International Studies Quarterly (2007) 51, 927–943

Islam’s Bloody Innards? Religion and

Political Terror, 1980–2000
Indra de Soysa
Norwegian University of Science and Technology (NTNU) and Centre for the Study
of Civil War (CSCW), International Peace Research Institute, Oslo (PRIO)

Ragnhild Nordås
Centre for the Study of Civil War (CSCW), International Peace Research
Institute, Oslo (PRIO)

Culturalists claim that political outcomes, such as respect for human

rights, are deeply rooted in culture. Some have singled out Islam as par-
ticularly problematic. We assess whether Muslim societies suffer higher
levels of political terror compared with others. Our results show that
countries containing larger shares of Catholics, and those dominated by
Catholics, fare the worst. The share of the population Muslim and
Membership in the Organization of the Islamic Conference predicts
lower levels of political terror. Claims about the uniqueness of Islam for
accounting for political repression seem to be exaggerated. Consistent
with the findings on religion and democracy, our results indicate that it
is the Arab region, not religion that matters, but Latin America shows
the largest impact. Substantively, political and economic factors matter
a whole lot more than do the variables on religion. This is good news
for policy that seeks to end the scourge of political repression.

In July 2004, a prominent Norwegian politician, Carl I. Hagen, in an address to

a religious group, commented that Jesus Christ was kind to children whereas the
Prophet Mohammed would turn them into suicide bombers. Such rhetoric reso-
nates with a large segment of public opinion across the globe. The portrayal of
the Prophet Mohammed as a terrorist by a Danish cartoonist, which led to mas-
sive protests in many parts of the world, is emblematic of a Western public’s
image of Islam. Even the Pope became embroiled in controversy for quoting an
11th Century text that suggested that Islamic teachings incite violence. In the
post-9 ⁄ 11 world, ‘‘Islamist’’ terrorists have used suicide-bomb tactics from Bali to
London, which has spawned a public discourse that ties Al Qaeda’s ideology to a
larger Muslim cultural trait—the use of violence for political ends.1 These events

Author’s note: We are grateful to Nils Petter Gleditsch, Rodwan Abouharb, Halvard Buhaug, Erich Weede, Rich-
ard Pape, and Ulrika Mårtensson for comments and suggestions. We deeply appreciate the comments and sugges-
tions of the editor and several anonymous reviewers, who were unusually constructive. Only we are responsible for
errors. The replication data are available at http://www.isanet.org/data_archive/. Equal authorship. We dedicate
this article to the memory of Steven Poe, who has inspired us and taught us much.
1
The European Union decided to ban the term ‘‘Islamic terrorism’’ because it was too emotive, as terrorism
by Protestant fundamentalists, Hindus, Buddhists, and Catholics are rarely prefaced by the religious denomination
(Beunderman 2006).

 2007 International Studies Association.

Published by Blackwell Publishing, 350 Main Street, Malden, MA 02148, USA, and 9600 Garsington Road, Oxford OX4 2DQ, UK.
1 928 Islam’s Bloody Innards?

play into anti-Muslim feelings, even within bastions of liberal Western states, such
as the Netherlands, Denmark, and France. Discussion on the Turkish bid to join
the European Union is illustrative. Europeans view Turkish membership as prob-
lematic largely because it is a majority-Muslim country.2 Violent fundamentalist
groups, such as Al Qaeda, also claim that they represent those oppressed by
‘‘false’’ Islamic governments, vowing to restore ‘‘true’’ Islam. Discussions in the
West on questions of tolerance and human rights are often reduced to contrast-
ing a progressive liberal West with a ‘‘backward’’ Islamic world, which is mired
in tradition (Esposito 1999:245). Do Muslim societies have particularly ‘‘bloody
innards,’’ as some culturalists have claimed? (Huntington 1996). This study
addresses this debate by empirically examining whether countries with larger
shares of Muslims, and those demographically dominated by Muslims, are more
likely than others to experience state-sponsored repression.
Recently, the influence of religious–cultural factors have received much atten-
tion in economics, political science, and peace studies (e.g., Diamond 1994;
Esposito 1999; Fox 2000; Hasenclever and Rittberger 2000; Russett, Oneal, and
Cox 2000; Fish 2002; Norris and Inglehart 2004). Apparently, norms, values, and
attitudes conditioned by religion determine political and economic life to a sig-
nificant extent (North 1990; Landes 1999). Many argue that ‘‘culture matters’’
(Price 1999; Harrison and Huntington 2000; McTernan 2003), and some depict
Islamic culture in particular as worrisome, largely because of Islam’s inability to
deal with a secularizing, modern world (Barber 1995; Huntington 1996; Juer-
gensmeyer 2000; Hutchinson 2005).
We use the 5-point Political Terror Scale (PTS), which is based on country
reports by Amnesty International and the United States’ State Department, as our
2 main measure of political repression.3 Using these data,4 we find that countries
with higher shares of Muslims suffer much lower levels of political repression, rel-
ative to states with higher shares of Catholics. The results are robust to different
operationalizations of the religion variables, including dichotomized measures
capturing Muslim dominance and membership in the Organization of the Islamic
Conference (OIC).5 Fortunately, the substantive impact of religion seems to be
very slight when compared with political-economic variables, such as democracy,
per capita income, oil wealth, and involvement in civil war. Our results are robust
to different measures of repression, such as the physical integrity rights (PIR)
measure of the CIRI data project (Cingranelli and Richards 1999), alternative
specifications of the model, sample size, and several appropriate testing proce-
dures for cross-section, time-series data, and continuous and ordinal data.

