ONE GOD FOR ALL:
FUNDAMENTALIST RELIGIOUS GROUPS
AND TERRORISM
Jóhanna Kristín Birnir
Associate Professor
Department of Government
Center For International Development and Conflict Management
University of Maryland
0145 Tydings Hall
College Park MD, 20742
Nil S. Satana
Assistant Professor
Bilkent University
Department of International Relations
Faculty of Economics, Administrative and Social Sciences
Bilkent 06800, Ankara, Turkey
Prepared for the annual meeting of the International Studies Association.
New Orleans, February 2010.
© Draft, please cite but don't quote.
1
INTRODUCTION
In May 2002 a car bomb outside the Karachi Sheraton Hotel in Pakistan killed 16
people. In June that same year three Islamic militants were sentenced to death for the
bombing. (Mickolus and Simmons, 2006). [In February 1994 a Zionist settler from Qiryat
Arba entered a mosque on the West Bank. He killed 29 and wounded 150 with an
automatic assault rifle, before being beaten to death.] (Ranstorp, 1996). In 1996 Erik
Rudolph, who has been linked to the Christian Identity Movement, set off a backpack
bomb at the Centennial Olympic Park in Atlanta, killing one and injuring 111 others. In
1997, he is alleged to have carried out a bombing at a family-planning clinic and a
nightspot frequented by lesbians (Eggen, 2005).
Such examples of terrorism seemingly motivated by religion are all too common.
Generally these instances involve purported representatives of religious minorities who
carry out an extreme act against some majority population. The apparent proliferation of
this „new‟ breed of terrorism (Newsweek, 1993) has spurred the development of a
substantial corresponding scholarly literature that aims to distinguish this „new terrorism‟
from older types and explain why the two differ (Hoffman, 1998; Kibble, 2002; Azzam,
2005; Wiktorowicz, 2006; Dobrot, 2007; Hegghammer, 2006; Schmid, 2004; Morgan,
2004; Rapoport, 2002). In contrast, some of the most prominent experts in the literature on
terrorism argue that the idea of a new type of religious terrorism is misleading. Broadly
speaking they maintain that a closer examination of the goals, structure and means of
terrorist organizations perpetrating this „new‟ terrorism reveal a pattern akin to the 'old'
terrorist organizations more active before the 1990s (Crenshaw, 2000, 2005, 2006;
Duyvesteyn, 2004). Furthermore, this literature critiques studies that find causes for
2
violence within a particular religion, and point out that violent acts are perpetrated by
adherents of most if not all religions worldwide (Hadden and Shupe, 1991; Crenshaw
2006).
Given the dearth of cross national data on religious motivations of terrorist incidents
to date, most of the literature on the topic of religion and terrorism is in the form of
qualitative analysis of one or a handful of cases with little cross national empirical testing.
This paper contributes to the literature by first considering theoretical arguments that take
into account the empirical observation that terrorist violence is common to many if not all
religions. More specifically drawing on Almond et al. (2003) and the Fundamentalist
Project, we posit that it is not content of religious doctrine per se that matters but rather
whether the religious setting includes religious fundamentalist elements, which aim to
make their particular worldview predominate. Furthermore, in line with Birnir and Şatana
(2007), we suggest that the strategic interests of fundamentalist majorities are at least as
likely to lead to situations of conflict as the strategic interests of fundamentalist minorities.
Second, using the Global Terrorism Database (GTD) and composite variables on religious
identities, we empirically test the idea that religious identities diverge in their impact on
terrorist violence. Moreover, using original quantitative data on fundamentalism we
empirically test the notion that fundamentalist groups, irrespective of creed, are associated
with higher incidents of terrorism than their non-fundamentalist counterparts. Our results
clearly confirm the idea that by and large religious doctrine is a poor predictor of terrorist
violence. Fundamentalism does, however, emerge as a strong and consistent predictor of
terrorist violence for majorities and minorities alike. Thus, our findings show no support
3
for the growing new terrorism literature, which emphasizes religious doctrine in general
and Islam in particular as the primary emerging cause in terrorism of the new era.
We would like to note at the outset, and later we discuss in greater detail, that this
paper is work in progress. Consequently, while we have good reason to believe our
findings are indicative of the true underlying relationships, the data collection and
empirical analysis are not complete (as explained in greater detail later) and all of the
subsequent analysis is preliminary.
THE LITERATURE
The general literature on religion and conflict is deeply concerned about the rise of
religious violence. Juergensmeyer (2003) argues that religious violence is on the rise
globally. He argues that a common culture of violence fosters religious violence in the
communities of Christianity, Judaism, Islam, Hinduism, Sikhism and Buddhism. The most
important aspect of this common culture of violence is that the rebels often perceive
themselves and their way of life under attack. When this kind of worldview combines with
other historical and circumstantial factors, Juergensmeyer finds that there is an “odd
attraction between religion and violence” (p. 14, 240). Huntington (1996) predicts a clash
of civilizations in the international system in the post-Cold War era. Religion is a part of
the definition of civilization along with language and culture. Along similar lines, Jelen
and Wilcox (2002: 16) suggest that “the similarities and differences between creeds in
pluralistic religious settings may provide some insight as to the sources and nature of
religious conflicts.”1 Others agree that the content of religion directly or indirectly
produces conflicts (Seul, 1999; Stark, 2001; Fox, 2000, 2002, 2004; Rashid, 2002).2 One
idea, for example, is that monotheism poses particular difficulties for peace (Stark 2001;
4
Jelen and Wilcox, 2002). It is important to note that the above authors do not argue that
politics involving religion are necessarily conflictual. For instance, Fox (2002) argues that
religion becomes a salient political issue only when the group harbors desires for autonomy
or independence.3 Seul (1999) and Fox (2002) also state that religion provides a
framework that helps people understand the world around them. Fox then goes on to argue
that when challenged, the reality provided by this framework must be defended.
Furthermore, he suggests that religious frameworks provide codes of conduct for adherents.
In some cases, these codes are even thought to require conflictual behavior by followers.
Both Seul and Fox agree that religious institutions may facilitate conflict and Fox adds that
religious authority grants the conflict legitimacy. Stark (2001: 367, 376) further
underscores the strength of world-shaping view of religion.
