by Kyle T Kattelman
What prompted states to participate in the War on Terror? Conventional wisdom concludes that the endeavor is an unpopular exercise in US imperialism, yet this argument is juxtaposed with the overwhelming amount of international support in its initial stages. Additionally, while there is a great depth and breadth of information on aggregate terrorist attacks and their theoretical motivation, there is relatively little with regards to counterterrorist behavior. This study represents the first of its kind to examine from a global perspective the counterterrorist behavior of states by linking it to the conflict theories of general and immediate deterrence. The results will show how democratic characteristics inhibit military commitment while alliance obligations act as an outside constraint that engenders preemptive behavior. However, once committed militarily, state capabilities are the main influence on the level of preemptive action applied. This analysis supports the utilization of traditional conflict theories when examining state counterterrorist behavior.
Keywords: terrorism, counterterrorism, alliances, collective action
Introduction Shortly after the attacks on 9/11, President George W. Bush, in The National Security Strategy of the United States of America (NSS 2002, 1), outlined three crucial tasks to protecting the US from further violence, stating, “We will defend the peace by fighting terrorists and tyrants. We will preserve the peace by building good relations among the great powers. We will extend the peace by encouraging free and open societies on every continent.” These goals stand in stark contrast to those outlined under the Clinton administration in 1998, which argued that the main objective of the US was to bolster economic prosperity by promoting democracy and human rights abroad (NSS 1999). President Clinton’s statement was broad in comparison to the pointed and focused goals of President Bush. More important, Clinton’s statement implied peace, an attitude not present in the declaration of the Bush administration. Given the course of events that just transpired, its absence should come as no surprise. The attacks on 9/11 were an unparalleled global event that singlehandedly shifted US security strategy and the global system seemingly overnight. No single international crisis can rival the scope and depth of global consequences that the hijackings of four commercial airliners harkened in that day. Events that come close, such as the assassination of Archduke Franz Ferdinand and the concomitant commencement of WWI, were as much a product of the complicated web of international alliances and diplomatic blunders as it was an act of terrorism (Gaddis 2004). Indeed, 9/11 truly stands alone in terms of its tactical success and strategic implications.
A global insurgency of this scale warrants global action, and global action has been taken (Kilcullen 2005). The invasions of Afghanistan and Iraq have resulted in trillions of dollars in expenditures, thousands of casualties, and a countless amount of international protest. How has a single incident triggered such a severe response? How do threatened nations react to al-Qaeda and their affiliates?
The theoretical conclusions reached by Arce and Sandler (2005) and Enders and Sandler (2006) state that deterrence is a problem of liberal democracies. What can be argued, however, is that global terrorism is a problem of all states due to the fact that attacks could happen anywhere in the world. However, democracies do feel the brunt of this threat and have experienced a disproportionate amount of violence from international terrorism (Li 2005). So what aspects of liberal democracies and, more broadly, all threatened states, affect counterterrorist actions? Why do some states engage in military offensives, while others stay on the sidelines to shore up defenses on the home front? The theoretical differentiation between general and immediate deterrence (Zagare and Kilgour 2000; Quackenbush 2011) and its ties to regime characteristics can help lend insight into these puzzles. To date, scholars analyzing terrorism have examined the causes, targets, and types of attack (Ginges 1997; Rees and Aldrich 2005; Miller 2007). However, they have not examined counterterrorist measures on an aggregate scale, mostly relying instead on case studies. Conversely, aggregate-level analyses focusing on the characteristics of terrorists have led to conclusions that can be generalized across a broad spectrum of terrorist activities regarding their causes and strategies (Crenshaw 1981; Hoffman 1995; Newman 2006; Benmelech, Berrebi, Klor 2012). The research conducted here represents an attempt to offset this disparity by employing large-N quantitative analyses on the counterterrorist activities of countries in an effort to attribute state characteristics to general counterterrorist behavior.
This analysis represents an attempt to explain the dilemma faced by liberal democracies when combating terrorism by tying it into more traditional conflict theories of deterrence. What follows is a discussion of the perceived belligerence of US foreign policy and its activities in the Middle East. The purpose is to explain why traditional power balancing theories do not account for the coordinated multilateral operations seen today in modern conflict. The reason behind this is that most of the world feels the threat of global terrorism and does not balance against the US because they share its overarching goal of eradicating al-Qaeda. What follows next is a discussion of general and immediate deterrence, with an effort put forth explaining how it is possible for states to deter terrorism, and a discussion of why advanced liberal democracies are in a state of immediate deterrence.
Next, discussions of the dilemma of liberal democracies and resulting hypotheses are outlined, examining the paradox of how expressly targeted states are the ones most reluctant to act, how alliance commitments work to overcome the misgivings of democratic leaders on behalf of military intervention, and how capabilities drive the level of preemptive action once a state commits. Finally, the results are presented, followed by a conclusion noting both the strengths and weaknesses of applying aggregate analyses to counterterrorist behavior.
Traditional power balancing theories argue that the US should be experiencing international backslash due to its rank as the top military power and its subsequent interventions into the Middle East (Leiber and Alexander 2005). What is to be argued is that the lack of balancing stems from the shared sentiment of the international community that a global terrorist threat exists. That threat sets the stage for conceptualizing the multilateral counterinsurgency under more traditional conflict theories of general and immediate deterrence.
Though the events of 9/11 precipitated the mass invasions of Iraq and Afghanistan under the politicized misnomer of the “Global War on Terror,” the actual declaration of war came in a much subtler form from the perpetrators of the attacks a few years earlier. On February 23, 1998, Osama bin Laden issued a fatwa declaring war on the US, Israel, and all liberal democratic nations, which prompted a following two-phase strategy for a global jihad against the West (World Islamic Front 1998). Though the events were treated indifferently from the West, the credibility of their intentions crystallized in the unprecedented and historic destruction of WTC Towers 1 and 2, the crash of a Boeing 757 into the Pentagon, and the failed hijacking of Flight 93. An outlier in terms of destruction (Mueller and Stewart 2012), the attacks on 9/11 signaled that the perpetrators intended on carrying out a global insurgency. And though such violence is unlikely to be replicated, al-Qaeda has a presence in at least 40 countries, indicating the extent of its operations and its ability to invoke fear (Kilcullen 2005).
Viewing this theoretical reasoning in light of the military measures of the US and its coalition partners, the goal of the operations in Afghanistan attracted participants because the motive of the US was not entirely self-serving. Arguably, the US was acting on what it saw as an immediate and global threat. This assertion is further justified by the fact that NATO (2005) invoked Article 5 of its self-defense pact for the first time in its history. International terrorism threatens a host of target states that would all be better off if the source of the threat was eliminated. In this sense, the removal of al-Qaeda and its affiliates can be thought of as a public benefit (Enders and Sandler 2006), with the broader goal of global security precipitating a united coalition.