Why Repress?
It is not obvious why religion should predict the level of political terror in a soci-
ety. We first outline a model of repression from the extant literature on the sub-
ject and then discuss the ways in which Islam may relate to such a model.
Human rights practices depend primarily on decisions made by leaders from a
menu of choice about how to respond to a perceived or real threat. According
to Poe (2004), repression is likely when state leaders try to decrease threats

2
Ironically, Serbia, a country against which NATO went to war for preventing genocide, has a greater chance
of joining Europe than Turkey, a NATO ally!
3
This measure was first introduced by Stohl, Carleton, and Johnson (1984) and compiled by many researchers,
foremost among them Gibney and Dalton (1996). We use the Hafner-Burton replication data available at http://
www.prio.no/jpr/datasets. We refer the reader to her article for details.
4
We also run robustness tests using the Physical Integrity Rights (PIR) index (Cingranelli and Richards 1999).
The results using PIR support the findings using PTS. The PIR data are described at http://ciri.binghamton.edu/
index.asp. All results are available in the web Appendix.
5
Membership in the OIC is obtained from the OIC website: http://www.oic-oci.org/.
 INDRA DE SOYSA AND RAGNHILD NORDA˚S 929

or ⁄ and increase its strength relative to domestic adversaries. Others, too, have
variously argued some version of this parsimonious model when they have sug-
gested that political leaders are most likely to use repression as a means to gain
control over serious dissent (Gurr 1986; Gartner and Regan 1996; Moore 2000;
Landman 2005). Such practices as political murder, torture, disappearances, and
imprisonment for political activity, reflected in the standard measures of political
terror, are more likely to occur when some level of political dissent threatens
incumbent rulers. We do acknowledge, of course, that some rulers may also
repress without facing any level of dissent and use other means, such as reform
and cooptation, to deal with threats. Nevertheless, if in fact Islam is uniquely
‘‘bloody’’—our main focus of study—then these data are ideally suited for test-
ing the proposition that ordinary people face harsher conditions for political
activity in Muslim states.
Given this model, the importance of religion for repression can be conceptual-
ized in two broad ways: on one hand, particular aspects of a religion may
increase the level of dissent from below because religious dogma provides the
impetus for challenging rulers in violent ways. On the other hand, certain fea-
tures of religious institutions, or social attitudes and beliefs, may legitimize some
state actions against individuals over others. With this broad schema in mind, we
examine arguments about why Islam may matter for predicting the dissent–
repression nexus.

Why Islam?
Culturalist see political practices and outcomes as deeply rooted in cultural
practices and beliefs, shaped by theology (Huntington 1996; Lewis 2003; Pipes
2003). Huntington (1996:258) argues that there are sharp cultural differences
between the political values of a western Christian heritage and Islam—appar-
ently Islam has bloody borders as well as ‘‘bloody innards.’’ Rather than geo-
strategic issues, it is cultural values that make Muslim states prone to violence.
Others suggest that state repression relates to structure, political strategies,
short-term strategic questions of power and profit, and that culture is merely
an epiphenomenon (Jackman and Miller 1998; Booth and Seligson 2004; Nasr
2005).
In the first instance, we examine possible ‘‘bottom up’’ effects of Islam on the
level of state repression. Particular teachings of Islam could empower ordinary
people to high levels of dissent. The most commonly discussed aspect of Islamic
teaching is the socially sanctioned right of Jihad, where every individual is bound
by faith to fulfill this duty.6 Some Islamist leaders, such as Ayatollah Khomeini
and Osama Bin Laden, have issued fatwahs that include suicide bombing as a
legitimate tactic, and these events are seen as a civilizational trait of Islam (Fried-
man 2005). Anti-western sentiments, particularly over the West’s support for the
state of Israel, may allow radicals to rally the masses against their ruling elites.
According to Huntington (1996:175), Muslims are less loyal to the state because
of the concept of the Ummah, which emphasizes allegiance to a larger religious
community, as opposed to a territorially bound, secular state. The rise of
so-called Islamic fundamentalism is often equated with the desire of true believers
to bring about the realization of the sovereignty of Allah over secular political
authority. All of this suggests that ruling elites within Muslim states could face
uniquely high levels of dissent from below.
Does religion also determine state responses and strategies from the ‘‘top
down’’? State elites and the ruling class could be overly sensitive because

6
On whether it is Islam rather than other convictions that may have driven the September 11th hijackers to
crash planes into buildings in the U.S.A., see Holmes (2005).
1 930 Islam’s Bloody Innards?

