The Selectorate Theory: Power of Predicting Terrorism

________________________________________________________________________ The Selectorate theory, developed by Bueno de Mesquita, Smith, Siverson, and Morrow in The Logic of Political Survival (2003), recognizes that political institutions have a variety of consequences for a country’s economic, political, and social structure. This paper extends this logic to investigate the explanatory power of the Selectorate theory in understanding terrorism. Specifically, I test the hypothesis that the Selectorate model can be used to explain levels of international terrorism produced by countries. The independent variables tested are states’ winning coalition size (W), selectorate size (S), and ratio of W/S, as defined in The Logic of Political Survival database. The dependent variables are states’ levels of international state-sponsored and non-state sponsored terrorism as in the ITERATE database (Mickolus et al. 2003.) I find that from 1968 1989, with control variables, winning coalition size has a significant positive impact on levels of terrorism by non-state actors. Without controls (1990-2000), there are significant negative correlations between winning coalition size and W/S and statesponsored attacks, and a significant negative correlation between selectorate size and attacks by non-state actors. These findings suggest that the Selectorate theory is perhaps not the best framework within which to explain international terrorism as a function of a country’s political institutions, but that it may be useful in identifying trends in the relationship between political institutions and terrorism. ________________________________________________________________________

Robin Morse

1

Table of Contents

I.

Introduction.........................................................................................................

II.

Roots of Terrorism ................................................................................................ 5

3

A.

Oppression and Low Civil Liberties as Causes of Terrorism.....................

6

B.

Low Public Goods as a Cause of Terrorism..............................................

12

C.

Causes of State-Sponsored Terrorism.......................................................

14

D.

Perceptions as a Cause of Terrorism.........................................................

16

E.

Other Considerations................................................................................

16

III.

Causal Model ......................................................................................................

17

IV.

Testable Hypotheses ............................................................................................

20

V.

Description of Data..............................................................................................

21

VI.

Empirical Method................................................................................................

25

VII.

Results.................................................................................................................

27

VIII.

Analysis and Conclusions ....................................................................................

37

IX.

References...........................................................................................................

41

2

I.

Introduction According to Bueno de Mesquita et al. (The Logic of Political Survival 2003), the

main objective of leaders is to stay in power. They always face challengers and stay in power by raising government revenue through taxation and spending it on public goods, which benefit everyone in society, and private goods, which benefit members of the winning coalition. To stay in office, leaders must prevent members of their winning coalition from defecting. The winning coalition is composed of members whose support endows the leadership with political power; it is a subset of the selectorate. The remaining members of the selectorate have the possibility of becoming members of a future winning coalition. People are chosen to become part of a coalition in part based on their affinity. If they back a challenger, then they cannot be sure that they will be part of his coalition. Risk of being replaced increases as the selectorate increases and the coalition decreases. Those who are in the winning coalition beyond the transition period can be confident that because of their high affinity for the leader, they will remain in his long-term coalition. Bueno de Mesquita et al. discuss how institutional arrangements influence policy and political survival. The size of the winning coalition determines whether policies have a public or private focus. If the coalition is small, then leaders focus on providing their supporters with private goods. If the coalition is large, then leaders are unable to reward their supporters with high levels of private goods and must provide public goods to survive. Thus, when W is small, members of the winning coalition are much better off than those outside the coalition. But as the coalition’s size increases and leaders switch to

3

policies with a public focus, the differences between the welfare of these groups diminish. When W is small, its members fear leadership turnover because this could cut off their access to private goods, and the cost of losing this privilege is high. If a challenger comes to power, he may not maintain the coalition that brought him to power; this creates a risk of exclusion. The risk of exclusion increases as W decreases and S increases. When W is small and S is large, the loyalty norm is strong because both the cost and risk of exclusion are high. Bueno de Mesquita et al. (2003) assert that oppression is most common in small coalition systems. When the private benefits of office or coalition membership are large, people have more incentive to oppress others to ensure that they do not lose these private benefits. The authors suggest that countries with small winning coalitions and large selectorates have the greatest intensity and magnitude of oppression. These systems provide the strongest incentives to challenge the leader, incentives for the leader to do anything possible to retain power, the ability to recruit people to carry out oppression, and credibility because of long tenure for leaders. Strongest oppression is directed at members of the winning coalition because they represent the greatest threat to the leaders. The oppressors are usually members of the winning coalition or members of the selectorate seeking to be promoted into the winning coalition. In systems with a small winning coalition and small selectorate, other members of the selectorate are also willing to carry out the oppression. Democratic leaders do not use oppression because the benefits of holding leadership and the costs of losing it are not very great. Also, in democratic systems, it is difficult to recruit followers to carry out the oppression.

4

Civil liberties are considered one of the most important public goods that a government can provide. When citizens are secure against domestic tyranny, their welfare is enhanced. Bueno de Mesquita et al. (2003) describe civil liberties as a “core public good.” They use the Freedom house scale to assess civil liberties and their association with institutional arrangements. The authors argue that civil liberties are dependent on coalition size: the larger the coalition or the weaker the loyalty norm, the stronger we expect civil liberties to be.

II.

Roots of Terrorism The potential usefulness of the Selectorate model in explaining and predicting

levels of terrorism becomes evident through an examination of the causes of terrorism and their relations to institutional arrangements. Lai (2004) modifies the concept of opportunity and willingness developed by Most and Starr (1989) to apply the theory to terrorism. Most and Starr define opportunity as “possibilities that are available within any environment” while willingness “captures an actor’s motivation to choose a particular action”.

Thus, opportunity factors make terrorism more likely, while willingness

encompasses the circumstances that lead people to participate in terrorism. A number of factors that affect opportunity and willingness to engage in terrorism are examined below. Factors that might increase willingness to engage in terrorism include oppression, low civil rights, low public good provision, and grievances. Lack of constraints on leaders’ public policy may increase the opportunity for terrorism, as this freedom allows leaders to implement policies without regard for the welfare of citizens and to neglect national security.

5

Thus, institutional arrangements in a country affect levels of transnational and international terrorism in two distinct but related ways. Transnational terrorism depends on institutional arrangements because these arrangements determine whether the citizens in the country suffer from oppression, a lack of civil rights, and few public goods provided by the government; these deficiencies in welfare have been shown to be causes of transnational terrorism. It is expected that the selectorate theory should also be useful in predicting levels of international terrorism because certain institutional arrangements constrain leaders to implement good public policy that is supported by the citizens, while other institutional arrangements require only that a leader be able to buy off a small number of key supporters to stay in power.

A.