Hence, a more accurate appraisal would be that states do not balance because they lack the motivation. As argued previously, the attacks of 9/11 can be seen as an unprecedented and devastating attack on the international community, with liberal democratic states serving as the specific target. Yet all states felt some degree of threat as evidenced by the lack of balancing. A more general view of international policy supports this assertion. Lieber and Alexander (2005) find no change in the Cold War alliance patterns following the US’s aggressive shift regarding missile defense in the late 1990s, arguing that there was little need to change the alliance patterns in the post-Cold War environment. Russia and China did give vague remonstrance to US involvement in Afghanistan and Iraq in 2002 and 2003 but took no further action. Additionally, the opposition to these actions was not cleanly delineated along cultural lines; states with large Muslim populations were also active participants in the global War on Terror, with countries such as Jordan, Kuwait, and Saudi Arabia lending cooperative efforts (Country Reports on Terrorism 2005). The main point is that all states, even predominantly Muslim ones, view international terror as a threat. Indeed, most al-Qaeda activity occurs in the historic caliphate region in the Middle East and shows evidence of broadening beyond its original borders (Kilcullen 2005). In general, the global consensus may be dissatisfied with US foreign policy, but they are inactive in response to specific demonstrations of perceived US aggression (Voeten 2004). Even the most belligerent of US actions, such as its national missile defense policies, fail to elicit a response (Quackenbush and Drury 2011). This does not mean that the global community is satisfied with US foreign policy. It does, however, imply that there is no one policy that serves as the fulcrum upon which to successfully leverage international pressure.
Hence, the absence of balancing in regards to US interventions into the Middle East stems in part from the prevailing fear that particular states have regarding international terrorism, specifically concerning al-Qaeda. This mutual concern is analogous to the conflict situation of general deterrence. The global threat of international terrorism puts all states on alert (Arce and Sandler 2005; Enders and Sandler 2006). That does not mean that the level of threat is felt equally or that state reactions are homogenous. What it does signal is that the fatwa issued by bin Laden theoretically put all threatened states on a higher level of alert, a situation of general deterrence. When states are directly challenged, as the US was after 9/11, the threat of an international terrorist attack is greater. In this instance, the US realized that al-Qaeda possessed a credible threat. Since terrorists do not claim a particular state and rely on fear, threats of an attack are often not credible (Bapat 2006). The attacks of 9/11 indicated to the US and other liberal democracies that al-Qaeda possessed a credible threat and thus selected liberal democracies into a situation of immediate deterrence. This study represents a closer examination into the institutional characteristics that motivate offensive state action. In order to examine this in more detail, a closer look at the specifics of deterrence is necessary.
In order for deterrence to work, two conditions must be present. The first is that the defender must be aware that there is a challenge to the status quo. In other words, a threat must exist. The second is that the decision making process on behalf of the adversary must be influenced by costs and benefits (Quackenbush 2011). This is all predicated on the base assumption that the actors are rational, an assumption that is not without its detractors. If terrorists are irrational they do not perceive costs and benefits and thus cannot be deterred. The popular argument championed by the press is that terrorists do not measure costs in the traditional sense that states use (Trager and Zagorcheva 2006). Since the global insurgency consists largely of radical Muslims who view themselves as martyrs, the threat of costly retribution would only confirm their goal of a glorious death (Pape 2003). Framed in this logic, terrorists do not fear punishment.
The above reasoning implies a reliance on procedural rationality, which argues that rationality is defined by the ability of actors to have access to all information, process it accordingly, and not be affected by their cognitive limitations or emotions (Lebow and Stein 1989, 1990; Rhodes 1989). In this regard terrorists are not rational because they are religious fanatics that often engage in suicidal missions. However, the assumption used in this article, as well as in rational deterrence theory and rational choice theory in general, is that of instrumental rationality. Broadly stated, an instrumentally rational actor is one who, when confronted with two choices, chooses the one that yields the highest outcome. This definition accounts for any preference, however “irrational” it may seem, as well as for incomplete information, which can often lead to mistakes in choices of strategy. Preferences are subjective in nature and are shaped by beliefs, cognitive limitations, and emotions, but this does not mean that an actor is irrational (Quackenbush 2004, 2011). Rather, actors are rational merely if they choose strategies in an effort to obtain their highest possible preference. Hence, to flesh out the interplay between al-Qaeda, international terrorism, and targeted states using the deterrence framework, it is assumed here that both sides possess instrumental rationality.
For the purposes of this article, there are two stages of deterrence that actors can reside in. General deterrence is the broader conceptualization of the two and is less focused on crisis management than it is with everyday decision-making in international politics to maintain defense (Quackenbush 2011). This form of deterrence simply embodies the defensive stance that every state exhibits when it maintains secure borders or that terrorist organizations display when they maintain an arsenal of weapons or armed combatants. More threatening is the event of immediate deterrence, a situation in which at least one actor is seriously considering mounting an attack while the other side is threatening retaliation to prevent it. A situation of immediate deterrence indicates that general deterrence has failed, for a state’s natural defenses have been unable to avoid an international crisis (Quackenbush 2011).
When examining general and immediate deterrence in the context of the War on Terror, the fatwa issued by bin Laden in 1998 reorganized the global system into those states that are targeted by this declaration, essentially putting them into a state of general deterrence. However, this action alone was not sufficient to place states into the context of immediate deterrence, given the lack of attention initially paid to this pronouncement. Terrorists often lack the credibility to appear sincere in their goals. Therefore, they rely on violence to signal their intentions. The attacks on 9/11 served as a credible signal to the US and its allies that they were sincere in their intent of driving foreign troops out of Muslim territory and establishing an Islamic state, for it is unlikely that the US would have removed their troops from Saudi Arabia by the mere threat of an attack on the WTC towers (Kydd and Walter 2006). In this regard, 9/11 placed the US and its allies in a state of immediate deterrence, which ultimately led to a deterrence failure in the wake of the invasion of Afghanistan. To explain the breakdown into international conflict it is important to examine another critical set of actors, the targeted states, and the factors that conditioned their response.