This concern about religion as the root cause of violence is echoed in the literature
that focuses more specifically on terrorism. Because terrorist groups and terrorism are
poorly defined and in controversial terms both in the literature and in common discourse
(Feldman and Peralta 2004, Walzer 2002, Stohl 1988) it is important to define our use of
the term before proceeding further. This paper follows Crenshaw‟s (1972) definition,
which states that: “(1) Terrorism is part of a revolutionary strategy - a method used by
insurgents to seize political power from an existing government. (2) Terrorism is
manifested in acts of socially and politically unacceptable violence. (3) There is a
consistent pattern of symbolic or representative selection of the victims or objects of acts of
terrorism. (4) The revolutionary movement deliberately intends these actions to create a
psychological effect on specific groups and thereby to change their political behavior and
attitudes." (p. 385). Returning to the concern about religion as the root cause of violence in
5
the literature on terrorism, Rapoport (1984: 673), for example, compared three religious
terrorist groups of Hindu, Muslim and Judaist faith, and found that „sacred terror‟ had
common characteristics even though these three religions were fundamentally different.
Furthermore, Rapoport (2002) argues that the period after 9/11 is the period of the religious
wave in four waves of terrorism, also including the „anarchist wave,‟ „anti-colonial wave,‟
and the „new left wave‟ Similarly, Hoffman (1998) argues that although leftist and ethno-
national terrorism have been very destructive, religion is the most important terrorist
motivator today. Morgan (2004) adds that old terrorism was politically motivated as
opposed to religiously motivated new terrorism.
Importantly, however, one strand of the terrorism literature also diverges from the
main-stream literature on religion and conflict (conventional and terrorist) in its emphasis
on Islamic terrorism. Hoffman (1993), for example, finds that Islam is a prominent cause
of terrorism and argues that „holy terror‟ will be the most dangerous type of violence in the
new millennium. Rapoport (2002: 424) posits that religious transcendence of state
boreders is particularly important for Islam since Sunnis are dispersed among many states
and religious beliefs are quite strong in all. Rapoport, for example, ties the September 11
attacks to this very fact. Kibble (2002) also argues that Islamic fundamentalism is what
lead to the 9/11 terrorist attacks and a more moderate form of Islam is needed to solve the
clash between the West and the Muslim world. Azzam (2005) finds a connection between
Wahhabi Islam and rise of militant Islam. He connects the very nature of Wahhabism to
religious fundamentalism. According to Azzam, Al Qaeda of Saudi origin is Wahhabist
thus anti-Christian and anti-Jewish in its very nature. Wiktorowicz (2006), on the other
hand, argues that the Salafi Movement (also referred to as Wahhabis) has factions that
6
differ greatly in motivation and in action. It is the jihadis that can be tied to violence and
not the purists and politicos. Thus, the author recommends that the two non-violent Salafi
groups should be allied with by the West against the jihadis, which is in line with Kibble‟s
recommendations. Dobrot (2007) posits three causes of Islamist terrorism: „the lack of
wealth-sharing in Islamic countries, resentment of Western exploitation of Islamic
countries, and a U.S. credibility gap within the Islamic community.‟ Thus he agrees with
Hoffman that Islam is indeed an underlying cause of global terrorism. Furthermore,
Hegghammer (2006) argues that Iraqi invasion of the U.S. increased global jihadist
terrorism because the invasion increased the ideological development of Islamist militants
as it increased frustration in the Muslim world towards the West.
Operationalizing the effects of divergent religions on terrorism for testing along the
lines of this strand of the literature we posit that it is possible that Sunni Muslims, for
example, are likely to consider the Muslim sect of of Shi‟iism less incompatible with their
world view than with a more divergent worldview such as Christian Catholicism.
Although there are significant differences between the practices and perceptions of the
Sunni and the Shi‟i Muslims, the fundamental building blocks within the religious family
of Islam (i.e., a single god, Allah and a single prophet, Mohammad) are common for both
sects. However, the building blocks of Christianity (i.e., the Trinity) significantly differ
from Islam. Similarly, according to this idea, the Catholic and Protestant branches of
Christianity are less likely to perceive their “world order” under attack from each other
than from other religious families that are fundamentally different from Christianity.4 If
this is true then we expect that in any given country:
7
H1. The probability of terrorist attack is higher between two groups whose religions are
from different families than between two groups who belong to the same family of
religions.
Countering this idea of new Islamic terrorism, Crenshaw (2005) argues that
examination of the goals, organization and means of new and old terrorism shows that new
terrorism is not new in terms of redefinition of these three criteria but in degrees of them.
Crenshaw (1981, 1995) argues that terrorism is caused by „psychological considerations
and internal bargaining, as well as by reasoned or strategic reactions to opportunities and
constraints, perceived in light of the organization‟s goals.‟ Crenshaw (2006) further
criticizes the new terrorism literature that underemphasizes political and social causes of
terrorist violence and blames terrorism on religion. Specifically she points out that
religious extremism is not monopolized by Islam but shared by other religions. Similarly,
Duyvesteyn (2004) counters the juxtaposition of new and old terrorism. Duyvesteyn agrees
that one of the main motivations of new terrorism is religion and fanaticism but points out
that older types of terrorism were not areligious, either. Furthermore, Duyvesteyn believes
that national and territorial characteristics still play an important role in contemporary
terrorism.
Furthermore, Pape (2005) argues that religious extremism is not a good variable to
predict terrorism because the terrorists are rational and they carry out violent attacks for
clear strategic objectives. Rather he posits that terrorists try to get modern democracies to
withdraw forces from the territories that the terrorists view as their homeland.
Consequently, Pape finds that the literature that links Islam to terrorism is fundamentally
wrong. In fact, Pape‟s data show that the world leader in suicide terrorist attacks is the Sri
8
Lankan Tamil Tigers, not an Islamist terrorist organization such as Hezbollah. Other cases
confirm this assesment. For example, since 1994 Sunni Muslim Kurds have carried out
tens of suicide attacks against Sunni Muslim Turks in Turkey. As the two groups belong to
the same family of religion explanations that center on religious doctrine as a cause for
Kurdish suicide attacks clearly fail to explain the case of Turkey.5
How then are we to reconcile these two observations that religious violence is
currently employed as a prominent motivator of terrorist violence but this method of
motivation is common to many religions? A possible solution presented by Almond et al.