religious opposition to secular rule is ever present; a pattern identified in the

Islamic world (Hafez 2003). According to Smock (2002:8), authoritarian regimes
in Turkey, Malaysia, and Algeria have employed harsh measures to suppress Isla-
mic movements that were gaining support. Even more liberal forms of political
Islam have been subjected to repression by state authorities as ‘‘undesirable
expressions of political dissent’’ (Smock 2002). Apparently, one cultural trait
attributed to Islam is the preference for order over chaos, which legitimizes the
use of harsh methods against any forms of organized opposition (Esposito
1999). As a result, Price (1999:162) claims that ‘‘the abusive treatment of citizens
by governments became acceptable in many Islamic societies.’’ Moreover, in
Egypt, ‘‘opposition groups are tolerated only as long as they remain relatively
weak, or accept government control and do not threaten the regime’’ (Esposito
1999:244). Thus, Muslim states may engage in uniquely high levels of repression
because of topdown processes, where the violent actions of rulers against society
are accepted as legitimate measures for restoring social order.
Muslim societies also lack separation of religion and state, which some suggest
is problematic (Fox 2006). According to them, the alliance between religion and
state is key to understanding the status of human rights in Muslim societies
(Mayer 1988). The dissent–repression cycle could occur when state-sponsored
religious dogma clashes with the rights of minorities. Religion is often formally
integrated into government policies, as religious laws regulate the personal
sphere, such as dress codes and dietary laws, and bans on ‘‘infidels’’ from hold-
ing certain political posts is a common feature (Price 1999:2). According to sys-
tematic data collected by Jonathan Fox and associates, ‘‘both Catholic and
Protestant states engage less in religious regulation,’’ while Muslim states are
identified as uniquely interventionist (Fox 2006:560). While all, or many, of the
features of Islam identified above may matter, we are not aware of any study that
has directly examined whether Muslim societies are any worse than states with
majorities of other believers in terms of predicting the level of state-sponsored
repression.7 If it is Islam, rather than other factors, that really matters, then
countries with larger shares of Muslims (and those dominated by Muslim popula-
tions) should correlate with the adverse outcomes many expect in the political
sphere.
The empirical literature on the effects of Islam on democracy confirms some
of the negative arguments. Apparently, Islamic attitudes toward women make
Muslims culturally non-conducive to modernization (Fish 2002; Karatnycky
2002). Donno and Russett (2004) found that repression of women’s rights is
indeed more widespread in Islamic countries, particularly in the Arab world.
Some find that Islam has an independent effect on lower levels of democracy
(Ross 2001; Fish 2002). However, Donno and Russett (2004) suggest that the
effect of culture has changed dramatically over time. Before 1980, predominantly
Catholic countries were ‘‘even less likely than Islamic ones to have democratic
governments, but this relationship subsequently turned strongly positive’’ (Don-
no and Russett 2004:583). They argue that rather than an effect of religion, it is
cultural factors peculiar to the Arab region that may matter.
Mayer (1993) warns that the Western tendency to believe that Islam is special
can ‘‘lead to a failure to recognize the common dimensions of human rights
issues in Muslim and non-Muslim societies’’ (Mayer 1993:117). An overemphasis
on Islam can obscure the mechanisms that explain human rights violations in
these societies, particularly if other factors more salient to human rights go

7
Some political scientists (Chehabi and Linz 1998) identify a unique brand of autocratic regime, which they
term ‘‘sultanistic.’’ Despite the reference to the Ottoman Empire and its heritage, the large majority of their cases
are Catholic countries in Latin America and Africa. Only Iran is presented as a ‘‘sultanistic’’ regime that contains a
majority of Muslims.
 INDRA DE SOYSA AND RAGNHILD NORDA˚S 931

unacknowledged. Following 9 ⁄ 11 and subsequent events, the tendency to focus

negatively on Islam has grown. Addressing this important issue systematically,
thus, is even more pertinent today, using a methodology that holds factors con-
stant across several cultural traits, so that Islam’s uniqueness for explaining
repression of human rights can be estimated more precisely. Given recent discus-
sions about the effects of religious fundamentalism and religious discrimination
on terrorism and violent conflict, testing how the share of a religious population
predicts repression is crucial for gaining further insight into how and why
religion features in violent political outcomes (Marty and Appleby 1995;
Reynal-Querol 2002; Hafez 2003; Fox 2004; Nordås 2004).

Methods, Variables and Data

Our dependent variable, the PTS (Stohl et al. 1984; Gibney and Dalton 1996;
Hafner-Burton 2005), captures negative rights, as it measures arbitrary physical
harm and state coercion with threat of harm against ordinary people.8 Generally,
there is high congruence among differently defined scales of repression (Land-
man 2005). We follow the conservative strategy of testing our models with both
the PTS and a measure of PIR (Cingranelli and Richards 1999), but we report
only tests using the PTS because of space limitations (see web Appendix for
results using PIR).
The PTS scale rates the yearly performance of countries on an interval scale of
5 categories (Hafner-Burton 2005:689).