The opening salvo and ensuing military offensive of Operation Enduring Freedom were spearheaded by the US and a fighting force composed largely of advanced liberal democracies. However, theories of counterterrorist action specify that liberal democracies are situated in a prisoners’ dilemma with regard to military action; democratic mechanisms restrict leaders in their pursuit of violent solutions due to the costs inflicted on the invading state (Enders and Sandler 2006; Arce and Sandler 2005). Hence, states, particularly democracies, should refrain from engaging in a military offensive in the face of a terrorist threat. This assertion is strengthened when examining the impact of war costs. Democratic leaders depend on a wider base of electoral support to maintain power. As a result, democracies are especially sensitive to war costs, since the electorate it depends on shoulders the burden of conflict (Valentino, Huth, Croco 2010). As the theatres of Afghanistan and Iraq illustrate, preemptive actions against terrorist groups can evolve into a counterinsurgency, which is a costly endeavor. Fought against a non-state entity, these strategies call for a large contribution of troops to meet objectives with a long-time horizon (Nagl 2002). Additionally, these interventions feature complex goals that leaders must articulate clearly to their electorate if they are to gain public support (Tellis 1996). This suggests that democratic leaders will have a difficult time convincing their constituents to support massive military contributions in pursuit of counterterrorist and counterinsurgent designs, which are inherently abstract (Record 2007).
Yet this is clearly not the case as the current campaigns in Afghanistan and Iraq demonstrate. How is this so? One solution to a prisoners’ dilemma is the imposition of outside constraints (Lipson 1993). In the international community, alliances can serve as this constraint. From a liberal perspective, collective security organizations such as the UN and NATO are designed with the precise goal of overcoming collective action problems (Keohane 1993). Additionally, alliance formation allows democracies to minimize costs and increase capabilities by spreading the burden across a coalition of states while simultaneously mitigating the potential political backlash (Valentino, Huth, Croco 2010). Hence it can be expected that alliance commitments will have an important role to play in the decision to commit military forces.
The ties between democracy and development say little of the connection between development and the level of military action a state applies. It is clear from al-Qaeda rhetoric that the target of their ire is Western democratic nations, and liberal democracies do suffer more international terrorist attacks (Li 2005). However, the bulk of terrorist attacks against liberal democracies are not actually carried out within their borders, and there has not been an attack that has approached the magnitude of 9/11 at the time of this writing (Wilkinson 2011). This suggests that terrorists are opportunists that seek to influence powerful states by carrying out attacks against the citizens of liberal democracies and other targets in less-secure venues, where their chances of success are higher (Enders and Sandler 1993; Pluemper and Neumayer 2010). Due to the fact that these weak states are often either unable or unwilling to eradicate the terrorist threat from their territory (Byman 2005), and the fact that the intended targets are advanced democracies, it is expected that, once a state chooses to engage in preemptive action, more advanced countries that are the intended targets will demonstrate the highest levels of preemption.
The goal of this study is to apply the theories of general and immediate deterrence to the decision calculus of states threatened by the global insurgency composed of al-Qaeda and their affiliated organizations. Although many states are in a stage of general deterrence, liberal democracies are argued to be the main target of terrorist violence. Thus, they feel the threat of terrorism more acutely than other states. The dilemma faced by liberal democracies is that, although they are the main targets, they are subject to a populace that is averse to violence and the restrictions in civil liberties that often follow military action and security issues, which leads to the first hypothesis.
H1: The more democratic a state, the less likely it is to engage in preemptive action.
This situation resembles a prisoners’ dilemma. Self-interested states are unwilling to work together to eliminate a terrorist threat due to the costs involved, instead allowing the organization to carry on with its violence, making everyone worse off. A mitigating factor is the existence of alliances that act as an outside enforcement mechanism to ensure cooperation, while also serving as a mechanism to minimize wartime costs. Though not all alliances are reliable, the fact that NATO invoked the self-defense pact of Article 5 of its C9harter serves as a strong test for the reliability of alliances as a solution to the prisoners’ dilemma. This leads to the second hypothesis.
H2: Alliance members are more likely to engage in preemptive action.
So far, the hypotheses apply to states in a situation of general deterrence. When looking at all threatened states, those that are democracies are less likely to cooperate and those that are in an alliance are more likely to cooperate. However, once a state is selected into the conflict by choosing preemptive action, it enters into another decision process. This supply side view of preemption is also supported from a demand oriented, terrorist-motivated factor, for liberal democracies, particularly advanced liberal democracies are the main target; their level of development equates to a larger amount of influence on the international stage. This leads to the third hypothesis.
H3: States with greater capabilities lead to higher levels of preemption.
With the hypotheses outlined, a discussion of the research design will follow. In order to correctly model the situations of general and immediate deterrence, it is necessary to outline the method of selecting cases to construct the universe of states under general deterrence.
Case selection on which states are threatened from al-Qaeda and their affiliates is a particularly thorny issue. In order to test for the effects of democracy and development, there needs to be sufficient variation, yet the cases need to properly approximate the group of threatened states. Islamic terrorism makes clear from their rhetoric that Western states are the primary source of their angst, particularly the US and its ties with Israel and the stationing of their troops on Muslim soil. However, the term “Western” is highly ambiguous, and international terrorist attacks are distributed unevenly (Global Terrorism Database, 2012).
To mitigate this problem, states will be selected based on their appearance in the Patterns of Global Terrorism (Country Reports on Terrorism after 2004). Countries included in the report consist of those where significant terrorist acts occurred and countries where the US has sought cooperation during the previous five years in the prosecution or investigation of acts of international terrorism against the US. Since this study concerns the immediate actions of states after the attacks of 9/11, the cases consist of state actions regarding the decision to join coalition forces for the invasion of Afghanistan for the year 2002 and comprise 103 observations.
The first dependent variable is constructed to explicitly determine which states selected to commit to military action and which states did not. It is coded as 1 for those states that either committed troops for the purpose of participating in Operation Enduring Freedom or allowed for the stationing of coalition forces on their soil and a 0 for any other action. States that allow for the construction of military bases are included as engaging in military action due to the evidence indicating that states are the target of terrorist attacks when they allow for the stationing of troops (Byman 2005). The method used to estimate the first model will be a simple logit. This is the most effective method of determining the characteristics that foster preemptive cooperation with the US, for the data set constructed is set up to represent the universe of cases in general deterrence that could be selected into immediate deterrence. Simply examining the characteristics of the coalition forces does not account for the factors that led to their decision to preempt and thus induces selection effects (Fearon 2002).
Another valid method of modeling the first equation would be to utilize an ordinal logit, taking into account non-military deterrent actions such as installing metal detectors or increasing surveillance and arrests. Conceptually, military preemption could be viewed as an extension of deterrent action, as both moves are directed at a common enemy. The only difference is that deploying troops inherently entails more costs, however they are both conducted to achieve the same end result, namely the prevention of a future terrorist threat. However, framed differently, deterrent and preemptive strategies could be conceptually distinct. Whereas deterrent actions regarding the War on Terror are in response to a specific threat, namely al-Qaeda, their effects are broad and do not serve as a pointed effort to specifically eliminate a threat. Generally speaking, deterrent actions could be undertaken to prevent any attack and do not impose as much costs on the international community. National defense is a hallmark of state sovereignty; efforts to uphold it are only met with negative reaction when they significantly intrude on individual rights (Enders and Sandler 2006). Military offensives, while directed at the same end, is a different form of means altogether. Public outlook on foreign affairs is fairly minimal, supporting a general aversion to a state involving itself beyond its borders. Additionally, military action almost certainly guarantees the loss of life, a fact that many states are unwilling to tolerate. Crossing the threshold into the sacrifice of human life is a heavier concept than the decision to install more metal detectors at an airport. In light of these arguments, the two concepts of deterrent and military action may be theoretically distinct and linked only indirectly through their focus on a common threat.