(2003) and the five-volume Fundamentalism Project, is that it is not content of religion per
se that matters but rather whether the religious setting includes religious fundamentalist
elements that aim to make their particular worldview against modernity predominate. The
scholars participating in this project define fundamentalism as a “discernible pattern of
religious militancy by which self-styled „true believers‟ attempt to arrest the erosion of
religious identity, fortify the borders of the religious community, and create viable
alternatives to secular institutions and behaviors” (Almond et al. 2003: 17). They posit
that in democracies fundamentalist religious strategy is “designed first in order to create a
„defensive perimeter‟ and later as a mode of enlarging the hold over civil society, with the
hope of achieving hegemony there.” Almond et al. suggest that the source of conflict is
first the fundamentalist minority‟s fight for survival and second the majority's fear of losing
hegemony.
Bruce (2000) also posits that fundamentalism is on the rise because of the effects of
modernization on social, cultural and individual values of different societies. Thus he
offers a structural clue as to what distinguishes between Islamic and Protestant
9
fundamentalisms because Protestant fundamentalism generally takes place in democratic
settings such as the U.S. where the fundamentalists become a pressure group within the
political and social system. Islamic fundamentalists, on the other hand, are often active in
non-democracies where the only way of existence for fundamentalists may be to overturn
the system and to establish their own. In such instances violence is an expected corollary
of fundamentalism.
There is support in the literature for the idea that cutting across religions,
fundamentalism plays a role in motivating religious violence in general and terrorism in
particular (Emerson and Hartman, 2006). Hadden and Shupe (1998), for example, note that
fundamentalism is not in the domain of Islam only. They argue that fundamentalism is a
way for people to deal with changing values of the modern world, thus globalization is the
major cause of fundamentalist movements all over the world. Juergensmeyer (1995: 353)
argues in the Fundamentalism Project that „antiFundamentalism‟ is dangerous because it
can be used to justify human rights abuses and limitation of civil liberties in an effort to
„stop fundamentalism before it spreads.‟ Schmid (2004) conceptualizes motives of
terrorism in various ways such as criminal, political, tool for warfare, communication and
religious fundamentalism. Schmid does not necessarily link terrorism to religious
fundamentalism as he thinks secular terrorism is a wide-spread phenomenon but he does
not dismiss religious fundamentalism as a cause of terrorism either (p. 210). Schmid also
notes that Juergensmeyer (2003) shows how religious fundamentalism can indeed lead to
terrorism because of the concept of martyrdom in religion – which is quite different than
secular terrorism. According to a quantitative study by Enders and Sandler (2006) “today‟s
fundamentalist terrorists are members of all the major religions; however, the primary force
10
is Islam (Hoffman 1998; White 2003). [Furthermore, since] the rise of fundamentalist
terrorism, the proportion of incidents with deaths or injuries has increased greatly” (p. 47).
The difference between the implications of the literature that emphasizes
fundamentalism as a cause of terrorism and the other strands in the literature that pinpoint
particular religious doctrine as the root cause is quite significant. Most importantly,
according to the fundamentalist perspective any particular religious group is no more or
less likely to engage any other particular religious group in conflict. For example, the
worldview of Muslims is not particularly incompatible with that of Christians, Hindus or
Jews. Rather the answer to our question of how religion affects conflict comes down to the
probability of cooperation. If the objective of a particular group is to change the dominant
worldview and who controls that worldview, this group is more likely to engage other
groups in conflict - no matter weather the other group includes a fundamentalist element or
not. It bears clarifying that we consider "worldview" to be synonymous to political order
and posit that these are strategic considerations rather than more esoteric ones. This tension
between opposing strategic political objectives might result in terrorism in multiple ways.
For example, a fundamentalist majority might antagonize and threaten a minority to such
an extent that the minority resort to terrorism to drive off the majority advances. The
majority might even support terrorist attacks against a minority within its borders.
Conversely, a fundamentalist minority might initiate terrorism against a majority or another
minority in the same country in an effort to spread its worldview. At a minimum, the
particulars of the religion do not matter but the method of motivation does, we propose that
in any given country:
11
H2. The presence of a religious fundamentalist faction in a minority or a majority group
increases the probability of terrorist conflict between these groups.
TESTING
The study of terrorism is not new and cross national empirical testing of the causes
of terrorism across cases is emerging. This includes most prominently a number of studies
that examine the causes of suicide terrorism (Pape, 2005; Asal, 2006; Berman and Laitin,
2005; Hafez, 2006; Moghadam, 2006; Horowitz, 2006; Wade and Reiter, 2007). Others
have examined the effect of economic globalization on the number of transnational terrorist
incidents (Li and Schaub, 2004) and the effect of regime type on transnational terrorism
(Li, 2005). Supporting prior analysis of Crenshaw (1981) and Taylor (1988), Krueger and
Maleckova (2003) use quantitative data to show that terrorists are as likely to be educated
and wealthy as un-educated and poor. Enders and Sandler (2006) find no quantitative
evidence of an income based post-9/11 transfer of attacks to low-income countries except
for attacks with U.S. casualties, but do find that there is a significant transference to the
Middle East and Asia where U.S. interests are, at times, attacked.
However, while Fox (2002) examines the general effect of religion on conventional
violence against the state the effects of divergent religions on terrorism remains mostly in
the domain of qualitative case studies. The same is true for the effect of fundamentalism.
Indeed, the only empirical analysis that we are aware of that explicitly aims to account for
the effect of fundamentalism does so by assuming we are now in an era of fundamentalist
terrorism and denotes this effect as an intercept shift “starting with the 4 November
takeover of the US embassy in Teheran” (Enders and Sandler, 2006:81).
The Research Design
12
The question we ask in this paper is what effect religious differences and religious
fundamentalism have on terrorist incidents within a country‟s borders? The ideal research
design would account for all possible configurations of groups, their religious affiliations
and fundamentalist elements in those religions. We would then like to match each group
up with every terrorist incident, separating between intra-state and inter-state
(transnational) terrorism. Finally, through the use of multiple regression we would
compare violent groups to non-violent ones on the dimensions on religious families and
fundamentalism while holding a number of other important elements constant. The current
version of this paper falls short of the ideal as outlined below. Nonetheless, as we also
explain we have made sufficient progress to be confident that the basic results we show
here are not likely to change as we gather additional data – the details of the story,
however, will be better fleshed out in subsequent versions.