Oppression and Low Civil Liberties as Causes of Terrorism

Pillar (2004) suggests that two types of antecedent conditions produce terrorism. These are “the issues expressed directly by the terrorists and those who sympathize with their cause: political repression, a lack of self-determination, the depravity of their rulers” and “the living standards and socioeconomic prospects of populations that are, or may become, the breeding stock for terrorists.” Pillar qualifies this theory, however: “No one has produced a good algorithm for the many variables that, in combination, breed terrorists. In the nineteenth century, terrorism frequently emerged in direct response to repression, but the correlation between political grievances and terrorism in more recent times is less obvious. In fact, terrorism today appears more often in free than in unfree societies.”

6

Krueger and Laitin (2003) examine the economic motivations of terrorists. Much literature on terrorism focuses on the economic motivations of terrorists, and George W. Bush has claimed that fighting poverty will decrease terrorism. However, the data does not support poverty as a cause of terrorism. The amount of terrorism does not appear to depend on macroeconomic shifts. At the microeconomic level, studies have failed to find a connection between poverty, education, and participation in terrorism. Some poverty theorists claim that well-to-do citizens participate in terrorism out of public spiritedness for the low standards of living of other citizens—a “Robin Hood” model of terrorism participation. Terrorist organizations may choose those individuals best suited to carry out missions—those with education and economic means. The correlates of participation in terrorism at the national level have been studied, but little evidence of a relationship between economic factors and the incidence of terrorism has been established. Utilizing the State Department’s data on international terrorism and a new dataset on suicide attacks, Krueger and Laiten (2003) link the country of origin of the terrorists and the target country in their analysis. Controlling for political regime, the evidence does not suggest that terrorist origins are founded on economic conditions. Terrorism appears to originate in countries that suffer from political oppression—that is, countries with low scores on the Freedom House index of civil liberties, and the targets are economically successful countries. Krueger and Laitin do not propose a theory to explain the correlation between oppression and terrorism; they simply conclude that “the most salient patterns in the data on global terrorism that we presented suggest that, at the country level, the sources of international terrorism have more to do with repression than with poverty” Id. International terrorism appears to be caused by repression rather than

7

by poverty. Neither country GDP nor illiteracy can predict terrorist origins. Thus, because Krueger and Laitin (2003) demonstrate that oppression is a main cause of international terrorism, and the Selectorate model claims to predict institutions that will encourage oppression, the model should be able to account for countries’ levels of transnational terrorism. Why might oppression cause terrorism? One possible explanation is that citizens with strong ideology regarding other countries who are denied political participation in their own countries are drawn to terrorism to achieve the goals of their ideology. Another explanation is that citizens in oppressive regimes may have more grievances toward other countries because they may view democracies as the cause of their oppression. Thus, oppression in a country will increase citizens’ willingness to engage in terrorism; countries with high oppression will have higher levels of transnational terrorism than will countries with low oppression. Martin (2003) suggests that “Adapting Frantz Fanon’s analysis that oppressed people psychologically need to engage in revolutionary resistance against their oppressor, terrorists can use the international arena as a means to “liberate” themselves and to thereby obtain vengeful justice against an adversary.” International terrorism may therefore serve as a means for oppressed peoples to combat their oppressors. Crenshaw (1981) also suggests oppression as a cause of terrorism. She asserts that oppression creates weakness and restriction of choice, which encourage terrorism: “Resistance organizations who lack the means of mounting more extensive violence may then turn to terrorism because legitimate expression of dissent is denied.” When citizens are denied legitimate participation, they may turn to terrorism to achieve their political

8

goals. This argument could also apply to international terrorism because if citizens with strong foreign policy beliefs are denied participation in the politics of their nation, then they may use terrorism as an alternate way to promote their ideology regarding international affairs. This supports the hypothesis that terrorism levels would be highest in states with a small W and large S because people in these states are subjected to the most oppression. Krueger and Maleckova (2003), in their examination of education and poverty as causes of terrorism, also conclude that, in fact, oppression and low civil rights are better predictors of motivation for terrorism. Indeed, the authors note that once a country’s Freedom House score on civil liberties is taken into account, the state’s income level appears to be unrelated to the number of terrorists it produces. They note that since September 11, several prominent observers and academics have called for increased aid and educational assistance to fight terrorism. For Krueger and Maleckova, such claims are problematic given the paucity of evidence for this view. The authors conclude that terrorism is not directly or strongly connected to poverty or education; rather, it is a response to political conditions and long-standing feelings of frustration. Thus, the Selectorate model may be a useful predictor of terrorism insofar as it captures aspects of a state’s political atmosphere that contribute to such frustration and repression. Krueger and Maleckova (2003) explore the theoretical considerations involved in the occupational choice to participate in terrorist-type activities. They contend that the rational-choice model of participation in terrorism is in some ways useful: terrorism, for example, might be considered an extreme and violent form of political engagement, which is more common among educated people from privileged backgrounds who have

9

the interest, expertise, and commitment to participate. Terrorist organizations may prefer educated, committed individuals, and these individuals may be better able to carry out acts of international terrorism. They also examine determinants of hate crimes and utilize public opinion data from the West Bank and Gaza Strip on Palestinians’ attitudes toward violence and terrorism, statistical analysis of the determinants of participation in Lebanon, biographical evidence on Palestinian suicide bombers and the backgrounds of Israeli Jews involved in terrorist activities, and a new cross-country data set as to whether a country’s economic conditions are related to the likelihood that citizens from that country will become involved in international terrorism. Despite the fact that definitions of terrorism often include nation states as potential perpetrators, Krueger and Maleckova (2003) do not examine state actors because they believe that the process underlying state terrorism is very different than the process underlying substate terrorism and would require different data and analysis. I hypothesize that although terrorism by state sponsors and terrorism by non-state actors have different root causes, the Selectorate theory will be able to account for both forms of terrorism with political institutions. The hypothesis here is that states with a large W and large S will have the lowest levels of both forms of terrorism, while states with a small W and large S will have the highest levels of both forms. Martin (2003) suggests that relative deprivation theory can be used to understand terrorism. Relative deprivation theory holds that feelings of deprivation relative to other groups underlie a decision to engage in collective action. Martin states, “when a group’s rising expectations are met by sustained repression or second-class status, the group’s reaction may include political violence.” If individuals from countries

10

with a small winning coalition and large selectorate feel deprived relative to countries with a large winning coalition and large selectorate, then they may be more likely to engage in international terrorism. Crenshaw (1981) also recognizes this connection between lack of opportunity for political participation and terrorism. This suggests that regimes with a small W and large S would have high levels of terrorism as these regimes allow participation in politics only by a very small number of citizens. This theory might account for the correlation between oppression and national terrorism, and it could also account for the correlation between oppression and transnational terrorism if oppressed citizens turned to terrorism as a means of promoting their foreign policy ideology and objectives. Grievances are one of the most obvious and most important causes of terrorism. Crenshaw contends that “the first condition that can be considered a direct cause of terrorism is the existence of concrete grievances among an identifiable subgroup of a larger population . . .” (1981). Grievances on the part of a country’s citizens might lead to terrorism by non-state actors, while grievances on the part of a leader and his key supporters might fuel state-sponsored terrorism. Citizens in states with a small W and large S that suffer from oppression and low civil liberties might have strong grievances against democracies if they view these states as contributors to the misery of their country. For example, Bueno de Mesquita and Smith (Unpublished 2004) demonstrate that democracies that provide foreign aid to autocracies do not improve conditions in those countries but keep unpopular dictators in power for longer.