A second dependent variable measures the level of preemptive action and is simply the number of troops an individual state committed to Operation Enduring Freedom in Afghanistan. It is possible to picture the invasion in Operation Iraqi Freedom as a preventive attack to prevent Iraq from acquiring weapons of mass destruction (Levy 2008). This logic implies that the US did not see Iraq as an immediate threat but an eventual one, emboldened by the al-Qaeda attacks of 9/11 to acquire WMDs and use them against the US. Many in the international community view the acquisition of nuclear weapons as a threshold that warrants a different treatment toward an opposing enemy. Indeed, such an acquisition by Iraq would certainly improve its capability to do irrevocable harm against powerful countries against the US. By contrast, the attacks against Afghanistan during Operation Enduring Freedom more closely resemble a preemptive attack, for these operations were seen as a direct response to the attacks of 9/11. This could also explain the timing of the events, as the invasion of Iraq began a full year after the invasion of Afghanistan. Thus the dependent variable modeled here is the number of troops deployed to Afghanistan in 2002, as it is a more appropriate test of the factors that foster state coordination in the face of a terrorist threat.
To test the effect of regime type, Li’s (2005) measures of democracy will be used. The first component is government constraints, a measure of the institutional constraints a chief executive faces, taken from the POLITY IV dataset (Marshall and Jaggers 2000). This measure is based on a 7-point scale, with 1 indicating complete executive authority and 7 indicating executive parity or subordination. The second component operationalizes democratic participation and is a combination of a binary indicator of a democracy (6 or higher difference between POLITY IV DEMOC and AUTOC score) and a measure of electoral participation. If the state has a POLITY score of 6 or higher, political participation measuring the percentage of the population that voted in the general election will be used from the International IDEA Voter Turnout database. Otherwise, a score of 0 will be assigned. The third measure used captures the overall effects of democracy through a standard POLITY score.
The log of a state’s GDP represents the measure of development. It is expected to have a positive effect in both models. In order to account for the link between development and military capabilities the logged ratio of a state’s military capabilities to the US is included to isolate the effects of the GDP variable. Finally, an alliance variable is included indicating whether a state is a NATO member.
The remaining independent variables represent an attempt to capture the various costs and benefits that are consequential of a decision to commit troops to combat a terrorist threat. The most explicit example of a state overcoming its apprehension to preemptively strike is when it bears the brunt of terrorist attacks. The logic behind this is straightforward. The benefit that a state derives from the elimination of a terrorist organization is a function of the level of threat that terrorist organization presented to the state. To accurately capture this threat, with regards to this study, only attacks from al-Qaeda and affiliated organizations will be used. Though this greatly narrows the scope of the terrorist threat, it can be reasonably concluded that the invasion into Afghanistan was not undertaken to eliminate all terrorist organizations. Limiting the inclusion of attacks in this manner may decrease the significance of this variable, as there might not be substantial variation of attack data among the cases to detect an influence. Nonetheless, specification of the threat is paramount, and it is difficult to make a case that violence from unrelated terrorist organizations would have any effect on such an important policy decision. Identification of al-Qaeda and its affiliates is taken from Stanford University’s Mapping Militant Organizations. Additionally, since international terrorist attacks are rare relative to the amount of domestic terrorist attacks, a casualty count will also be included. It is expected that as both these variables increase counterterrorist action should increase as well. Casualty and attack data are taken from the Global Terrorism Database. Also included are measures of the percentage of Muslims within a state’s population and a dummy variable indicating if a state is located in the Middle East.
The results of the first model are presented in the first column of Table 1. The simple logit results demonstrate that the decision to commit is largely a function of alliance commitments. The coefficient on NATO membership is of a large magnitude and a high significance, suggesting that alliance commitments are the driving force behind the initial invasion of Afghanistan and that democracies minimize war costs by engaging in coalition warfare. This also supports the theoretical argument that alliances are a possible solution to the prisoners’ dilemma of counterterrorism. This finding should come as no surprise, given the fact that the ISAF forces are NATO-led. However, looking beyond the correlation between NATO and ISAF, a more substantive explanation can be established. As mentioned before, the attacks on 9/11 marked the first time that NATO invoked the self-defense pact of Article 5. This article outlines the conditions for a collective response, stating that an attack against one or more allies on European or North American soil is considered an attack on all. The decision to act per the conditions of Article 5 lends considerable explanatory leverage to alliances and the decision to militarily intervene. Originally conceived as a Cold War institution, NATO has evolved to handle an array of conflict dilemmas that are increasingly directed at non-state actors. What is of particular note is the event that triggered Article 5 and the insignificance of the attack and casualty variables. As the first model shows, the number of attacks and casualties a state has suffered the previous year are insignificant predictors of the decision to commit forces. Given that the events of 9/11 are responsible for the invocation of Article 5 and the ensuing “Global War on Terror,” how do these two variables figure into the equation?
First it should be noted that, statistically, the event of 9/11 is an outlier. No terrorist attack before or since has killed more than a few hundred in a single event (Mueller and Stewart 2012). It is also worth noting that the method of attack is highly unique, as shrewdly put by Russell Seitz that “9/11 could join the Trojan Horse and Pearl Harbor among stratagems so uniquely surprising that their very success precludes their repetition.” (Mueller and Stewart 2012, 90). Hence, the variation of attack and casualties included here are obscured by the events that transpired on 9/11, effectively demonstrating the psychological threat that terrorism is capable of inflicting, however rare the event may be. Such an outcome confirms the solution to the collective dilemma outlined by Arce and Sandler (2005) where a particularly destructive attack motivates state action. In this case, the situation of immediate deterrence created by 9/11 prompted the triggering of Article 5, precipitating the resulting invasion and deterrence failure. in this light, the insignificance of attacks and casualties becomes easier to grasp. Annual death tolls average around 1,000 a year from international terrorism, supporting the notion that 9/11 was a watershed event, though the influence it has on other attacks is minimal (Sandler 2003). This suggests that attacks are at the very least an incomplete measure of the threat of terrorism.