1 If countries are under secure rule of law, political imprisonment and

torture are rare, and the political murders are extremely rare.
2 If imprisonment for non-violent political activities is limited, torture and
beating are exceptional, and political murder rare.
3 If political imprisonment is extensive, execution and political murder
may be common, and detention for political views are acceptable.
4 If the practices of level 3 are expanded to a larger segment of popula-
tion, murders, and disappearances are common, but terror affects pri-
marily those who interest themselves in political practices and ideas.
5 If level of terror are population wide, and decision makers do not limit
themselves by which they pursue private and ideological goals.

Our main independent variables are the percentages of the major religions that
make up a state. The variables are constructed as percentage of Protestant,
Catholic, Muslim, and other religions, for each country, summing to 100
(La Porta, Lopez-de-Silanes, Schleifer, and Vishny 1998). Secondly, we create
dichotomous measures to capture aspects of dominance of one or another
religion within a country. We create a Muslim-dominance dummy variable taking
the value 1, if Muslims are >80% of the population, and 0 if below.9 We created
similar dummy variables for Catholics, leaving out Protestant- and ‘‘other’’-domi-
nance as reference categories. The data on the share of the population Muslim
correlates at r = .98 with Fearon and Laitin’s (2003) measure of the size of the
Muslim population, which is reassuring. Thirdly, we test a dummy variable coded
according to membership in the OIC (see note 5). As expected, this variable
correlates at r = .83 with the share of the population Muslim (the first indicator).

8
In robustness checks, we also conduct all out analyses with the PIR data (Cingranelli and Richards 1999). The
correlation between the PTS and PIR is r = .82. The results are in the web Appendix.
9
We also test the dominance variables at the 70% and 90% thresholds respectively, and the results are the
same.
1 932 Islam’s Bloody Innards?

Controls
We control for several relevant factors gleaned from previous studies. Poe and Tate
(1994) and Poe, Tate, and Keith (1999) find that past levels of repression (the
human rights legacy), the level of formal democracy, population size, economic
standing, the threat of organized violence in the form of involvement in interna-
tional and civil wars, have significant effects on repression (Davenport and
Armstrong 2004; Poe 2004; Landman 2005). Civil wars are a real threat to leaders,
but the justification Poe and associates provide as to why international war matters
is somewhat vague. One possible explanation might be that the state leaders see an
increased need for instituting measures, such as states of emergency, out of con-
cerns over internal dissent, traitors, and saboteurs. However, external crises may
lead to ‘‘rally round the flag’’ effects when states shift their focus from domestic to
international threats. The empirical evidence on whether international war leads
to human rights repression is mixed (Richards, Gelleny, and Sacko 2001; Landman
2005). We include controls for both types of armed conflict—civil and interstate
war. Using the Uppsala ⁄ PRIO data set that includes all conflicts with at least 25
battle-related deaths (Gleditsch, Wallensteen, Eriksson, Sollenberg, and Strand
2002), we compute the number of years of peace since the last conflict and include
ongoing conflict coded as dummies. We do this for both types of conflict.10
Democracy is an important predictor of respect for human rights (Davenport
and Armstrong 2004; Landman 2005). The existence of an organized, legal, politi-
cal opposition, a free press, and civil society are constraints on leaders who may be
tempted to violate citizens’ rights. Established democracies contain norms of non-
violent means of conflict resolution, such as protest, strikes, and other civil actions
to make demands on government. Authorities responding to demands are con-
strained by laws of due process and political oversight. Some argue a curvilinear
effect of democracy on repression, where states at intermediate levels of democ-
racy are most likely to be repressive (Fein 1995). Davenport and Armstrong
(2004) find, however, that the effect of democracy is linear and negative, and most
significant when the democracy score is relatively high. Following these writers, as
well as others, we define our control for regime type as a dummy variable taking
the value 1 if the Polity IV (variable Polity 2) score is above 6 on the 11-point scale
(i.e., 7–10) and 0 if below. In our sensitivity analyses, we test the robustness of our
main variables by including democracy defined at various thresholds.
Whether or not countries are petroleum exporters is a crucial control variable
when assessing Islam’s effects on repression. As oil-wealthy countries are largely in
the Middle East and North Africa, tend to be undemocratic, and are largely Mus-
lim, we isolate religion’s effects from this potential confounding factor. We enter a
dummy variable that takes the value 1, if oil exports are <1 ⁄ 3 of Gross Domestic
Product (GDP) and 0 if not, taken from Fearon and Laitin’s (2003) replication
data set.11 Several find that natural resource wealth is a problem for governance
and peace (Auty 2001; de Soysa 2002). Resource wealth raises the stakes for state
capture and lowers the incentive of rulers to reform (Jensen and Wantchekon
2004; Acemoglu and Robinson 2006). We expect, therefore, that states that are
resource wealthy, such as oil-rich ones, will be more willing to resort to human
rights violations than those that are not (de Soysa and Binningsbø 2006). Despite a
correlation of only r = .30 between our oil-exporter dummy and the share of the