11

B.

Low Public Good Provision as a Cause of Terrorism

Public good provision also affects potential terrorists’ willingness to engage in terrorism. Berman and Laitin (2005) model the choice of terrorist methods by rebels, asking what a suicide attacker would have to believe to be deemed rational. They then devise a club goods model, which emphasizes voluntary religious organizations as providers of local public goods. The sacrifices demanded by these groups solve a freerider problem regarding public goods and make these groups well suited for organizing suicide attacks where defection by operatives endanger the entire organization. Thus, radical religious groups are well suited to dispatch suicide bombers. Perhaps in countries with few public goods (small W, large S), terrorist groups will be more powerful because they provide public goods that the government does not. For instance, as Berman and Laitin (2005) suggest, “public safety is an example of a pure public good which could be provided by government or by a club, perhaps as a religious obligation. Welfare services, schools, hospitals and mutual insurance are examples of excludable, partially rival activities commonly provided by religious communities.” Berman and Laitin (2005) analyze the choice of suicide attacks as a tactic and predict that such attacks will be used when targets are well protected and damage is great. The authors determine that state competence, economics, and topography have a greater effect on suicide attacks than theology. Their finding implies that weakening the benign activities of clubs reduces their ability to carry out attacks. This could be accomplished by targeting their social service provision networks or by strengthening competing networks through government or markets. This suggests that countries with a large W and

12

large S will have lower levels of terrorism because governments will provide more public goods and will be better able to compete with terrorist social service provision networks. Goldstone (2002) also emphasizes the role that lack of public goods plays in the origins of terrorism: “Where IIT [International Islamic Terrorism] groups are more successful than states in providing such basic services as protection, education, and welfare, they are naturally more influential over the local population.” Goldstone’s paper supports the hypothesis that people from states with a small W and large S will be most likely to engage in terrorism because these states provide the fewest public goods to compete with terrorist organizations. Martin (2003), noting that structural theory “has been used in many disciplines to identify social conditions (‘structures’) that affect group access to services, equal rights civil protections, freedom, or other quality-of-life measures,” suggests that it can be used to understand terrorism. Such “structures” include government policies and access to social institutions. Goldstone (1986) argues along similar lines: “weaknesses in state structures encourage the potential for revolution.” Applying this concept to international terrorism, one might predict that countries with poor government policies and low access to social institutions might produce more international terrorism than other countries. To use the selectorate theory to explain levels of transnational and international terrorism, one must evaluate not only how political institutions affect willingness for terrorism, but also how institutions affect opportunity for terrorism. Bueno de Mesquita et al. (2003) argue that in order to stay in power, states with a large W and large S must provide public goods—including good public policy—to their constituency. As national security is an important public good, countries with a large W and large S will generally

13

have better security measures and will be more motivated and competent to eradicate terrorists in their nation. Potential terrorists in countries with a small W and large S, on the other hand, will have greater opportunity to engage in terrorism because leaders in these countries are focused on private good provision and do not implement effective national security.

C.

Causes of State-Sponsored Terrorism

Because leaders in states with a large W and large S must provide good public policy to stay in office, these leaders and governments are not able to engage in international terrorism that would be detrimental to the nation as a whole. They must appeal to the median voter in order to stay in power, and as Colb (2001) points out: “the vast majority of any ethnic or social group is made up of people who abhor terrorism.” In states with a small W and large S, on the other hand, leaders only have to provide enough private goods to buy off their key supporters in order to have the opportunity to engage in international terrorism. Martin (2003) writes, “many governments have used terrorism as an instrument of foreign policy. These policies have been characterized by state sponsorship of terrorist movements and direct state involvement in terrorist incidents or campaigns.” Terrorism is thus an option in foreign policy. To use the Selectorate theory to understand statesponsored terrorism, one must examine the constraints that political institutions impose on leaders. State sponsorship of terrorist activities, it is argued, is poor foreign policy because it produces a variety of consequences detrimental to the country. As a means of

14

combating such policy, Pillar (2004) suggests that “[p]unishing terrorists through prosecution or retaliatory strikes, for example, might have some deterrent effect . . . The posture that the United States takes toward the political aspirations of groups it has officially branded as terrorist affects the intentions of those groups. The same could be said of state sponsors of terrorism.” The United States government uses the “State Sponsors of Terrorism” list as an economic counterterrorism tool. The Secretary of States places countries on this list if they have repeatedly provided support for acts of international terrorism; this triggers a variety of economic sanctions on that state, including “restrictions on US foreign assistance; a ban on defense exports and sales; certain controls over exports of dual use items; and miscellaneous financial and other restrictions” (US Department of State 2004). There are currently six states designated as State Sponsors: Cuba, Iran, Libya, North Korea, Sudan, and Syria. Griset and Mahan (2003) point out that “sanctions experienced by a country as a result of the US designations of state sponsors of terrorism are significant, and they have contributed to famine, economic stagnation, and other deprivations for the citizens of some sanctioned states.” If the leader of a state with a large winning coalition and a large selectorate engaged in a policy that had such detrimental effects on his citizens, he would be thrown out of office. It is only in states with a small winning coalition and a large selectorate that a great majority of citizens have no say in selecting a leader, and the leader is thus unconstrained by their opinions.

15

D.

Perceptions as a Cause of Terrorism

Martin (2003) suggests that perceptions of international terrorism influence an actor’s propensity to engage in it. “Western democracies,” he observes, “regularly abhor and denounce international terrorism, whereas many regimes and leaders elsewhere in the world have been either weak in their denunciation or have on occasion expressed their approval of international extremist violence.” Martin says that terrorism in unacceptable from the perspective of Western governments because they have adopted the norm of democratic justice, according to which terrorism is perceived as criminal behavior. Also, it must be noted that the West is often the target of terrorism. Martin suggests that terrorism is often acceptable from the perspective of governments in the developing world for several reasons: many anticolonial extremists became national leaders, terrorism has been used as a matter of practical choice during times of crisis, and many developing world insurgents have crafted an effective fusion of ideology and warfare. Western democracies are generally characterized by large winning coalitions and selectorates, while developing governments are often associated with small winning coalitions and large selectorates. The selectorate theory should be able to account for a relationship between political institutions and perceptions of international terrorism, and ultimately the relationship between political institutions and propensity to engage in terrorism. E.