The appropriate universe of cases for this project contains all religious groups in the
population of independent countries in the world. For a catalog of all groups we rely on the
new lists of groups in the Minorities at Risk (MAR) project (Birnir and Wilkenfeld, 2007).
In this project a list is collected of all groups world-wide that meet the below criteria in
countries with a population of at least ½ million inhabitants in 2007:
1 Membership in the group is determined primarily by descent by both members and
non-members.
a) The group may be a caste if membership is determined by descent and
precludes public social mobility.
2 Membership in the group is recognized and viewed as important by members and/or
non-members. The importance may be psychological, normative, and/or strategic.
3 Members share some distinguishing cultural features, such as common language,
religion, occupational niche, and customs.
4 One or more of these cultural features are either practiced by a majority of the
group or preserved and studied by a set of members who are broadly respected by
the wider membership for so doing.
13
5 The group has at least 100,000 members or constitutes one percent of a country‟s
population.
The total number of groups that meet the above criteria is around 1300. The basic design
of the research in this paper includes all of these groups. The current sample we analyze is
curtailed due to uncertainties about the independent variable of fundamentalism but we do
not anticipate that the full data will fundamentally alter the general conclusions we reach in
this paper for reasons explained below. Furthermore, the longitudinal research design in
this paper calls for matching each of these groups up with terrorist incidents they have
perpetrated each year. We have begun matching and future versions of this paper will
complete this work, we are less certain about the effects of matching.
Dependent variables:
Our dependent variables come from the Global Terrorism Database (GTD) covering
59503 terrorist incidents from 1970-1997. The database codes intra and interstate terrorism
by perpetrator, location and year in addition to information describing the attack further. In
this version of the paper we use three dependent variables. First we counted the overall
number of terrorist attacks perpetrated in a given year in a particular country. Furthermore,
by country and year we counted the number of successful and unsuccessful attacks. An
attack is successful when it is carried out in full, while an unsuccessful attack is prevented
before it results in violence. Finally, we created a simple dummy variable accounting for
whether there were any terrorist attacks in a given country in a given year.
Independent Variables:
The independent variables of interest, accounting for differences in group religion
and religious fundamentalism, are original data. Our collection of these data is a work in
14
progress. We have finished the coding of religious differences but would like to code a
second fundamentalism variable for reasons of case selection.
Religious differences.
For minority groups we coded the following religious families: Islam (with Sunni,
Shi‟i, Ibadhi sects), Christianity (with Catholic, Protestant, Orthodox, Anglican,
Independent sects), Baha‟ism, Hinduism, Sikhism, Jainism, Yazidism, Judaism, Buddhism,
Animism, Taoism and Shinto. We code Methodism as a movement in Christian
Protestantism; Pentecostalism and Letter Day Saints (Mormons) as movements in the
Christian Independent sect. We also have a variable for non-religious minority groups.
The religious families coded for majority groups: Islam (with Sunni, Shi‟i, Ibadhi sects),
Christianity (with Catholic, Protestant, Orthodox, Anglican, Independent sects), Hinduism,
Judaism, Buddhism, Animism, Taoism, Shinto. Also, we code Methodism as a movement
in Christian Protestantism; Pentecostalism as a movement in the Christian Independent
sect.
Our sources for this variable are the World Factbook (www. cia.org) for majorities
and the Joshua Project (www.joshuaproject.com), Library of Congress and World
Directory of Minorities (1997) for minorities from 1945 to 2007. Indigenous / traditional /
ethnic beliefs in African states mentioned in these sources are entered as „Animism‟
following the U.S. Department of State web site on country studies, and the data set we
assembled for an earlier paper (Birnir and Şatana, 2007). Finally, we cross-checked against
alternative sources and local web sites to confirm the minority religion.
Fundamentalism.
15
In coding fundamentalism we relied on Almond et al. and the Fundamentalism
Project volumes. They define fundamentalism as a “discernible pattern of religious
militance by which self-styled „true believers‟ attempt to arrest the erosion of religious
identity, fortify the borders of the religious community, and create viable alternatives to
secular institutions and behaviors” (Almond et al. 2003: 17). Both majority and minority
groups in a country can contain fundamentalist elements and this is irrespective of religious
family to which the majority and minority belong. In many cases, however, neither the
majority nor the minority group contains fundamentalist elements, and Almond et al.
describe fundamentalism as a “rare phenomenon.” In the sample of all groups from 1970-
1997 we coded two fundamentalist variables.6 The first variable accounts for whether the
majority religion contains a fundamentalist element with reference to majority religions in
other countries that do not have a fundamentalist element. The second variable accounts
for whether the minority religion contains a fundamentalist element with respect to
minority religions that do not. To clarify, the operationalization of these variables depends
on our reading of all of the individual cases in the fundamentalist project. If, using the
above definition, the author of the particular case makes the argument that a minority or a
majority contains or does not contain a fundamentalist faction we code the groups
accordingly.
To our knowledge, the Fundamentalism Project volumes discuss all
fundamentalist movements that fit the controversial definition of religious fundamentalism
set out by the project. Since there are differing and debatable definitions of
fundamentalism, we preferred to use one definition and one source for this variable.7
16
The problem, however, is that the fundamentalist project does not define explicitly the
universe of cases that were selected for the project. Consequently, we cannot be certain
that the project identifies all fundamentalist religious groups and, therefore, cannot code
religious groups that are not discussed in the project as lacking a fundamentalist element.
Assuming that the Fundamentalist project does not identify all religious fundamentalist
groups the concern becomes one of selection bias. Possibly the project contains only the
most prominent fundamentalist groups while fundamentalist groups that are less visible are
left out. If this is the case the analysis is unreliable if left out groups differ systematically
in their propensity for violence from groups that are included. Our cursory survey suggests
that not many fundamentalist groups were excluded from the project. Second, we are not
particularly concerned that groups that possibly were left out differ significantly from
groups in the sample because many of the fundamentalist groups discussed in the
Fundamentalist project are small and not violent.