What are surprising are the insignificance of the logged variables GDP and the ratio of Military Expenditures, which capture the effect of state capabilities. With regards to the War on Terror, alliance commitments provide most, if not all, of the driving force that prompted states to engage in military offensives against terrorist organizations. Development appears to have played little to no role in the decision to engage. This can also be explained through the prism of alliance commitments. The Country Reports on Terrorism note that states can engage in a variety of counterterrorist actions, including freezing assets, sharing intelligence, providing flight oversights, participating in joint military exercises, and offering diplomatic initiatives. These effects are not captured in the dependent variable, as it is theorized that contributing military troops is an inherently distinct form of participation that garners a higher amount of both military and political costs. As such, variables consistently correlated with power, GDP and military expenditures, are also co-opted under the power effect of alliance commitments, once again highlighting the impact of catastrophic transnational terrorist events and the ability of states to overcome a collective action problem to present a unified front at the outset.
Also contrary to popular opinion is the obstructive pull that democracy has with regard to preemptive action. This is effect is particularly robust, given both the method of case selection and the control variable of NATO membership. Democratic states are less likely to support preemptive action, a reflection of the resident population’s aversion to casualties and costs from conflict. This problem is exacerbated by the amorphous threat that terrorism presents. As previously stated, transnational terrorism is an exceedingly rare event, and the threat of attack varies from state to state. The difficulty in quantifying a benefit from participating in the elimination of a nebulous non-state enemy implies that contributing forces is an unpopular option among many. More so than autocracies, democracies seek to minimize the costs of participation in conflict (Valentino, Huth, Croco 2010). Given that contributing military forces almost always involves casualties, and that retaliatory attacks could result from participation, democracies should be generally unwilling to participate as the costs become more manifest than the benefits. This result compounds the dilemma argued by Enders and Sandler (2006). Although liberal democracies are the main target of international terror, they are the most reluctant to take preemptive measures to eliminate the threat. Not only are the citizens of liberal democracies generally against curtailing civil liberties to fight terrorism, they also oppose the mobilization of troops to eliminate the threat. These findings are surprising, given that the fatwas issued by bin Laden explicitly target wealthy liberal democratic states, particularly the US, Israel, and their allies.
Muslim Population shows a positive effect on the decision to engage in preemptive action, though it remains unclear as to the relationship between a state’s Muslim population and its decision to preempt. This could be a reflection of the alliance links between the US and Middle Eastern countries such as Egypt and Saudi Arabia, or it could be a reflection of the Muslim population within liberal democracies themselves. Looking at the Muslim population within liberal democracies points to the possibility that they are different from their Middle Eastern counterparts. It could be the case that Muslims immigrating to Western democracies are highly skilled and of a socioeconomic status that facilitates mobility. Norris and Inglehart (2012) find evidence of this self-selection effect at work regarding Muslim immigration to Western democracies, noting that upon relocation, immigrants gradually adopt some of the cultures, ideas, and customs of their new host country. The allure of these nations, with their increasing labor demands and high regard for human rights, has prompted a boom in international migration (Hunter 1998, 2002; Israeli 2008). These factors facilitate assimilation into a democracy more readily and suggest that Muslims, more than other religions and races, condemn the actions of their extremist brethren.
However sound the logic, empirically this is difficult to support. Although Boyle (2010) finds evidence that a majority of Muslims do not share the extremist ideology that radical groups such as al-Qaeda champion, Muslim immigrants to non-Muslim states are comprised of a wide variety of nationalities, ethnicities, and socioeconomic backgrounds, indicating that a general explanation may be elusive (Kurth 2002; Kahera 2002). Additionally, public opinion polls on Muslims in the US regarding thoughts on al-Qaeda and the War on Terror find that there is no statistical difference of opinions between the US and European countries as well as between the US and the Middle Eastern countries of Morocco, Indonesia, Pakistan, and Egypt (McCauley and Scheckter 2008).
Larger numbers of Muslims in troop-contributing nations do suggest that their presence is a factor, though it is not clear what the connection is. It could be a defensive action at the state level in which countries with larger Muslim populations contribute for fear of a recruitment factor within their borders. It has been postulated that the evolving hardline policies France has taken against Islamic extremism in Libya, Mali, Syria, and Iran stem in part from its problem regarding the recent growth in militant Islamism that has been precipitated by the large number of Muslims that comprise France’s underclass (Wallerstein 2013). This exemplifies a distinct connection between a country’s domestic demographic and its international policy. Hence, hardline policies are a defense mechanism against the spread of radical Islamism. Terrorist actions send a message to their target audiences with the intention of both spreading fear and generating recruitment. For example, when al-Qaeda executed 9/11, their actions carried two separate messages simultaneously to two target audiences. The first was sent to the US and its allies conveying the intention that more attacks were on the way if it did not change its foreign policy in the Middle East. The intended audience and message that extended beyond the immediate sphere of those actually killed and injured in the attack are crucial for an act to be defined as terrorism (Enders and Sandler 2006; Hoffman 2006). The second served as a rallying call to arms to the umma throughout the world, a message that al-Qaeda works to cultivate very diligently through propaganda and video documentaries (Amble 2012). This concern for the target audience is what compels a state to temper its counterterrorist actions, for attacks that are too severe could result in a swell of terrorist recruitment and a fear of reprisal attacks. Rosendorff and Sandler (2004) argue that states responding too harshly to terrorist organizations run the risk of arousing a latent group of individuals who share the terrorists’ views, angering them to the point that they become recruited into the organization. The hypothesized negative effect was seen as an artifact of states with large Muslim populations shoring up defenses to deter potential terrorist attacks occurring against foreigners on their own soil. However, here it appears that Muslim population has a positive function regarding preemptive commitment. These findings warrant further examination, and are beyond the scope of this study.
Turning now to the next model, the distribution of the second variable measuring the number of troops committed to ISAF is highly skewed to the right, indicating that many states contributed no troops, while others contributed a significant number. A tobit model is used for the second set of results. The theory states that a two-step process is at work regarding the number of troops to send. Additionally, though preemptive action includes troop contributions, it is merely the most visible form of action available. Allowing basing rights and military supplies can be thought of as another form of preemption, though it is difficult to compare a measure of basing rights against the number of troops committed to battle. Consequently, they are not included here. Preemptive action can also include the use of Special Operation Forces (SOFs) to complete covert military actions ahead of standard troop invasions. It is known that the US employed SOFs in its preliminary actions against Afghanistan and Iraq (Woodward 2002; 2004). However, due to the secrecy of these missions, they are often classified and difficult to quantify. These factors indicate that there could be a number of censored observations within the data. To account for this, as well as the two-step decision process behind troop contributions, a tobit model is used.