10
The peace years variables are defined as the number of years since the last conflict. The count is begun at
the time a country enters the system between 1946 and 2000. If a country enters the system with ongoing war, then
peace years would be 0, and if a country has never experienced conflict, peace years would be 54, and so forth. We
calculated the peace years using the Binary Time Series Cross Section (BTSCS) program working in Stata (Beck,
Katz, and Tucker 1998).
11
The data are available at: http://www.stanford.edu/~jfearon/
 INDRA DE SOYSA AND RAGNHILD NORDA˚S 933

population Muslim, we follow others who treat the question of religion and democ-
racy by controlling for oil wealth (Ross 2001). In sensitivity analyses, we also enter a
continuous measure of energy rents as a percentage of GDP (Bolt, Matete, and
Clemens 2002) to assess the strength of this relationship further.
Previous research has explicitly controlled for legal traditions of states. Appar-
ently, a British (common law) legal tradition that has independent bureaucracies
and court systems constrain arbitrary acts by government (Poe et al. 1999). Like-
wise, leftist governments are found to be kinder to human rights (Poe et al.
1999). As governments may go in and out of office, while repression scores are
relatively inert, we use a more permanent measure capturing a socialist legal tra-
dition—a measure that captures socialist law. Both measures are obtained from
La Porta et al. (1998).12 The exact reason for why socialist states are more likely
to respect human rights is unclear, except that they may be less likely to be chal-
lenged because of an ideology that emphasizes class equality and a unique capa-
city for social control. Yet, another argument could be that many of the socialist
states did not have open dissent out of fear of Soviet intervention.
Economic conditions can affect the likelihood of repression. Poe et al.
(1999:294) state that ‘‘in countries with economies characterized by scarcity,
regimes will be more likely to repress domestic threats.’’ Henderson (1991)
argues that in more developed states the population will be more content, and
thus less repression is needed to keep control. Richer states also have higher
capacity to deal with dissent through more organized channels, such as bureau-
cracies and public agencies (Fearon and Laitin 2003). Moreover, richer people
may have high opportunity costs for engaging in dangerous (violent) dissent that
threaten states (Collier and Hoeffler 2004). We use logged Gross National
Income per capita in PPP terms (Gleditsch 2002) and the growth rate of income
per capita to capture economic effects (World Bank 2005).
Studies find that countries with larger populations will be more repressive than
countries with smaller populations (Poe 2004; Landman 2005). Theoretically, a
large population implies larger numbers of potential dissenters and weaker state
capacity. A large population may also mean a larger geographical area, which
can be more difficult to control than a smaller area because of factors that affect
the technologies of insurgency, such as rough terrain (Fearon and Laitin 2003).
Poe et al. (1999:294) argue that ‘‘a large or dense population may increase the
occurrence of state terrorism by increasing the number of occasions on which
threats and coercive acts can occur,’’ and that having a large population also
affects repression because of the strain on available resources. We also control
for the degree of ethnic heterogeneity by a measure of ethnic fractionalization
taken from Fearon and Laitin (2003).13 The size of a population and the degree
of heterogeneity are correlated. As some argue, governance under conditions of
social heterogeneity is supposed to be difficult (Alesina, Baqir, and Easterly
1999; Easterly 2001). The results of ethnicity on repression, however, are mixed
(Walker and Poe 2002; Lee, Lindström, Moore, and Turan 2004). We also test
population density in our sensitivity analyses for gauging the robustness of the
basic model. The demographic data are from the World Bank (2005), expressed
as total population (logged) and people per square kilometer (logged).
Others find that a lagged dependent variable (LDV) is highly significant (Poe
2004). The LDV controls for time dependence and serial correlation, and it pre-
sumably captures effects of those factors omitted in the models (Landman

12
We might also have entered a term for Sharia law. However, Sharia is a major argument for why majority-Isla-
mic society might suffer repression. Thus, entering a separate term would mask any independent effect that Islam
may have on repression.
13
This variable indicates the likelihood that two individuals drawn at random from the population belong to
the same ethnic group.
1 934 Islam’s Bloody Innards?