Other Considerations

Krueger and Laitin (2004) note that despite measurement inaccuracies and biases in the annual Patterns of Global Terrorism report, the database allows analysts to crossclassify attacks according to the country where they occurred and the perpetrators’

16

country of origin. From 1997 – 2002, terrorists tended to come from nondemocratic countries, both rich and poor, and target rich, democratic countries. As states with a large W and a large S are associated with democracy, this observation supports my hypothesis that these states will be least likely to engage in terrorism. When determining levels of terrorism originating in various states, it will be extremely useful to categorize these states using the parameters of the selectorate theory rather than using the broad labels of “democracy” or “autocracy.” The definitions of these forms of government vary according to who is defining them and for what purpose. Also, there are certainly vast differences in governments within each of these categories. The selectorate theory allows us to use measurements of political institutions to quantify characteristics of governments and to place them on a continuum based on how many people have an institutional say in who can be leader and how many people’s support is necessary for the leader to retain power. The selectorate theory will offer much more specific information about what political institutions cause terrorism than will the general categorizations “democracy” and “autocracy.”

III. Causal Model As we have seen, it is generally argued that the causes of terrorism are mainly political. Krueger and Laitin (2004) hypothesize that terrorism is caused mainly by oppression, and Krueger and Maleckova (2003) suggest that terrorism is related to civil war and to a lack of civil rights. Additionally, Berman and Laitin (2005) model terrorist groups as voluntary religious organizations that provide local public goods. State-sponsored terrorism is supported by leaders and members of their winning

17

coalitions. Members of the selectorate hoping to gain access to the winning coalition might also support state-sponsored terrorism. Non-state actors who engage in terrorism, on the other hand, are more likely to be members of the disenfranchised or possibly members of the selectorate who have little chance of entry in the winning coalition. I hypothesize that transnational terrorism, terrorism carried out by non-state actors, is most likely in states with a small W and large S: states that are associated with oppression, lack of civil rights, and few public goods and would provide the strongest incentives for citizens to turn to terrorism to improve their lots in life. Conversely, states with a large W and large S will be least likely to engage in transnational terrorism because they are less likely to suffer from oppression and weak civil rights and most likely to benefit from generous public good provision. In such a state citizens have few incentives to engage in terrorism. States with a large W and large S will be least likely to engage in international, or state-sponsored, terrorism, because the leaders of these states must keep the support of a large winning coalition. Leader selection in states with a large W and large S is driven by the median voter theorem, which holds that the candidate whose policy position is closest to that of the median voter will win the election. It is extremely unlikely that the median voter in a state would support a drastic, violent policy such as terrorism; thus, leaders in states with a large W and large S cannot pursue this policy or they will lose office. Leaders in states with large winning coalitions must provide good public policy in order to stay in power. Engaging in terrorism can decrease the welfare of the state because the benefits are small, as policy is rarely achieved through terrorism, while the costs can be substantial, as states that support terrorism are often subjected to various economic

18

sanctions. I further hypothesize that international terrorism would only be possible in states with a small W and large S, because leaders in these states depend only on a small winning coalition to retain power. Ganor (1997) notes that “ . . . most countries which sponsor terrorism are not democratic, and their decision-making processes are limited to the consideration of a single ruler . . . the lone decision maker tests the extent to which his activity will ultimately promote his goals, as he himself has defined them . . .” Leaders of states with a small W and large S do not need to provide good public policy to stay in power, they need only provide enough private goods to their supporters to buy their support. Thus, leaders in states with a small W and large S do not need to provide public policy that is attractive to a majority of their citizens; they can engage in terrorism as long as this policy is supported by their winning coalition or as long as they can provide enough private goods to their winning coalition to buy their support. I predict that states with a small W and small S, typically monarchies or juntas, will have levels of international and transnational terrorism that are higher than states with a large W and large S, but lower than states with a small W and large S. Transnational terrorism is associated with oppression, low civil rights, and low public good provision—political conditions induced by certain institutional arrangements in a state. Citizens in states with a small W and small S suffer from these conditions to an extent that is greater than citizens in states with a large W and large S yet less than citizens in states with a small W and large S. Thus, I expect levels of transnational terrorism in these monarchies and juntas to be consistent with political conditions

19

induced by their institutional arrangements; transnational terrorism will be more prevalent than in democracies but not as common as in autocracies. States with a small W and small S will also have levels of international terrorism that fall between those of democracies and autocracies. Leaders in states with a small W and small S are more constrained than leaders in democracies; like autocracies, they depend on a small winning coalition. The ratio of W/S in these states, however, is large; because the loyalty norm is weak, it is more difficult for leaders to retain the support of their winning coalition. Leaders are less likely to pursue extreme, violent policies because they may have to spend even more on their winning coalitions to compensate for these policies that are detrimental to the state.

IV.

Testable Hypotheses 1. States with a small W and large S will be most likely to engage in transnational and international terrorism. 2. States with a large W and large S will be least likely to engage in transnational and international terrorism.

I predict that as W increases, international and transnational terrorism will decrease. As S increases, international and transnational terrorism will increase. As W/S increases, international and transnational terrorism will decrease. I control for GDP per capita, log of population, ethnolinguistic fractionalization, GINI index of inequality, illiteracy rate, percentage population in the largest religious group, and percentage of the

20

population Moslem, because these factors could contribute to a state’s level of transnational or international terrorism.

V.

Description of Data The database from The Logic of Political Survival (Bueno de Mesquita et al.

2003) was used to determine each country’s winning coalition size, selectorate size, and ratio of W/S for each year from 1968 to 2000. Bueno de Mesquita et al. (2003) construct their winning coalition size variable, W, as a composite index based on the variables regime type, competitiveness of executive recruitment, openness of executive recruitment, and competitiveness of participation from the Arthur S. Banks Cross National Time Series Data Archive 2001. When the regime type variable in the Banks database, REGTYPE, is not equal to codes 2 or 3 so that the regime type is not a military or military/civilian regime, W receives one point. Because military regimes have particularly small coalitions, they are not credited with an increment in coalition size. When competitiveness of executive recruitment, XRCOMP, is larger than or equal to code 2, then another point is assigned to W. An XRCOMP code of 1 means that the chief executive was selected by heredity or in rigged, unopposed elections, suggesting dependence on few people. Code values of 2 and 3 refer to dependency on more supporters, implying a larger winning coalition. An additional point is added to W if the executive is recruited in a more open setting than heredity (that is, if XROPEN is greater than 2). Executives who are recruited in an open political process are more likely to depend on a larger coalition. Another point is added to W if competitiveness of participation, PARCOMP, is coded as a 5, meaning that "there