The results are presented in the second column of Table 1 and show strong support for the effects of capabilities on the number of troop commitments. Indeed, taking into account NATO membership, in which membership leads to a contribution of approximately 93 troops, the amount of preemptive action taken is a measure of state capabilities as captured by the logged variables of GDP and the Military Expenditures ratio. Tobit coefficients can be interpreted as OLS, however the effects are framed in terms of a latent dependent variable (Wooldridge 2009). Looking at the effects of GPD first, a one percent increase in a state’s GDP leads to an expected personnel increase of around 0.76. Since GDP is a widely dispersed variable, a more identifiable interpretation could be teased out. A ten percent increase in a state’s GDP would result in an expected increase of ISAF personnel by 7.6. Given that the average ISAF troop contribution in 2002 is 82.56, an increase in GDP by ten percent increases expected troop contribution by .09, almost a ten percent change. The military expenditures variable consists of a ratio of military expenditures between each state as compared to the US. The negative coefficient indicates that the larger the discrepancy between the US and a given state’s military capabilities the less the expected personnel contribution will be given from that state. This supports a behavior of buck passing whereby less capable states rely on the stronger members of their cohort to shoulder the responsibility (Arce and Sandler 2005). The average ratio in expenditures is 2.02, indicating the wide discrepancy in troop contributions that stem from the gulf in capabilities. Indeed, a one unit increase in the ratio leads to a decrease in expected troop contributions by approximately 163, lending strong support to the hypothesis that military expenditures are a major driving force behind participation. The growth in the spending disparity between a given state and the US leads to a significant drop in troop contribution. This should come as no surprise, as the US was the vanguard of the initial decision to invade. The results in the second column of Table 1 show that even after a state decides to participate, it still shirks responsibility to more capable allies. Its contributions are commensurate with its capabilities.
Overall, the results reinforce what the preceding theory states. The decision to engage in preemptive action is hindered by the level of democracy a state maintains. However, once states decide to engage in preemptive action, which is largely influenced through alliance commitments, the amount of force used is due to state capabilities. This is made all the more convincing by the complete lack of significance on all other variables within the model. The logic behind this relationship is straightforward; the amount of force used is a product of the amount of force a state is capable of producing.
The results shown here support the hypotheses presented. Democracy acts as a screening procedure that dismays leaders of democratic nations from taking preemptive action. This argument is made stronger when controlling for the alliance commitments through NATO membership, which largely consists of liberal democracies. Indeed, alliances exhibit a strong pull in persuading states to engage in preemptive action and also account for the large number of troops that a state commits. It appears that democracy captures the residual effects of the decision to commit and temper a state’s willingness to get involved. When viewed in this light, it should be no surprise that capabilities explain troop contributions, due to the fact that the invasion of Afghanistan was largely led by liberal democracies. Consequently, it seems likely that the populace of a liberal democracy cares more about the initial decision to get involved rather than the level of preemption. Indeed, the absence of democratic factors on troop contributions could be a reflection of the population’s initial and short-lived willingness to support military action and the inevitable casualties that follow once the decision was made to commit (Mueller 1970; Parker 1995).
Another interesting finding is the insignificance regarding the number of terrorist attacks a state receives as well as casualties suffered. Although the events of 9/11 were the most catastrophic international terrorist attacks ever conducted, they were not enough to generate a significant effect on state action. This contributes to the argument that terrorism best works through psychological rather than physical damage (Mueller 2006). The argument becomes even more apparent when one considers the fact that US expenditures on homeland defense alone since 9/11 exceed $1 trillion (Mueller and Stewart 2012) . Unfortunately, the psychological costs imposed on a state are much more difficult to measure and are not explicitly modeled here.
The interplay between state characteristics, democratic characteristics, and threat levels regarding decisions to engage in preemptive action are difficult to elucidate. The preliminary findings here suggest that the decision to engage is largely a function of alliance commitments tempered by the effects of democracy. There is no doubt that the targets of Muslim extremists are specifically aimed at Western-liberal democratic values, however, the fact that these variables lead to a decline in the probability of military actions reinforces the dilemma of liberal democracies outlined by Enders and Sandler (2006).
It is here that military capabilities take over. This study offers a preliminary look into the mechanisms driving counterterrorism and suggests separate levels of counterterrorism that draw upon separate influences. All capable states should take deterrent action, because these actions entail little retributive costs. A terrorist organization is unlikely to attack because a state upgrades its homeland security, but it is conceivable that it would attack if said state invades its territory to engage in combat. The findings indicate that democracies are pressured to stop at defense when deciding whether to cooperate with the US in taking the offensive. Alliance obligations are sufficient to overcome this, however, and serve as the impetus for military engagement of the enemy. The amount of force then applied in the preemption stage is largely a product of the resources at hand.
The evidence also lends support to the arguments of general and immediate deterrence as stated here. The data represent the universe of cases susceptible to international terrorist attacks from al-Qaeda and their affiliates, as they are all states that have either coordinated counterterrorist actions or been involved in a terrorist attack with the US in the last five years. Though terrorist organizations often lack the ability to project credibility like a state, the attacks of 9/11 sent a credible signal to the US that additional attacks were imminent if it did not soften its presence and influence in the Middle East. From the theory outlined here, it was argued that those states selected into immediate deterrence were mostly liberal democracies. However, instead of engaging the immediate threat of future terrorist attacks, these states were reluctant to strike, only to be prompted into action by alliance commitments. The unprecedented destruction of the attacks on 9/11 resulted in the NATO invocation of Article 5, which was sufficient to overcome the collective action dilemma facing the states in immediate deterrence. Thus, even in the face of an imminent threat, democracies rely on the external constraint of alliance commitments to engage the enemy, with capabilities driving the level of preemptive force applied.
The results demonstrate the feasibility of using aggregate empirical analyses to analyze counterterrorist phenomena while simultaneously linking this behavior to more traditional conflict theories. In light of this however, there are several additional points worth mentioning. A valid argument concerns the inability of alliance commitments to overcome international reluctance to intervene in Iraq in 2003, a point that US President George W. Bush circumvented by assembling an ad hoc “coalition of the willing.” In this case, the alliance formation was dictated by policies enacted unilaterally by the US (Gordon and Shapiro 2004). This displays the limits of alliance constraints when utilized for counterterrorist ends and underscores the necessity of context when framing a terrorist threat. In the case of Afghanistan, identifying the perpetrators of 9/11 and their capabilities and intentions were clear. In the case of Iraq, tying Saddam Hussein’s regime to WMDs and al-Qaeda proved more difficult and ultimately false (Woodward 2006). Additionally, the invasion of Iraq was seen as a preventive measure to prevent a possible attack in the future, which stands in stark contrast to the imminent threat that Afghanistan presented (Levy 2008), suggesting that states are better able to overcome an immediate threat rather than one on the distant horizon. Future aggregate analyses must be careful to identify the difference between preemptive action and preventive action when identifying the context of the counterterrorist environment.