The second concern about selection bias is that if there truly is a positive
relationship between fundamentalism and terrorist violence and we leave out most of the
cases that are not fundamentalist – we have effectively over-selected cases with a high
value on the dependent variable. Consequently, we will tend to have a disproportionate
number of cases that, because of other variables unrelated to the independent variable we
are examining, were high on the dependent variable of terrorism. The well known result is
that the true relationship between fundamentalism and terrorist violence would be flattened
and harder to detect (Geddes 2003, Shively 2006). As the results show this is not a
problem for detecting the effect of fundamentalism but serves as a caution against reading
17
anything into the magnitude of the coefficients other than their signs and size relative to
each other.
Control Variables
Due to the dearth of cross national empirical analysis of the causes of religious
terrorism it is not clear precisely what the most appropriate controls for these equations
should be.8 Presumably, however, ongoing conflicts are more likely to produce continued
attacks than situations where terrorism has never been used. Consequently, we begin by
including a lag of the dependent variable on the right hand side of the first two equations to
account for autocorrelation. It is not likely, however, that the third dependent variable –
number of successful terrorist attacks carried out in a particular country in a given year – is
as dependent on the preceding number of successful attacks as it is on the total number of
attacks carried out in the same year. However, the two variables; total number of attacks
and the number of successful attacks are highly correlated (over 0.9). Consequently, in the
equation that has as its dependent variable number of successful attacks we include, on the
right hand side, a variable accounting for the total number of terrorist attacks carried out in
a particular country in the previous year. Furthermore, the extant wealth of case studies
informs us that country specific conditions are instrumental to the onset, number, and
success of terror attacks. Therefore, all of the below analyses include country fixed effects.
The Methods
The dependent variables determine the method used in each case and we also run
Ordinary Least Squares Regression for each equation to probe robustness to alternate
methods of analysis. The first dependent variable, annual count of terror incidents per
country is a count variable. The number of incidents runs from 0 to 778 attacks in a single
18
year. The average is 43.66 and the standard deviation is 12.47. Poisson is the appropriate
form of estimation for count variables if the mean and variance of the dependent variable
are approximately equal (Long & Freese, 2001). The incident count variable is clearly
more dispersed than that. Therefore, we use Negative Binomial Regression to assess the
effect of the independent variables on the count outcome, the number of terror incidents
perpetrated in a country in a given year.
The second dependent variable, presence of any terrorist incidents in a given
country in a given year is a binomial variable. Consequently, we assess the effects of the
independent variables using Logit regression. The third dependent variable, number of
successful terror attacks in a year is also an over-dispersed count variable and we estimate
the effects with a negative binomial regression.
RESULTS
Table one shows the effects of religion on the number of terrorist attacks
perpetrated in a country in any given year between 1970 and 1997. The first specification
includes independent variables accounting for majority and minority religious family as
well as majority and minority fundamentalism. In addition this specification contains
control variables that account for the number of terrorist attacks perpetrated in the prior
year to account for possible autocorrelation of violence in addition to country dummies
whose coefficients are excluded due to space constraints. The second specification is
identical to the first but the estimation technique is different.
According to Table one religious family constellation of a country has no
systematic effect on the number of terrorist attacks perpetrated in that country in a given
year. Because of sign changes, the only variables denoting any significance of religious
19
family are unreliable between estimation techniques and must, therefore, be interpreted as
not having any clear relationship with the dependent variables. Interestingly however, both
minority and majority fundamentalism are associated with increased numbers of terrorist
attack. Notably the effect of majority fundamentalism is at least triple that of minority
fundamentalism. As expected, prior terrorist violence is an excellent indicator of
subsequent terrorist violence as are many of the country fixed effects.
[Insert Table 1 about here]
Table two shows the effects of religious family and fundamentalism on the
probability that a country experience any terrorist violence in a given year. Controlling for
violence in a prior year and country fixed effects the results suggest that countries with
Islamic and Hindu majorities are less likely to experience any terrorist violence. The
significance of the variable accounting for Islamic majorities is not, however, robust to the
arguably more correct estimation technique. Interestingly, Hindu majorities seem less
likely to experience terrorist violence but it is possible that the India country dummy was
dropped due to collinearity with this variable so that this result is only picking up the effect
of India. In general, dropping of country fixed effects was a problem when using the more
intensive estimation techniques. This suggests that regional fixed effects may be more
appropriate than country effects and that we need to examine in detail findings that suggest
a particularly religious combination has an effect on terrorism until we verify that this
result is not simply driven by a single country.
Again, minority fundamentalism is significantly and robustly associated with
increase in terrorist violence whereas the same is only true for fundamentalist majorities in
the specification analyzed using Logit.
20
[Insert Table 2 about here]
The third table accounts for the number of successful terrorist attacks perpetrated in
a given country in a particular year. Again the same pattern emerges with respect to
religious family – showing that religious family of neither the majority nor the minority
matters for the number of successful terrorist attacks carried out in a given country in a
particular year. The presence of a fundamentalist majority does, however, increase the
number of successful attacks whereas the presence of a fundamentalist minority has little
effect on the success of terrorist attacks.
[Insert Table 3 about here]
CONCLUSION
In addition to the fact that the sample likely over-represents fundamentalist groups
(violent and non-violent ones) it is important to remember that current runs include both
intra and interstate terrorist violence. Furthermore, terrorist incidents have not been
matched with groups so that the coding currently holds all groups accountable for violence
perpetrated in a particular country. It is not clear that this last coding convention sways the
results one way or another because it is just as likely that this increases the number of non-
fundamentalist groups inaccurately “held responsible” for violent attacks as it is that
fundamentalist groups are inaccurately “credited with” with violent attacks perpetrated by
others. Nonetheless, for theoretical clarity in subsequent versions of the paper we intend to
separate out intra vs. inter state terrorism, and for empirical clarity to match incidents with
particular groups.
What we can say about these results is that it seems pretty evident that religious
family orientation of minority and majority groups has no systematic effect across nations
21
on the presence or number of terrorist attacks carried out in a particular country. This
finding is consistent with a prior paper on motivations of ethnic minority rebellion against
the state where we found no systematic effect of divergence in majority minority religious
family orientation (Birnir and Şatana, 2007).
To the contrary, it also seems clear that fundamentalism, irrespective of doctrine,
matters for terrorism. Majority fundamentalism is significantly associated with an
increasing number of terrorist incidents as is minority fundamentalism, though that effect is
considerably smaller. Furthermore, minority fundamentalism is significantly associated
with the presence of any terrorist incident and majority fundamentalism may be as well
although that is less clear. These findings are also consistent with our prior work where we
found that fundamentalist elements in the majority or minority religion significantly
increase the probability that a minority engage in violent rebellion against the state (Birnir
and Şatana, 2007).