21

are relatively stable and enduring political groups which regularly compete for political influence at the national level" (Polity II, p. 18). This variable is used to indicate a larger coalition based on the assumption that stable and enduring political groups would not persist unless they believed they had an opportunity to influence incumbent leaders and thus to become part of a winning coalition. The indicator of W is then divided by 4 to create a five-point scale for W taking the possible values 0, .25, .5, .75, and 1. The selectorate size variable, S, is measured as LEGSELEC, the selection procedure for the legislature, as in Banks, 2001. This variable is also normalized to take the possible values of 0, .25, .5, .75, and 1. The ITERATE database was used to determine each country’s level of international and transnational terrorism for each year from 1968 to 2000. ITERATE stands for “International Terrorism: Attributes of Terrorist Events.” It is compiled by Mickolus, Sandler, Murdock, and Flemming (2003), and contains data from 1968 to 2002. ITERATE is a widely used database that codes all international terrorist events as reported in open news sources. Lai (2004) notes that “[d]emocratic states, which generally have more open news sources are likely to more accurately report terrorist incidents than autocratic states, which generally maintain tight controls over the media. As such, a finding that democracies are likely to experience greater degrees of terrorism could potentially be driven by the nature of the data collection process.” Thus, there may be a systematic bias in the ITERATE database to underreport terrorism against autocratic states. While this is a potential source of error, it should not have a great effect on the findings of this study, as I measure levels of terrorism originating from various states. Previous studies have shown that most terrorist attacks are directed toward democracies,

22

so the ITERATE database should provide a broad sample of attacks to use to examine levels of terrorism originating from various countries. Krueger and Laitin’s (2003) discussion of the political biases present in the State Department’s Patterns of Global Terrorism report suggests a benefit to using the ITERATE database. Because the data was gathered by independent researchers from news sources rather than by a government agency, the ITERATE database will have fewer political biases and be more accurate than other information sources. Mickolus et al. (2003) define international terrorism as “ . . . carried out by individuals or groups controlled by a sovereign state.” They define transnational terrorism as “ . . . carried out by basically autonomous non-state actors.” Thus, in their database, international terrorism is equivalent to state-sponsored terrorism, and transnational terrorism is equivalent to terrorism by non-state actors. The following variables were used in my research: - 1-3: Date of incident - 7: Evidence of state sponsorship o 1: Some evidence that the incident received state support o 2: No apparent evidence of state support o 9: Unknown - 17: First nationality of terrorists in attack force - 21: Number of victims - 30: Total individuals wounded In this study, the perpetrators of an attack are considered state sponsors if the evidence of state sponsorship variable is coded as 1 (some evidence that the incident received state support) in the ITERATE data. The perpetrators are considered non-state actors if the evidence of state sponsorship variable is coded as 2 (no apparent evidence of state 23

sponsorship.) I constructed a third category for those attacks whose state sponsorship variable is coded as 9 (unknown) and calculated levels of terrorism for this third category separately. Because my theory predicts that levels of both state-sponsored terrorism and terrorism by non-state actors will increase as W decreases, the fact that it is not known whether some attacks are state-sponsored should not hinder the experiment. If the variables “number of victims” or “total individuals wounded” are coded as 998 (unknown, but some victims), or 999 (unknown, but individuals were injured) in the ITERATE database, then these variables are coded as 1 victim or 1 individual respectively. This study is not concerned with the absolute number of victims or people wounded from terrorism from various countries, but with the comparison between these values for states with different institutional arrangements. Thus, as long as these variables are coded consistently across countries with different political institutions, these should not be a major source of error. Thus the independent variables are W, S, and W/S, and the dependent variables are levels of transnational and international terrorism as measured by number of attacks, number of victims, and total individuals wounded per year. I control for GDP per capita, log of the population, ethnolinguistic fractionalization, GINI index of inequality, illiteracy rate, percentage population in the largest religious group, and percentage of the population Moslem, as these characteristics have been suggested as factors in a country’s level of terrorism. Twenty-three country codes were present in the ITERATE database but not in the Selectorate theory database. These included Puerto Rico, Guadeloupe, Martinique, Indeterminate Latin American nation, Isle of man, Northern Ireland, Scotland, Corsica,

24

Indeterminate European nation, Latvia, Zaire, Cabinda, Canary Islands, Spanish Sahara and Western Sahara, Kurdisan, Indeterminate Arabs and Palestine, Yemen, Abu Dhabi, United Arab Emirates, Hong Kong, Bangladesh, South Molucca, and New Caledonia. The omission of most of these countries from the data set should not have had a strong effect on results, as most of the countries are not associated with high levels of terrorism. Four of the countries, however, were deemed crucial to the experiment because of their higher levels of terrorism. These were: Northern Ireland, Spanish Sahara and Western Sahara, Kurdistan, and Indeterminate Arabs and Palestine. For the purpose of this experiment, Northern Ireland was grouped with Great Britain, Indeterminate Arabs and Palestine was grouped with Israel, Spanish Sahara and Western Sahara was grouped with Morocco, and Kurdistan was grouped with Turkey.

VI.

Empirical Method As noted above, the dependent variables are states’ levels of international and

transnational terrorism, and the independent variables are W, S, and W/S. I performed a multiple regression and estimated the coefficients for each independent variable, controlling for GDP per capita, log of the population, ethnolinguistic fractionalization, GINI index of inequality, illiteracy rate, percentage population in the largest religious group, and percentage of the population Moslem. The model is expressed in the following equations: Level of international terrorism = a + ß1 (W) + ß2 (GDP per capita) + ß3 (log pop) + ß4 (ethnolinguistic fractionalization) + ß5 (GINI) + ß6 (illiteracy) + ß7 (religion) + ß8 (Moslem)

25

a + ß1 (S) + ß2 (GDP per capita) + ß3 (log pop) + ß4 (ethnolinguistic fractionalization) + ß5 (GINI) + ß6 (illiteracy) + ß7 (religion) + ß8 (Moslem)

a + ß1 (W/S) + ß2 (GDP per capita) + ß3 (log pop) + ß4 (ethnolinguistic fractionalization) + ß5 (GINI) + ß6 (illiteracy) + ß7 (religion) + ß8 (Moslem)

Level of transnational terrorism = a + ß1 (W) + ß2 (GDP per capita) + ß3 (log pop) + ß4 (ethnolinguistic fractionalization) + ß5 (GINI) + ß6 (illiteracy) + ß7 (religion) + ß8 (Moslem)

a + ß1 (S) + ß2 (GDP per capita) + ß3 (log pop) + ß4 (ethnolinguistic fractionalization) + ß5 (GINI) + ß6 (illiteracy) + ß7 (religion) + ß8 (Moslem)

a + ß1 (W/S) + ß2 (GDP per capita) + ß3 (log pop) + ß4 (ethnolinguistic fractionalization) + ß5 (GINI) + ß6 (illiteracy) + ß7 (religion) + ß8 (Moslem)

My theory predicts that levels of international and transnational terrorism can be predicted by the same equation, as both forms of terrorism are due to a country’s institutional arrangements. Different aspects of the selectorate theory are responsible for each form of terrorism, as explained in the causal model. I expect countries with a small W and a large S to have the highest levels of international and transnational terrorism. I expect countries with a large W and a large S to have the lowest levels of international

26

and transnational terrorism.