Another point concerns the nature of the prisoners’ dilemma situation when applying it to the complex and amorphous circumstances of terrorism. The theory and research design utilized here modeled states as relatively homogenous save for a few important variables concerning democracy, alliance status, and military capabilities. Aggregate analysis necessitates such an approach. However, analyzing the threat of a terrorist attack is fraught with a set of unique difficulties regarding the motivations of the terrorist organization, their capabilities, structure, support network, location, as well as identifying the very definition of terrorism itself. These characteristics are much easier to estimate in other states and suggest the possibility that a country’s evaluation of a threat stemming from a terrorist attack is inherently different from a threat that originates from a state. This intricate phenomenon is also complicated by the multiple strategies that states use to combat different terrorist networks, indicating that the multifaceted problem of international terrorism may require action on a case-by-case basis (Miller 2007; Bleich 2009; Meyer 2009).
Finally, the game outlined here is modeled as a one-shot play. This is a clear simplification of a complex decision process and neglects a number of possible alternative explanations concerning the possibility of repeated interactions, reciprocity, and the shadow of the future (Axelrod and Keohane 1985). Subsequent analyses could utilize more complex models of the prisoners’ dilemma by including a time dimension. Additionally, Bayesian methods could be applied to determine if states postpone counterterrorist action in anticipation of the decisions of more powerful countries. Nevertheless, the results show that generalizations are capable of being identified and point to the utility of aggregate analyses and the application of conflict theories to certain aspects of counterterrorist behavior.
About the author: Kyle T Kattelman is a Mizzou Advantage Postdoctoral Fellow at the University of Missouri. His research focuses on terrorism, counterterrorism, and US foreign policy. He currently resides with his wife in Columbia, Missouri.
 Pape (2003) notes that while individual terrorists are suicidal, terrorist organizations utilize suicide terror campaigns in a rational and strategic fashion.
 Arce and Sandler (2005) and Enders and Sandler (2006) model state strategies as a choice between “preemption,” signifying military action, and “deterrence,” signifying the strengthening of defenses without military commitment.
 A negative binomial was also employed, though convergence was difficult to achieve due to the small sample size.
 This estimate does not include the military operations in Iraq and Afghanistan. Mueller and Stewart (2012) also note that the odds of an American dying from a terrorist attack in a given year are approximately one in 3.5 million.
Amble, John Curtis. 2012. “Combating Terrorism in the New Media Environment.” Studies in Conflict and Terrorism 35: 339–353.
Arce M., Daniel G., and Todd Sandler. 2005. “Counterterrorism: A Game-Theoretic Analysis.” Journal of Conflict Resolution 49(2): 183–200.
Axelrod, Rober, and Robert O. Keohane. 1985. “Achieving Cooperation Under Anarchy: Strategies and Institutions.” World Politics 38(1): 226–254.
Bapat, Navin A. 2006. “State Bargaining with Transnational Terrorist Groups.” International Studies Quarterly 50(1): 213–229.
Benmelech, Efraim, Berrebi, Claude, Klor, Esteban F. 2012. “Economic Conditions and the Quality of Suicide Terrorism.” Journal of Politics 74(1): 113–128.
Bleich, Erik. 2009. “State Responses to ‘Muslim’ Violence: A Comparison of Six West European Countries.” Journal of Ethnic and Migration Studies 35(3): 361–379.
Boyle, Michael J. 2010. “Do Counterterrorism and Counterinsurgency Go Together?” International Affairs 82(2): 333–353.
Byman, Daniel. 2005. Deadly Connections: States That Sponsor Terrorism. Cambridge: Cambridge University Press.
Country Reports on Terrorism 2004. 2005. Washington, DC: US Department of State.
Crenshaw, Martha. 1981. “The Causes of Terrorism.” Comparative Politics 13(4): 379–399.
Crenshaw, Martha. 2013.Mapping Militant Organizations. Stanford University http://www.stanford.edu/group/mappingmilitants/cgi-bin/ (December 6, 2013).
Enders, Walter, and Todd Sandler. 1993. “The Effectiveness of Anti-Terrorism Policies: A Vector-Autoregression Intervention Analysis.” American Political Science Review 87(4): 829–44.
Enders, Walter, and Todd Sandler. 2006. The Political Economy of Terrorism. New York: Cambridge University Press.
Ex: The White House. 1999. A National Security Strategy for a New Century. Washington: The White House.
Ex: The White House. 2002 The National Security Strategy of the United States of America. Washington: The White House.
Fearon, James D. 2002. “Selection Effects and Deterrence.” International Interactions 28(1): 5–29.
Gaddis, John Lewis. 2004. Surprise, Security, and the American Experience. Cambridge: Harvard University Press.
Ginges, Jeremy. 1997. Deterring the Terrorist: A Psychological Evaluation of Different Strategies for Deterring Terrorism." Terrorism and Political Violence 9(1): 170–185.
Gordon, Philip H. and Jeremy Shapiro. 2004. Allies at War: America, Europe and the Crisis Over Iraq. New York: McGraw-Hill.
Hoffman, Bruce. 1995. “‘Holy Terror’: The Implications of Terrorism Motivated by a Religious Imperative.” Studies in Conflict & Terrorism 18(4): 271–284.
Hoffman, Bruce. 2006. Inside Terrorism. New York: Columbia University Press.
Hunter, Shireen T. 1998. The Future of Islam: Clash of Civilizations or Peaceful Coexistence? London: Praeger.
Hunter, Shireen T. 2002. Islam, Europe’s Second Religion: The New Social, Cultural and Political Landscape. London: Praeger.
IDEA: International Institute for Democracy and Electoral Assistance. http://www.idea.int/vt/viewdata.cfm. (December 10, 2013).
Israeli, Ralph. 2008. The Islamic Challenge in Europe. New Brunswick NJ: Transaction Publishers.
Kahera, Akel Ismail. 2002. “Urban Enclaves, Muslim Identity and the Urban Mosque in America.” Journal of Muslim Minority Affairs 22(2): 369–380.
Keohane, Robert O. 1993. Institutional Theory and the Realist Challenge After the Cold War. In Neorealism and Neoliberalism: The Contemporary Debate, edited by David A. Baldwin. New York: Columbia University Press.
Kilcullen, David J. 2005. “Countering Global Insurgency.” Journal of Strategic Studies 28(4): 597–617.
Kurth, James. 2002. “Domestic Security and Muslim Immigrants.” Journal of the Historical Society 2(3/4): 409–419.
Kydd, Andrew H., and Barbara F. Walter. 2006. “The Strategies of Terrorism.” International Security 31(1): 49–80.