2005). On the other hand, the LDV may soak up so much of the variance that it
masks potential causal factors explained by the other independent variables of
interest (Achen 2000; Plümper, Troeger, and Manow 2005). We will test our
models with and without LDVs.
In our sensitivity analyses, we test the effects of regional factors, as human
rights conditions, much like democracy, may cluster in space, and others report
that women’s rights issues are better predicted by regional factors rather than
religion (Donno and Russett 2004). Finally, we control for time trends in the
data. Human rights data may be affected over time by the increased sophistica-
tion of the technologies of detection. Some find that while the reporting of
violations by Amnesty International and the US State Department differed some-
what in early years, they have tended to converge in later years (Poe, Vazquez,
and Carey 2001). We enter year dummies to capture such potential time trends
in the data. Time dummies also take care of any unobserved factors, such as the
end of the Cold War, or global policy shifts that may affect human rights
through the process of diffusion and increased activism across the globe (Keck
and Sikkink 1998; Simmons and Elkins 2004).14
An examination of all of our variables did not show correlations that would
lead us to be concerned about multicollinearity (see correlations in Appendix).
As seen there, a regional dummy variable for the Middle East and North Africa
region (MENA) correlate at r = .64 with the share of the population Muslim.
Hence, we only use these variables together in our robustness checks. The next
highest correlation is that between income per capita and democracy (r = .51),
but as income and democracy are often used together and show statistical signifi-
cance independently, collinearity is probably not a factor.15 As seen in the table
in the Appendix, there is a positive but rather small bivariate relationship
between the share of the population Muslim and the level of repression
(r = .19).
We estimate the PTS scale using Ordered Probit analysis because the cut-off
points of this 5-point scale are easier to interpret.16 We use probit rather than
logit because the scale is very close to being normally distributed (Long 1997).
Additionally, time-series, cross-section data may contain complicated correlation
patterns because of the spatial and temporal nature of the data setup. With the
ordered probit models, we cluster the analysis on countries and estimate the
Huber-White corrected robust standard errors, a method which is robust to
heteroskedasticity and serial correlation (Wiggins 1999). By clustering on country,
we assume that the data is correlated within clusters but not across. The summary
statistics of each of the variables are described in the Appendix. The estimates are
based on data for 141 countries with over 1 million inhabitants covering the per-
iod 1980–2000. The data set is unbalanced (i.e., the number of countries is not
the same for each of the years), and the availability of data for all of the controls
determined the size of the data set (see Appendix for country list).

Results
In Table 1, column 1, our baseline model confirms many of the findings of pre-
vious studies. The high levels of significance reflect the fact that we did indeed

14
We could also have controlled for how committed a country is to the norms of human rights by adding a var-
iable measuring whether countries participate in human rights conventions, but the existing evidence suggests that
this variable does not matter (Keith 1999; Neumayer 2005).
15
Nevertheless, we examine variance inflation factor (VIF) scores in the sensitivity analyses (see web
Appendix).
16
We use the statistical software package stata version 9 in all estimations. The data and do files containing
all the results will be available at http://www.isanet.org/data_archive.html. The estimations using the PIR scale and
other diagnostics will be presented in a web Appendix.
 TABLE 1. Ordered Probit Estimates of Religion and Political Terror, 1980–2000. Dependent variable: PTS

(1) (2) (3) (4) (5)

Oil exporter 0.553*** (3.67) 0.582*** (3.84) 0.323*** (3.81) 0.583*** (3.87) 0.589*** (3.83)
GNI per capita )0.517*** (6.83) )0.515*** (6.77) )0.293*** (6.76) )0.524*** (6.89) )0.535*** (7.24)
Growth per capita )0.006 (1.19) )0.007 (1.30) )0.015*** (2.89) )0.007 (1.21) )0.006 (1.07)
Democracy )0.869*** (7.85) )0.953*** (8.30) )0.564*** (7.20) )0.924*** (7.97) )1.004*** (8.85)
British legal system )0.260* (1.92) )0.136 (0.91) )0.017 (0.19) )0.200 (1.37) )0.105 (0.74)
Socialist legal system )0.207 (1.30) )0.104 (0.49) )0.026 (0.22) )0.146 (0.69) )0.063 (0.33)
Ethnic fractionation )0.665** (2.55) )0.636** (2.35) )0.360** (2.41) )0.532* (1.95) )0.565** (2.04)
INDRA

Population sizea 0.241*** (5.67) 0.228*** (5.50) 0.115*** (5.00) 0.224*** (5.30) 0.231*** (5.70)
DE

Civil war 1.034*** (7.50) 1.035*** (7.67) 0.718*** (7.38) 1.024*** (7.73) 1.035*** (7.66)
Civil peace years )0.025*** (5.55) )0.025*** (5.45) )0.009*** (3.32) )0.024*** (5.30) )0.024*** (5.52)
SOYSA

Interstate war 0.469** (2.23) 0.511** (2.38) 0.252** (2.01) 0.469** (2.25) 0.489** (2.29)
Interstate peace years 0.014*** (3.27) 0.010** (2.23) 0.004* (1.75) 0.011*** (2.58) 0.010** (2.26)
AND

Muslim (%) )0.005** (2.29) )0.003** (2.25)

Protestant (%) )0.008* (1.82) )0.006** (2.15) )0.007* (1.67)
Other (%) )0.004 (1.40) )0.002 (1.31) )0.003 (0.98)
OIC member )0.322** (2.05)
Political terrorb 1.225*** (19.96)
Muslim dominancec 0.006 (0.03)
Catholic dominancec 0.567*** (2.93)
Wald chi-square 616.6 595.6 1400.9 623.3 623.8
RAGNHILD NORDA˚S

p > chi-square 0.00001 0.00001 0.00001 0.00001 0.00001

No. countries 141 140 140 140 141
Observations 2,576 2,567 2,541 2,567 2,576

Robust z statistics in parentheses. Two-tailed tests. Due to the lagged dependent variable, Model 3 covers the period 1981–2000.
a
Log-transformed.
b
Lagged 1 year.
c
Population share ‡80%.
Significant at *10%, **5%, ***1%.
935
1 936 Islam’s Bloody Innards?