VII.

Results: Without control variables, there is a significant relationship between state-

sponsored attacks and W, S, and W/S. These relationships are in the predicted direction. When control variables are included, these relationships are no longer significant. These results are displayed in Table 1 and Graphs 1-3, below. Table 1: State-Sponsored Attacks: 1968 - 2000 W W S With Without With Controls Controls Controls W -.0438 -.1402*** (.0762) (.0461) S .0247 (.0630) W/S rgdpch lpop ELF60 GINI Illiteracy RELIGION MOSLEM N= R2 =

3.62e-06 (.0000) -.0211 (.0189) -.0136 (.1028) -.0046* (.0027) -.0012 (.0017) .0008 (.0018) .0015** (.0007)

-

259 .0543

1301 .0071

-

S Without Controls -

W/S With Controls -

W/S Without Controls -

.0997** (.0420) -

-

-

-.0481 (.0753) 3.73e-06 (.0000) -.0212 (.0188) -.0132 (.1024) -.0046* (.0027) -.0012 (.0017) .0008 (.0018) .0015** (.0007)

-.1534*** (.0478) -

259 .0546

1244 .0082

2.43e-06 (.0000) -.0241 (.0191) .0014 (.1026) -.0042 (.0027) -.0013 (.0017) .0005 (.0018) .0016** (.0007)

-

259 .0536

1244 .0045

27

-

-

0

.05

.1

.15

.2

.25

Graph 1: As winning coalition size increases, state-sponsored attacks decrease.

0

.2

.4 .6 Winning Coalition size 95% CI

.8

1

Fitted values

-.05

0

.05

.1

.15

Graph 2: As selectorate size increases, state-sponsored attacks increase.

0

.2

.4 .6 Selectorate size 95% CI

Fitted values

28

.8

1

0

.05

.1

.15

.2

.25

Graph 3: As W/S increases, state-sponsored attacks decrease.

0

.2 .4 .6 .8 W/S: transformed to avoid division by zero: =W/(log((s+1)*10)/3) 95% CI

Fitted values

29

1

Table 2: Attacks by Non-State Actors: 1968 - 2000 W W S S With Without With Without Controls Controls Controls Controls W 2.9752** .8649* (1.3781) (.5110) S 1.3655 -.3195 (1.1462) (.4621) W/S rgdpch lpop ELF60 GINI Illiteracy RELIGION MOSLEM N= R2 =

.0001 (.0002) -.0403 (.3410) 1.9381 (1.8587) -.0029 (.0489) -.0361 (.0303) -.0340 (.0329) .0044 (.0126)

-

259 .0458

1301 .0022

-

.0001 (.0002) -.0308 (.3477) 1.6287 (1.8670) -.0155 (.0487) -.0326 (.0304) -.0238 (.0327) -.0011 (.0123)

-

259 .0335

1244 .0004

-

W/S With Controls -

W/S Without Controls -

-

-

2.7466** (1.3638) .0001 (.0002) -.0237 (.3408) 1.8126 (1.8549) -.0051 (.0489) -.0346 (.0303) -.0324 (.0329) .0032 (.0125)

1.0216* (.5261) -

259 .0436

1244 .0030

-

Without control variables, there is a significant relationship between W and attacks by non-state actors and W/S and attacks by non-state actors. However, these relationships are not in the predicted direction. When control variables are included, these relationships become more significant, as displayed in table 2, above.

30

Dummy year variables were added to control for a trend over time of increasing democratization. These regressions produced similar coefficients and standard errors, displayed below in Table 3. Thus, the unexpected results are not due to a trend over time of increasing democratization. Table 3: Attacks by Non-State Actors Controlling for Changes Over Time W: with controls and W/S: with controls and dummy year variables dummy year variables W W/S

3.1557** (1.3289) -

N= R2 =

259 .2280

2.9500** (1.3115) 259 .2261

After reviewing the history of international terrorism from 1968-2000, it was postulated that while the positive correlation between winning coalition size and W/S and attacks by non-state actors could be accurate for the earlier period of international terrorism, from 1968-1989, it could not account for the character of international terrorism during the 1990s. Another set of regressions was attempted including only data from the 1990s, but this could not be carried out with control variables. Although there is data on terrorist attacks, W, S, and W/S from 1968-2000, most of the control variables are missing data for the 1990s. Thus when previous regressions with control variables were run, observations from the 1990s were dropped, and results only reflected data from 1968-1989.

31

Regressions were run with control variables with data from 1968 - 1990. These produced results similar to those from 1968 - 2000. Regressions run without control variables with data from 1968 - 1990 also produced results similar to those from 1968 2000. These results are displayed in Tables 4 and 5, below. Table 4: State-Sponsored Attacks 1968 - 1989 W W S With Without With Controls Controls Controls W -.0426 -.1429** (.0774) (.0573) S .0257 (.0638) W/S rgdpch lpop ELF60 GINI Illiteracy RELIGION MOSLEM N= R2 =

3.65e-06 (.0000) -.0218 (.0193) -.0124 (.1049) -.0046* (.0028) -.0012 (.0017) .0008 (.0018) .0015** (.0007)

-

254 .0544

878 .0070

-

S Without Controls -

W/S With Controls -

W/S Without Controls -

.1198** (.0503) -

-

-

-.0471 (.0764) 3.77e-06 (.0000) -.0218 (.0193) -.0120 (.1046) -.0046* (.0027) -.0012 (.0017) .0008 (.0018) .0015** (.0007)

-.1552*** (.0594) -

254 .0546

836 .0081

2.45e-06 (.0000) -.0249 (.0196) .0033 (.1047) -.0042 (.0027) -.0013 (.0017) .0006 (.0018) .0016** (.0007)

-

254 .0538

836 .0067

32

-

-

Table 5: Attacks by Non-State Actors 1968 - 1989 W W S With Without With Controls Controls Controls W 2.8980** .6867 (1.3756) (.5405) S 1.3494 (1.1410) W/S rgdpch lpop ELF60 GINI Illiteracy RELIGION MOSLEM N= R2 =

.0001 (.0002) -.0012 (.3431) 2.1234 (1.8662) .0033 (.0489) -.0351 (.0303) -.0363 (.0328) .0042 (.0126)