The fact that these results are in line with our prior findings that fundamentalist
groups are associated with presence and greater number of conventional violence incidents
against the state is not all-together surprising as one might expect that groups who use
violence strategically will employ many different types of violence and may graduate from
using one type of violence to another when the first type outlives its usefulness or is not
practical for logistical reasons. To gauge the accuracy of this conjecture we examined the
sample of minority group rebellion against fundamentalism and Memorial Institute for the
Prevention of Terrorism (MIPT) and Department of Homeland Security Terrorism
Knowledge Database terrorism indicators. Interestingly we found that the average level of
rebellion against the state among non-fundamentalist minority groups is .475, the average
22
level of rebellion against the state among fundamentalist groups is 1.266, and the average
level of rebellion among fundamentalist groups that also engage in terrorist activities is
2.180.9
To conclude, therefore, it seems clear that doctrine simply is not a good predictor of
violence. Fundamentalism, however, is a strong consistent predictor but the precise
reasons for why this is the case remain the topic of future research.
23
Table 1
Estimation technique
Dependent variable:
Number of terrorist
attacks perpetrated in
a country in a year
OLS NBREG
Number of terrorist
attacks in the prior
year
0.664
(0.009)**
0.013
(0.001)**
Minority religion is
Islam
0.030
(1.168)
-0.000
(0.055)
Minority religion is
Christianity
-0.094
(1.152)
-0.007
(0.054)
Minority religion is
Hindu
0.352
(1.496)
0.016
(0.071)
Minority religion is
Sikh
-1.612
(4.074)
-0.065
(0.184)
Minority religion is
Judaism
-0.004
(3.149)
-0.002
(0.141)
Minority religion is
Buddism
0.196
(1.500)
0.008
(0.073)
Majority religion is
Islam
-18.902
(7.383)*
0.257
(0.487)
Majority religion is
Christianity
-0.551
(10.973)
3.698
(0.396)**
Majority religion is
Hindu
15.829
(10.502)
-1.532
(0.343)**
A minority religion
contains a
fundamentalist faction
5.534
(2.458)*
0.240
(0.111)*
The majority religion
contains a
fundamentalist faction
18.708
(4.208)**
1.799
(0.223)**
Country fixed effects
(161 countries)
coefficients not
reported due to space
Constant 1.269
(5.715)
-1.102
(0.306)**
Observations 7258 7258
R-squared 0.59
Standard errors in parentheses
* significant at 5%; ** significant at 1%
24
Table 2
Estimation technique
Dependent variable:
The country
experienced a terrorist
attack in a given year
OLS LOGIT
The country
experienced a terrorist
attack in the prior year
0.199
(0.011)**
0.928
(0.061)**
Minority religion is
Islam
0.000
(0.015)
-0.001
(0.083)
Minority religion is
Christianity
-0.001
(0.014)
-0.007
(0.083)
Minority religion is
Hindu
0.005
(0.019)
0.026
(0.101)
Minority religion is
Sikh
-0.025
(0.051)
-0.296
(0.370)
Minority religion is
Judaism
-0.000
(0.039)
-0.020
(0.307)
Minority religion is
Buddism
0.003
(0.019)
0.018
(0.108)
Majority religion is
Islam
-0.296
(0.092)**
-0.495
(0.879)
Majority religion is
Christianity
-0.117
(0.138)
-0.494
(0.819)
Majority religion is
Hindu
-0.084
(0.132)
-1.612
(0.582)**
A minority religion
contains a
fundamentalist faction
0.085
(0.031)**
0.584
(0.205)**
The majority religion
contains a
fundamentalist faction
0.098
(0.053)
0.718
(0.333)*
Country fixed effects
(161 countries)
coefficients not
reported due to space
Constant 0.652
(0.072)**
1.554
(0.686)*
Observations 7258 7234
R-squared 0.26
Standard errors in parentheses
* significant at 5%; ** significant at 1%
25
Table 3
Estimation techniques
Dependent variable:
Number of successful
terrorist attacks in a
country in a given year
OLS NBREG
The number of
terrorist attacks
perpetrated in a prior
year
0.653
(0.010)**
0.011
(0.000)**
Minority religion is
Islam
0.000
(1.460)
-0.000
(0.043)
Minority religion is
Christianity
-0.061
(1.451)
-0.002
(0.043)
Minority religion is
Hindu
0.159
(1.968)
0.005
(0.059)
Minority religion is
Sikh
-0.479
(4.495)
-0.016
(0.130)
Minority religion is
Judaism
0.002
(3.381)
-0.000
(0.097)
Minority religion is
Buddism
0.070
(1.928)
0.002
(0.058)
Majority religion is
Islam
0.367
(17.799)
-1.392
(0.515)**
Majority religion is
Christianity
-0.118
(12.260)
0.744
(0.467)
Majority religion is
Hindu
19.799
(9.017)*
0.512
(0.427)
A minority religion
contains a
fundamentalist faction
1.986
(2.793)
0.064
(0.081)
The majority religion
contains a
fundamentalist faction
29.531
(5.596)**
1.744
(0.182)**
Country fixed effects
(161 countries)
coefficients not
reported due to space
Constant 1.669
(10.249)
0.349
(0.271)
Observations 4815 4815
R-squared 0.65
Standard errors in parentheses
* significant at 5%; ** significant at 1%
26
REFERENCES
Almond Gabriel, Scott Appleby and E. Sivan. 2003. Strong Religion. The Rise of
Fundamentalisms around the World. Chicago: The University of Chicago Press.
Asal, Victor. 2006. “The Preconditions for Ethnic Suicide Bombing Campaigns 1991-
2003,” Paper presented at the annual meeting of the International Studies Association,
Town & Country Resort and Convention Center, San Diego, California, USA.
Berman, Eli, and David D. Laitin. 2005. “Hard targets: Theory and evidence on suicide
attacks,” Working paper no. 11740, Cambridge, MA, National Bureau of Economic
Research.
Birnir, Jóhanna K. 2007. Ethnicity and Electoral Politics. New York: Cambridge
University Press.