choose to include those variables others have also found to be robustly related
to repression.
Most of the results, except for the socialist legal system and the growth rate of
income, are significant in the baseline model.17 Both war variables are positive
and significant for predicting repression, but the years since the previous inter-
national war shows a positive effect. This suggests that countries with less repres-
sion have shorter periods of international peace, indicating perhaps that it is the
wealthy, democratic states that have banded together in collective security in
recent years, such as in the Persian Gulf and the Balkans. Interestingly, oil
export dependence, one of the key control variables, predicts higher levels of
repression, consistent with various theories of the ‘‘resource curse.’’
In column 2, we add the crucial independent variables. As seen there, the
share of the population Muslim has a negative and statistically significant effect,
relative to the share of the population Catholic, our reference category. In col-
umn 3, the result upholds when we enter a LDV, suggesting that this result is
quite robust to specification. In order to assess the substantive effect of this sig-
nificant result, we compute a baseline predicted probability of reaching level 4
and above of the dependent variable holding all the independent variables at
their mean values. We use PTS >3 because this is the point at which ‘‘large seg-
ments of the politically active population are affected by political imprisonment
and political murder’’ (see Methods above). Next, we vary each of our variables
of interest to its maximum value, holding the others at their means, to see how
this change compares with the baseline prediction. While holding all the vari-
ables at their means and raising the share of the population Muslim to its maxi-
mum value (99.8%) lowers the baseline predicted probability of reaching level 4
or above of political terror by 56%.
In column 4, we test the effect of membership in the OIC. As seen there,
being a member of the OIC predicts lower levels of repression. In column 5, we
include the religious dominance variables, defined at the 80% and above thres-
hold. Catholic dominance predicts higher levels of repression, a result that is sta-
tistically highly significant. Muslim dominance had no discernible effect on
higher repression relative to Catholics and others.
In summary, we find that, relative to countries with larger shares of Catholics,
countries with larger shares of Muslims, and those demographically dominated
by Muslims suffer lower levels of political repression. Moreover, membership in
the OIC, too, predicts lower political repression relative to everyone else. It
seems that Islam relative to Catholicism is a poor predictor of ‘‘bloody innards.’’
Interestingly, however, examining the substantive impacts of our political and
economic variables, relative to the religious ones, is illustrative. Holding all vari-
ables at their mean values, going from a non-democracy to a democracy lowers
the baseline predicted probability of level 4 and above of the PTS by 70%, while
a similar movement from no civil war (0) to civil war (1) increases the baseline
predicted probability by 280%—roughly five times the impact at the maximum
value of the share of Muslims. Raising per capita income to the maximum value
lowers the baseline predicted probability by 94%, again, slightly less than double
the impact of the religious variables. Notably, changing from being an oil expor-
ter (1) to being a non-oil exporter (0) lowers the baseline risk of repression at 4
and above on the PTS by 140%, which is roughly three times the impact of a
maximum change of the share of the population Muslim. Public and scholarly
discussion seems to be wrong about the uniqueness of Islam for predicting levels
of dissent and repression, and they may have overemphasized religion over
other, more important, factors based on political economy.

17
The effect of having a British legal system is statistically significant only at the 10% level in the baseline
model, but it is fragile.
 INDRA DE SOYSA AND RAGNHILD NORDA˚S 937

Robustness Checks
We conduct a series of sensitivity analyses to assess the robustness of our findings
(results in web Appendix).18 First, we test other thresholds for religious domi-
nance by running our models at the 70% and 90% share of the population
respectively. The basic results uphold. Next, we add regional dummies, leaving
out the ‘‘Western’’ region, which includes North America, Europe, Australia,
and New Zealand.19 The religion variables lose significance when the regions are
added, although the share Islam remains negative and barely misses significance
at the 10% level. Latin America, as a region, has the strongest impact of all, fol-
lowed by the MENA region. These results are similar to those reported by others
(Landman 2005). It is interesting that, of all the religions, the share of the popu-
lation Muslim retains its sign and remains close to significance. Thus, the effect
of Muslim societies becomes strong even against the share of Protestants, when
regional dummies are included. Clearly, it is the regional effects of the MENA
region, not the share of the population Muslim, that helps us understand the
nature of human rights violations, a point also made about women’s rights (Donno
and Russett 2004). Yet, the fact that the MENA region’s impact is not hugely
different from the others is heartening. While the level of human rights, like
democracy, may exhibit neighborhood effects, more sophisticated analyses are
warranted to examine specifically how levels of human rights may cluster in
space.
Given the discussions around Islam and violent conflict, we assess whether the
effect of the share of Muslim relative to Catholics changes when dropping the
conflict variables. Dropping civil and international war has little effect on the
basic results. Dropping oil, too, made only a small difference. We tested a contin-
uous measure of energy rents to GDP (Bolt et al. 2002), as it could be argued
that our choice of oil exports above one-third of GDP is arbitrary. This variable
is also positively and highly significantly related to the PTS.20 Despite this highly
significant result for the energy wealth variable, the share of the population Mus-
lim continues to show a negative and statistically significant effect on political
terror.
The significant effect of Islam is affected if democracy is dropped from the
models entirely. When we drop the democracy dummy variable, the statisti-
cally significant effect of the share of the Muslim population vanishes, suggesting
that the relative effect of Catholicism might be driven largely by majority-
Catholic states that have failed to democratize. However, dropping democracy
from the model is theoretically questionable, as regime type has such a strong
theoretical and empirical effect on levels of political repression. We tried
democracy at several different cut-offs on the Polity2 scale. Adding a dummy,
defining democracy above 8, made the Protestant variable insignificant, while
Islam retained its statistically significant effect, proving once again its robust-
ness. The reported effects did not change when defining democracy at
the low end of 3 and above, nor when adding the Polity2 scale without
dichotomizing.
Additionally, we ran all the reported models with the PIR data using linear
estimating techniques. Our basic findings uphold (see web Appendix). We also
conducted the usual diagnostics using the regression–cluster method. There was
little sign of multicollinearity, as the mean VIF scores were always <2, and none