-

254 .0444

878 .0018

-

S Without Controls -

W/S With Controls -

W/S Without Controls -

-.1943 (.4709) -

-

-

2.6660* (1.3607) .0001 (.0002) .0159 (.3428) 1.9945 (1.8619) .0012 (.0489) -.0335 (.0303) -.0348 (.0328) .0030 (.0125)

.7992 (.5549) -

254 .0421

836 .0025

.0001 (.0002) .0084 (.3499) 1.8126 (1.8735) -.0085 (.0487) -.0316 (.0304) -.0268 (.0325) -.0009 (.0123)

-

254 .0326

836 .0002

-

-

The regressions reveal that from 1968-1989, controlling for GDP per capita, log of the population, ethnolinguistic fractionalization, GINI index of inequality, illiteracy rate, percentage population in the largest religious group, and percentage of the population Moslem, there is a significant positive correlation between winning coalition size and attacks by non-state actors, and a significant positive correlation between W/S and attacks by non-state actors. Without control variables, there are significant relationships between W, S, and W/S and state-sponsored attacks in the predicted direction. From 1990-2000, using the database from The Logic of Political Survival, control 33

variables cannot be taken into consideration. Without control variables there is a significant negative correlation between W and attacks and W/S and attacks. Regressions with non-state actors and no control variables are not significant. The results from these regressions are displayed in Tables 6 and 7, below. Table 6: State-Sponsored Attacks 1990 - 2000 W W S With Without With Controls Controls Controls W (dropped) -.1310* (.0770) S (dropped)

S Without Controls -

W/S With Controls -

W/S Without Controls -

-

-

(dropped)

W/S

-

-

-

.0517 (.0076) -

rgdpch

0

-

0

-

0

-.1460* (.0801) -

lpop

(dropped)

-

(dropped)

-

(dropped)

-

ELF60

(dropped)

-

(dropped)

-

(dropped)

-

GINI

0

-

0

-

0

-

Illiteracy

0

-

0

-

0

-

RELIGION

(dropped)

-

(dropped)

-

(dropped)

-

MOSLEM

0

-

0

-

0

-

N= R2 =

5 -

423 .0068

5 -

408 .0011

5 -

408 .0081

34

Table 7: Attacks by Non-State Actors 1990 - 2000 W W S S With Without With Without Controls Controls Controls Controls W (dropped) 1.0660 (1.1067) S (dropped) -1.1491 (1.1063) W/S rgdpch

-

lpop

.0047 (-) (dropped)

ELF60 GINI

W/S With Controls -

W/S Without Controls -

-

-

(dropped)

1.3453 (1.1450) -

-

-

.0047 (-) (dropped

-

.0047 (-) (dropped)

(dropped)

-

(dropped)

-

(dropped)

-

-.1762 (-) .3463 (-) (dropped)

-

-.1762 (-) .3463 (-) (dropped)

-

-.1762 (-) .3463 (-) (dropped)

-

MOSLEM

-1.0000 (-)

-

-1.0000 (-)

-

-1.0000 (-)

-

N= R2 =

5 1.0000

423 .0022

5 1.0000

408 .0027

5 1.0000

408 .0034

Illiteracy RELIGION

-

35

-

-

-

Table 8: Total Attacks W With Controls W 5.3928*** (1.8685) S W/S rgdpch lpop ELF60 GINI Illiteracy RELIGION MOSLEM N= R2 =

W Without Controls -.6297 (.5812) -

-

-

-.0001 (.0003) .1094 (.4624) 2.8707 (2.5203) .0380 (.0663) -.1201*** (.0411) .0033 (.0447) .0430** (.0171)

-

259 .0699

1301 .0009

-

S With Controls -

S Without Controls -

W/S With Controls -

W/S Without Controls -

-1.0739 (1.5683) -

-.5327 (.5240) -

-

-

.0000 (.0003) .3471 (.4758) 1.4768 (2.5545) .0008 (.0667) -.1101*** (.0416) .0328 (.0447) .0270 (.0168)

-

5.7277*** (1.8423) -.0001 (.0003) .1182 (.4603) 2.7809 (2.5056) .0385 (.0661) -.1182*** (.0409) .0027 (.0445) .0425** (.0169)

.8422 (.5972) -

259 .0407

1244 .0008

259 .0746

1244 .0016

-

-

Attacks by state sponsors, non-state actors, and unknown entities were then pooled together to obtain the largest possible number of observations (Table 8, above). Without control variables, results were not statistically significant. With control variables, there were significant positive correlations between W and attacks and between W/S and attacks. Thus, the regression was probably dominated by the non-state actors from countries with large winning coalitions during 1968 - 1990.

36

VIII. Analysis and Conclusions A survey of international terrorism from 1968 - 1989 suggests that the correlation between W and W/S and attacks by non-state actors during this period is accurate. Hoffman (1998) asserts that in the late 1960s and early 1970s, “outside the Middle East, a combination of societal malaise and youthful idealism, rebelliousness and antimilitarism/anti-imperialism was rapidly transforming the collective political consciousness among the more affluent counties of Western Europe and North America.” Hoffman suggests that the unprecedented economic prosperity of this period permitted self-criticism that resulted in revulsion against the socialist-economic inequities of the capitalist state. Four radical movements resulted from this era: Germany’s Red Army Faction (RAF), France’s Action Directe (AD), Italy’s Red Brigades (RB), and the Belgian Communist Combatant Cells (CCC). In 1985, the RAF joined with AD in hopes of forming an umbrella group that would include the other two radical movements and represent the “anti-imperialist front of Western European guerillas” (Hoffman 1998). By the late 1980s/early 1990s, however, all four of these movements had fizzled out: terrorism by non-state actors from 1968 -1989 thus appears to have been dominated by individuals from democratic states. Corrado and Evans (1988) also examine ideological terrorism in Europe from the 1960s through the 1980s. They suggest that “ideological terrorists in Europe reject the economic and social structure of industrial capitalism; they want a new order.” These movements were thus associated with capitalist, democratic states. Corrado and Evans conclude that by 1988, left-wing terrorism was diminishing as “the pluralism of western democracies opened the door to peaceful participation in the political system and offered

37

opportunities for change.” The relationship between political institutions and state-sponsored terrorist attacks may not have been significant during the period 1968 - 1990 because attacks by authoritarian states with a small W and large S were balanced with attacks by democratic states with a large W and large S. Richardson (2004) asserts that “even impeccably liberal democracies might engage in state-sponsored international terrorism.” He cites US support for Chilean anti-Allende forces in the 1970s, the Nicaraguan contras in the 1980s, and for anti-Castro forces throughout that period as examples of terrorism as an instrument of foreign policy. Martin (2003) notes that “it is certainly true that democracies are less likely to engage in this type of behavior [state-sponsored terrorism] than are aggressively authoritarian states. However, as suggested by the cases of the Phoenix Program and French intelligence operations, democracies have been known to resort to terrorist methods . . .” Because both autocracies and democracies have engaged in state-sponsored terrorism, there is not a significant relationship between political institutions and statesponsored terrorism. The coding of the Israel-Palestine conflict was examined to determine its effect on the model. As mentioned before, Israel and Palestine were grouped together because of the format of the Selectorate data. Hoffman (1998) asserts that “between 1968 and 1980, Palestinian terrorist groups were indisputably the world’s most active, accounting for more international terrorist incidents than any other movement.” The ITERATE database records 20 attacks by non-state Israeli/Palestinian actors during this period. From 1968 2000, there are 7 state-sponsored attacks and 26 attacks by non-state actors. This includes

