Behavior at the Organizational Level. The project will also examine reporting of repressive behavior across different levels of the organization as specified in the above hypotheses. Specifically, we are interested in examining what types of information are captured in reporting at the different levels as well as how it relates to both behavior and reporting at the other levels. To evaluate the content of reporting at different levels we will again examine variation across battalions, but rather than analyze the behavior of the battalions we will examine the relationship between various independent variables and the reports filed by the battalions across time and space. We further extend the analysis by examining reporting at two additional levels of the organization—the brigade and Army HQ.
Our design employs this variation to identify the relationship between reporting at different levels and the hypotheses stipulated above. We focus on how patterns of reporting at each level are related to the directives issued by organizational principals as well as how reporting patterns at each level reflect local and national threats. This will allow the study to investigate directly how accountability metrics relate to both the directives issued nationally and the threats experienced locally.
Our design employs this variation to identify the relationship between reporting at different levels and the hypotheses stipulated above. We focus on how patterns of reporting at each level are related to the directives issued by organizational principals as well as how reporting patterns at each level reflect local and national threats. This will allow the study to investigate directly how accountability metrics relate to both the directives issued nationally and the threats experienced locally.