Birnir, Jóhanna K. and Jonathan Wilkenfeld. 2007. "Minorities at Risk: Addressing
Selection Bias Issues and Group Inclusion Criteria for Ethnopolitical Research." Funded
by The National Science Foundation grant #SES-0718957
Birnir Jóhanna K. and Nil S. Şatana. 2006. “Incompatibles jarring, or sparring of the sacred
and the secular? How minority religious dimensions affect a groups‟ propensity for access
to governing coalitions.”
Birnir Jóhanna K. and Nil S. Şatana. 2007. “Religious Violence.” Presented at the annual
meeting of International Studies Association (ISA), Chicago, IL, March 2007.
Bruce, Steve. 2000. Fundamentalism. Malden, MA: Blackwell Publishers Ltd.
Crenshaw, Martha. 1972. “The Concept of Revolutionary Terrorism,” Journal of Conflict
Resolution, 16 (3): 383-96.
Crenshaw, Martha. 1981. “The Causes of Terrorism,” Comparative Politics, Vol. 13: 379–
99.
Crenshaw, Martha. 1995. Terrorism in Context. University Park, PA: Pennsylvania State
University Press.
Crenshaw, Martha. 2000. “The Psychology of Terrorism: An agenda for the 21st century.”
Political Psychology 21 (2): 405-420.
Crenshaw, Martha. 2005. “‟New‟ vs. „Old‟ Terrorism.” Presented at the meeting sponsored
by Woodrow Wilson Center‟s Division of International Studies and Middle East Program,
the RAND Corporation and the U.S. Army‟s Eisenhower National Security Series. May
23rd. “http://www.eisenhowerseries.com/pdfs/final_05/final_2005-5-23.pdf
27
Crenshaw, Martha. 2006. “Old and New Terrorism – Lessons Learned” Presented at the
Second IRRI Conference on International Terrorism. February 13th.
Dobrot, Laurence A. 2007. “The Global War on Terrorism: A Religious War?” Strategic
Studies Institute, November.
http://www.strategicstudiesinstitute.army.mil/pdffiles/pub822.pdf
Duyvesteyn, Isabella (2004) “How New is the New Terrorism,” Studies in Conflict and
Terrorism, Vol. 27, No. 5: 439-454.
Eggen, Dan. 2005. “Rudolph To Plead Guilty to Bombings:‟96 Olympics Killer To Get
Life in Prison,” The Washington Post April 9.
Emerson, Michael O., and David Hartman. Forthcoming, 2006. "The Rise of Religious
Fundamentalism." Annual Review of Sociology 32: 127-44.
Enders, Walter and Todd Sandler. 2006. “Distribution of Transnational Terrorism Among
Countries by Income Class and Geography After 9/11,” International Studies Quarterly 50
(2), 367–393.
Feldmann, Andreas E. and Maiju Perälä. "Reassessing the Causes of Nongovernmental
Terrorism in Latin America." Latin American Politics and Society. 46(2):101-132.
Fox, Jonathan. 2000. “Religious Causes of Discrimination Against Ethno-Religious
Minorities,” International Studies Quarterly 44(3):423-450.
Fox, Jonathan. 2002. Ethnoreligious Conflict in the Late 20th Century: A General Theory.
Lanham, MD: Lexington Books.
Fox, Jonathan. 2004. Religion, Civilization, and Civil War: 1945 through the Millenium.
Lanham, MD: Lexington Books.
Geddes, Barbara. 2003. Paradigms and Sand Castles: Theory building and research design
in comparative politics. Ann Arbor: University of Michigan Press.
“Global Terrorism Database, START / CETIS, accessed via Global Terrorism Database
CD, June 2007 edition.”
Hadden, Jeffrey K. and Anson Shupe. 1991. “Secularization and Fundamentalism
Reconsidered: Religion and the Political Order,” Journal for the Scientific Study of
Religion, 30 (1).
Hafez, Mohammed M. 2006. “Suicide Terrorism in Iraq: A Preliminary Assessment of the
Quantitative Data and Documentary Evidence,” Studies in Conflict and Terrorism 29: 591-
619.
28
Hegghammer, Thomas. 2006. “Global Jihadism After the Iraq War,” Middle East Journal,
60 (1).
Hoffman, Bruce. 1993. “‟Holy Terror‟: The Implications of Terrorism Motivated by a
Religious Imperative.” http://www.rand.org/pubs/papers/2007/P7834.pdf
Hoffman, Bruce. 1998. Inside Terrorism. London: Colombia University Press.
Horowitz, Michael. 2006. “The Diffusion of Military Power: Causes and Consequences for
International Politics”, Ph.D. dissertation. Harvard University, Cambridge, MA.
Huntington, Samuel P. 1996. The Clash Of Civilizations And The Remaking Of World
Order. New York: Simon & Schuster.
Jelen Ted, G. and Clyde Wilcox (eds). 2002. Religion and Politics in Comparative
Perspective: The One, the Few, and the Many. New York: Cambridge University Press.
Joshua Project. http://www.joshuaproject.net/
Juergensmeyer, Mark. 1995. “Antifundamentalism” in Martin E. Marty and R. Scott
Appleby, eds. Fundamentalisms Comprehended, 353–366.
Juergensmeyer, Mark. 2003. Terror in the Mind of God: The Global Rise of Religious
Violence. University of California Press.
Kibble, David G. 2002. “The Attacks of 9/11: Evidence of a Clash of Religions?”
Parameters, Autumn. http://carlisle-www.army.mil/usawc/Parameters/02autumn/kibble.pdf
Krueger, Alan B. and Jitka Maleckova. 2002. “Education, Poverty, Political Violence and
Terrorism: Is There a Causal Connection?” NBER Working Paper No. W9074.
http://www.nber.org/papers/w9074
Li, Quan and Drew Schaub. 2004. “Economic Globalization and Transnational Terrorism,”
Journal of Conflict Resolution 48 (2): 230-258.
Li, Quan. 2005. “Does democracy promote or reduce transnational terrorist incidents?”
Journal of Conflict Resolution 49 (2): 278-97.
Library of Congress. http://www.loc.gov/index.html
Long, J. S. and J. Freese. 2001. Regression Models for Categorical Dependent Variables
using Stata. College Station, TX: Stata Press.
Marty, Martin and R. Scott Appleby, eds. Fundamentalisms Observed. The
Fundamentalism Project. Volume 1. Chicago: University of Chicago Press.