38

6 state-sponsored attacks and 23 attacks by non-state actors from 1968 - 1989, and 1 state-sponsored attack and 3 attacks by non-state actors from 1990 - 2000. The coding of the Israel-Palestine conflict thus contributes to the correlation between W and W/S and attacks by non-state actors from 1968 - 2000. Israel/Palestine has a large winning coalition, a large selectorate, and high levels of terrorism by non-state actors. It thus strengthens the relationship between W and W/S and attacks by non-state actors. Religion may play a decisive role in the correlation between political institutions and international terrorism during the 1990s. Hoffman (1998) remarks, “ . . . during the 1990s the growth in the number of religious terrorist groups as a proportion of all active international terrorist organizations has not only continued but increased appreciably.” Hoffman suggests that religion has become a more common motivation for terrorism since the end of the Cold War, as the collapse of the Soviet Union discredited Communist ideology but the promised benefits from the democratic state failed to materialize in many countries around the world. He highlights the violence of religious terrorism by citing the fact that the most serious terrorist acts of the 1990s all had a significant religious dimension. Thus the trend toward religiously motivated terrorism during the 1990s likely contributed to strengthening the relationship between political institutions and international terrorism. The separation of church and state in democracies would make state-sponsored international terrorism by countries with a large W and large S extremely unlikely. Some autocracies, however, are governed by religion and might engage in statesponsored religious terrorism. For example, the US considers the Islamic Republic of Iran to be the most active state-sponsor of terrorism.

39

Religion is also expected to motivate more international terrorism by non-state actors in autocracies than in democracies. Although religiously motivated terrorism does exist in democracies, it seems to be mainly domestic. In autocracies, on the other hand, religiously motivated terrorism often has international targets, such as Western democracies. No concrete conclusions about the relationship between political institutions and international terrorism during the 1990s can be drawn because of the lack of control variable values during this period. Without control variables, however, there is a significant negative correlation between W and W/S and state-sponsored attacks. Future research would involve determining control variable values for this period and including them in regressions with political institutions and terrorism levels. It is predicted that the original hypothesis will be true for the 1990s; that is, there is a negative correlation between W and W/S and state-sponsored attacks and attacks by non-state actors, and a positive correlation between S and state-sponsored attacks and attacks by non-state actors. While the selectorate theory is thus unable to adequately explain international terrorism as a function of a country’s political institutions, it may be useful in identifying trends in the relationship between political institutions and terrorism.

40

IX.

References

Berman, Eli and Laitin, David, 2005. “Hard Targets: Theory and Evidence on Suicide Attacks.” NBER Working Papers 11740, National Bureau of Economic Research, Inc. Bueno de Mesquita and Smith. “Foreign Aid and Policy Concessions.” Unpublished, 2004. Bueno de Mesquita, Smith, Siverson, Morrow, The Logic of Political Survival. Cambridge, MA: The MIT Press, 2003. Crenshaw, Martha. “The Causes of Terrorism.” From Comparative Politics 13, no. 4 (July 1981) pp. 379-99. Colb, Sherry F. (10-10-01). “The New Face of Racial Profiling: How Terrorism Affects the Debate.” Find Law’s Legal Commentary. Corrado, Raymond and Rebecca Evans. (1988). “Ethnic and Ideological Terrorism in Western Europe.” In Michael Stohl (ed.), The Politics of Terrorism. New York: Dekker. Ganor, Boaz, “Countering State-Sponsored Terrorism.” (Herzlia: ICT Papers, The International Policy Institute for Counter-Terrorism, The Interdisciplinary Center, 1997). Goldstone, Jack A. “States, Terrorists, and the Clash of Civilizations.” From Craig Cahoun, Paul Price, and Ashley Timmer, eds., Understanding September 11 (New York: The New Press, for the Social Science Research Council, 2002), pp. 139-58. Goldstone, Jack A. Revolutions: Theoretical, Comparative, and Historical Studies. San Diego, CA: Harcourt Brace Jovanovich, 1986. Griset, Pamala A. and Mahan, Sue. Terrorism in Perspective. Thousand Oaks, CA: Sage Publications, Inc., 2003. Hoffman, Bruce. Inside Terrorism. New York: Columbia University Press, 1998. Krueger, Alan B. and Laitin, David D., "Kto Kogo?: A Cross-Country Study of the Origins and Targets of Terrorism." Princeton University, November, 2003. Krueger, Alan B. and Laitin, David D., “’Misunderestimating’ Terrorism,” From Foreign Affairs, September/October 2004 Krueger, Alan B. and Maleckova, Jitka, "Education, Poverty and Terrorism: Is There a Causal Connection?" From Journal of Economic Perspectives, Volume 17, No. 4, Fall 2003. Lai, Brian, 2004. “Explaining Terrorism Using the Framework of Opportunity and Willingness: An Empirical Examination of International Terrorism.” Research Paper, Department of Political Science, University of Iowa, April 2004. Martin, Gus. Understanding Terrorism: Challenges, Perspectives, and Issues. Thousand Oaks, California: Sage Publications, Inc., 2003.

41

Mickolus, Sandler, Murdock, Flemming, “International Terrorism: Attributes of Terrorist Events 1968-2002.” Dunn Loring, VA: Vinyard Software. Most, Benjamin and Starr, Harvey, Inquiry, Logic, and International Politics. Columbia: South Carolina University Press, 1989. Pillar, Paul R. “The Dimensions of Terrorism and Counterterrorism.” From Howard and Sawyer, Terrorism and Counterterrorism: Understanding the New Security Environment. Guilford, Connecticut: McGraw-Hill/Dushkin, 2004. US Department of State. The Office of the Coordinator for Counterterrorism. 2004. Patterns of Global Terrorism 2003. Richardson, Louise. “Global Rebels: Terrorist Organizations as Trans-National Actors.” From Howard and Sawyer, Terrorism and Counterterrorism: Understanding the New Security Environment. Guilford, Connecticut: McGraw-Hill/Dushkin, 2004. White, Jonathan R. Terrorism and Homeland Security. Belmont, CA: Thomson Wadsworth, 2006.

42