Lebow, Richard Ned, and Janice Gross Stein. 1989. “Rational Deterrence Theory: I Think, Therefore I Deter” World Politics 41(2): 208 - 24.
Lebow, Richard Ned and Janice Gross Stein. 1990. “Deterrence: The Elusive Dependent Variable” World Politics, 42(3): 336 - 369.
Levy, Jack S. 2008. “Preventive War and Democratic Politics.” International Studies Quarterly 52(1): 1–24.
Li, Quan. 2005. “Does Democracy Promote or Reduce Transnational Terrorist Incidents?” Journal of Conflict Resolution 49(2): 278–297.
Lieber, Keir A., and Gerard Alexander. 2005. “Waiting for Balancing: Why the World is not Pushing Back.” International Security 30(1): 109–139.
Lipson, Charles. 1993. International Cooperation in Economic and Security Affairs. In Neorealism and Neoliberalism: The Contemporary Debate, edited by David A. Baldwin. New York: Columbia University Press.
Marshall, Monty G., and Keith Jaggers. 2000. Polity IV Project: Political Regime Characteristics and Transitions, 1800–2010 dataset users manual.
McCauley, Clark, and Sarah Scheckter. 2008. “What’s Special about U.S. Muslims? The War on Terrorism as Seen by Muslims in the United States, Morocco, Egypt, Pakistan, and Indonesia.” Studies in Conflict and Terrorism 31(11): 973–980.
Meyer, Christoph O. 2009. “International Terrorism as a Force of Homogenization? A Constructivist Approach to Understanding Cross-National Threat Perceptions and Responses.” Cambridge Review of International Affairs 22(4): 647–666.
Miller, Gregory D. 2007. “Confronting Terrorisms: Group Motivation and Successful State Policies.” Terrorism and Political Violence 19: 331–350.
Mueller, J. E. 1970. “Presidential Popularity from Truman to Johnson.” American Political Science Review 64(1): 18–34.
Mueller, John. 2006. “Is There Still A Terrorist Threat?: The Myth of the Omnipresent Enemy.” Foreign Affairs 85(5): 2–8.
Mueller, John, and Mark G. Stewart. 2012. “The Terrorism Delusion.” International Security 37(1): 81–110.
Nagl, John A. 2002. Learning to Eat Soup with a Knife: Counterinsurgency Lessons from Malaya and Vietnam. London: Praeger.
National Consortium for the Study of Terrorism and Responses to Terrorism (START). (2011). Global Terrorism Database. http://www.start.umd.edu/gtd/search/. (December 10, 2013).
Newman, Edward. 2006. “Exploring the ‘Root Causes’ of Terrorism.” Studies in Conflict & Terrorism 29(8): 749–772.
Norris, Pippa, and Ronald F. Inglehart. 2012. “Muslim Integration into Western Cultures: Between Origins and Destinations.” Political Studies 60: 228–251.
Parker, S. L. 1995. “Toward Understanding of ‘Rally’ Effects: Public Opinion in the Persian Gulf War.” Public Opinion Quarterly 59(4): 526–546.
Pape, Robert A. 2003. “The Strategic Logic of Suicide Terrorism.” American Political Science Review 97(3): 343–361.
Patterns of Global Terrorism 2001. 2002. Washington, DC: US Department of State.
Pluemper, Thomas, and Eric Neumayer. 2010. “The Friend of My Enemy is My Enemy: International Alliances and International Terrorism” European Journal of Political Research 49(1): 75–96.
Quackenbush, Stephen L. 2004. “The Rationality of Rational Choice Theory.” International Interactions 30(2): 87–107.
Quackenbush, Stephen L. 2011. “Deterrence Theory: Where Do We Stand?” Review of International Studies 37(2): 741–762.
Quackenbush, Stephen L., and A. Cooper Drury. 2011. “National Missile Defense and (Dis)satisfaction.” Journal of Peace Research 48(4): 469–480.
Record, Jeffrey. 2007. Beating Goliath: Why Insurgencies Win. Washington, D.C.: Potomac Books, Inc.
Rees, Wyn, and Richard J. Aldrich. 2005. “Contending Cultures of Counterterrorism: Transatlantic Divergence or Convergence?” International Affairs 81(5): 905–923.
Rhodes, Edward. 1989. Power and MADness: The Logic of Nuclear Coercion (New York: Columbia University Press.
Rosendorff, B. Peter, and Todd Sandler. 2004. “Too Much of a Good Thing? The Proactive Response Dilemma.” Journal of Conflict Resolution 48(5): 657–671.
Sandler, Todd. 2003. “Collective Action and Transnational Terrorism.” The World Economy 26(6): 779–802.
Tellis, Ashley J. 1996. “Terminating Intervention: Understanding Exit Strategy and U.S. Involvement in Intrastate Conflicts.” Studies in Conflict and Terrorism 19: 117–151.
Trager, Robert F., and Dessislava P. Zagorcheva. 2006. “Deterring Terrorism: It Can Be Done.” International Security 30(3): 87–123.
Unknown author. 2005. “NATO and the Scourge of Terrorism.” February 18. http://www.nato.int/terrorism/five.htm. (December 10, 2013).
Valentino, Benjamin A., Huth, Paul K., and Sarah E. Croco. 2010. “Bear Any Burden? How Democracies Minimize the Costs of War.” Journal of Politics 72(2): 528–544.
Voeten, Erik. 2004. “Resisting the Lonely Superpower: Responses of States in the United Nations to U.S. Dominance.” The Journal of Politics 66(3): 729–754.
Wallerstein, Immanuel. 2013. “France’s Aggressive Foreign Policy.” Aljazeera America December 2. http://america.aljazeera.com/opinions/2013/12/france-foreign-policymilitaryintervention.html (December 9, 2013).
Wilkinson, Paul. 2011. Terrorism versus Democracy: The Liberal State Response, 3rd ed. New York: Routledge.
Woodward, Bob. 2002. Bush at War. New York: Simon and Schuster.
Woodward, Bob. 2004. Plan of Attack: The Definitive Account of the Decision to Invade Iraq. New York: Simon and Schuster.
Woodward, Bob. 2006. State of Denial: Bush at War, Part III. New York: Simon and Schuster.
Wooldridge, Jeffrey M. 2009. Introductory Econometrics: A Modern Approach, 4th ed. Ohio: Cengage Learning.
World Islamic Front. 1998. “Jihad Against Jews and Crusaders.” Statement. http://www.fas.org/irp/world/para/docs/980223-fatwa.htm (December 9, 2013).
Zagare, Frank C., and D. Marc Kilgour. 2000. Perfect Deterrence. United Kingdom: Cambridge University Press.