29
Marty, Martin and R. Scott Appleby, eds. 1993. Fundamentalisms and Society. The
Fundamentalism Project. Volume 2. Chicago: University of Chicago Press.
Marty, Martin and R. Scott Appleby, eds. 1993. Fundamentalisms and the State. The
Fundamentalism Project. Volume 3. Chicago: University of Chicago Press.
Marty, Martin and R. Scott Appleby, eds. 1994. Accounting for Fundamentalisms The
Fundamentalism Project. Volume 4. Chicago: University of Chicago Press.
Marty, Martin and R. Scott Appleby. eds. 1995. Fundamentalisms Comprehended. The
Fundamentalism Project, Volume 5. Chicago: University of Chicago Press.
Mickolus Edward and Susan L. Simmons. 2006. “Terrorism, 2002-2004: a Chronology.”
Praeger Security International.
Minority Rights Group International (eds.) World Directory of Minorities. 1997. MRG.
Moghadam, Assaf. 2006. “Suicide terrorism, occupation, and the globalization of
martyrdom: A critique of „dying to win‟,” Studies in Conflict and Terrorism 29: 707-29.
Morgan, Matthew J., (2004), “The Origins of the New Terrorism,” Parameters 34 (1), 29-
43.
Pape, Robert. 2005. Dying to Win: The Strategic Logic of Suicide Terrorism. Random
House Publishers.
Ranstorp, M. 1996. Terrorism in the Name of Religion. Centre for the Study of Terrorism
& Political Violence. Retrieved March 4, 2008 from Columbia International Affairs Online
http://www.ciaonet.org/wps/ram01/index.html
Rapoport, David C. 1984. “Fear and Trembling: Terrorism in Three Religious Traditions,”
American Political Science Review, 78, 668-72.
Rapoport, David C. 2002. “The Four Waves of Rebel Terror and September 11”
Anthropetics,Vol 8, no 1.
Rashid, Ahmed. 2002. Jihad: The Rise of Militant Islam in Central Asia. New Haven: Yale
University Press.
Şatana, Nil. 2006. "Third Party Intervention in Civil Conflicts: Sorting Out the Effect of
Domestic Factors". Unpublished Manuscript.
Schmid, Alex P. 2004. “Framewoks for Conceptualising Terrorism.” Terrorism and
Political Violence 16:197-221.
Seul, Jeffrey R. 1999. “„Ours is the Way of God‟: Religion, Identity, and Intergroup
Conflict.” Journal of Peace Research 36 (5), 553-569.
30
Shively, Phillips. 2006. “Case Selection: Insights from Rethinking Social Inquiry.”
Political Analysis 14(3):344-347.
Stark, Rodney. 2001. One True God: Historical Consequences of Monotheism. New Jersey:
Princeton University Press.
Stohl, Michael. 1988. The Politics of Terrorism. New York: Marcel Dekker.
Taylor, Maxwell. 1988. The Terrorist. London: Brassey‟s.
“The New Terrorism. The FBI Cracks A New York Bomb Plot. But America Is Now More
Vulnerable To Attacks At Home.” Newsweek July 5, 1993.
The MAR Users Manual. www.cidcm.umd.edu/inscr/mar/margene/mar-
codebook_040903.pdf
Victoroff, J., (2005), “The Mind of the Terrorist: A Review and Critique of Psychological
Approaches.” Journal of Conflict Resolution 49(1):3−42.
Wade, Sara Jackson and Dan Reiter. 2007. “Does Democracy Matter? Regime Type and
Suicide Terrorism,” Journal of Conflict Resolution 51 (2): 329-348.
Walzer, Michael. 2002. “Five Questions About Terrorism.” Dissent 49(1):5–11.
White, Jonathan R. 2003. Terrorism: 2002 Update, Belmont, CA:
Wadsworth/Thomson Learning.
Wiktorowicz, Q. 2006: Anatomy of the Salafi Movement. Studies in Conflict and
Terrorism 29: 207-239.
31
Notes
1 Jelen and Wilcox (2002) go on to suggest that religious competition may only be a
problem where at least one of the religious traditions is monotheistic.
2 Neither Huntington (1996) nor Fox (2004) argue that civilizations are composed entirely
by religious affiliation. However, they both acknowledge that religion is an important
component of civilization.
3 Fox (2004) also tests Huntington‟s clash of civilizations argument by coding a variable to
account for civilizations. He does not find support for Huntington‟s paradigmatic thesis.
Fox does, however, find support for the idea that religion affects domestic conflict as an
intervening variable: certain religions have fought more often than others and religious
conflict has increased since around 1980.
4 Clearly as the case of Northern Ireland demonstrates this is not an iron clad rule.
However, religious differences in Northern Ireland are also reinforced by economic
cleavages, which might exacerbate the conflict.
5 In the latest attacks in Diyarbakir in January 3, 2008, 7 people died of whom 6 were
young high school students. Interestingly, socio-economic explanations do not fair much
better than religious explanations. Interviews with convicted Kurdish terrorists in 2006
suggest that terrorism is used as a rational strategy by the PKK to force the hand of Turkish
state for recognition of more cultural and political rights for the Kurds (Satana, 2006).
6 We have coded the fundamentalism variables starting from the year of independence of
each country or 1945 if the country got independence before 1945. The data end in 2003
since the Almond et al. book covers fundamentalist groups until 2003.
32
7 Almond et al. 2003 or the Fundamentalism Project volumes do not discuss case selection
criteria in great detail. For example, Pentecostal fundamentalists in various Latin American
countries are mentioned in the Project but Almond et al. (2003) mentions Guatemala in a
general discussion of fundamentalism and dismisses Ecuador Pentecostals as
fundamentalists. We coded positively only countries and groups that were explicitly
described as having a fundamentalist group. Countries that are not discussed are coded as
not having fundamentalist groups.
8For example, the most common socio-economic controls related to the terrorist profile
such as education of the terrorist, social inequality, employment situation etc. are not
available cross-nationally. Subsequent, versions will, however, include macro-variables to
account for state repression (Polity 4) and change in GDP per capita for relative change in
economic deprivation.
9 If groups tend to progress from one type of violence to another in an attempt to get an
unresolved grievance addressed, research on group grievances such as political access
(Birnir, 2007) may inform our future revisions of theory and tests of the relationship of
religion and terrorism.