18
The results are not published because of considerations of space, but they can be generated with the do files
made available with the replication data and as a web Appendix (see above). We use both the PTS and PIR in the
sensitivity tests.
19
The regional dummies are taken from the Fearon and Laitin (2003) replication data.
20
Logged and non-logged terms yield similar results.
1 938 Islam’s Bloody Innards?

of the individual variables reached the problematic score of 10 (Hamilton 1992;

Stata Corporation 2003). As heteroskedasticity consistent robust-standard errors
are obtained when clustering on country (Stata Corporation 2003), we ran the
models without clustering and examined the residuals. The standardized residu-
als did not depart from normality. We also checked Cook’s D values to eliminate
problems of influential observations. Running the model without several data
points with values above the cut-off of 4 ⁄ n made little difference to the results
reported above.

Conclusion
Popular opinion, high political discourse, and several segments of scholarly
discussion in the social sciences point to the uniqueness of Islam, often in
rather disparaging terms. Discussion on the war against Al Qaeda, immigra-
tion-related problems, Middle Eastern wars, and the question of Turkey’s pos-
sible entry into the European Union are laced with concerns about Islam’s
propensity for breeding religious fanaticism. Islam is apparently antithetical to
individual rights and the respect for personal liberties. We also address a par-
allel debate about whether or not culture, rather than strategic behavior, mat-
ters when explaining political life. The policy import of this issue is
enormous—if religion is a major driver of repression in a society, then there
is little room for intervention from a policy perspective, and the process of
change, if it ever gets in motion, could span generations. Fortunately, our
estimates show that there is a lot of room for policy that could have more
immediate effects. Compared with others, Islam is not associated with greater
levels of repression, and religious affiliation is not a strong predictor of politi-
cal terror.
We have only examined religion’s effect on a particular variety of human
rights violations. Future studies may also look at other aspects of human rights,
such as economic and gender rights. Case-study based research should examine
the institutional bases of social control and interest mediation in Islamic socie-
ties. Why such factors might be weaker in majority-Catholic countries from a
comparative perspective is also interesting. Questions of the quality of institu-
tions, and institutional factors relating to colonial heritage based on recent work
on European settler patterns, may also be extended fruitfully to human rights
studies (Acemoglu, Johnson, and Robinson 2001). Our empirical analyses clearly
support those who argue that there is little in Islam that militates against the
possibility of decent governance, for political compromise, and institutionalized
political competition, as many others have also argued (Mayer 1993; Tessler
2002; Hafez 2003; Nasr 2005).
Apparently, one of the strongest predictors of why people are intolerant
comes from real and perceived ‘‘threats to a way of life’’ (Gibson 2006). The
current discussion on Islam contains such a tone. Our results certainly ques-
tion the utility of framing issues around religious ⁄ cultural generalities. Policies
based on the fear of Islam may itself lead to the violation of human rights
and undermine democracy, particularly when Muslim immigrants are treated
differently, when Muslims applicants for asylum and sanctuary face added bar-
riers, and people face unfair treatment at the hands of governments simply
because they profess a particular faith. As recent reports indicate, based on
the fear of Islam, the actions of Russian authorities against some Muslim
regions mirror the old Soviet Union, which may be counterproductive to peace
and justice (Myers 2005). Our empirical results refute those who view Muslim
societies as containing uniquely ‘‘bloody innards’’ relative to the other major
religions.
 INDRA DE SOYSA AND RAGNHILD NORDA˚S 939

Appendix
Correlation matrix of selected variables
OIC Muslim (%) Prot. (%) Cath. (%) Other (%) MENA Democ Income ⁄ pc Oil exp.

PTS 0.20 0.18 )0.28 )0.02 )0.01 0.08 )0.42 )0.46 0.13
OIC – 0.83 )0.33 )0.45 )0.24 0.43 )0.46 )0.27 0.28
Muslim (%) – )0.35 )0.51 )0.35 0.64 )0.40 )0.16 0.30
Prot. (%) – )0.10 )0.12 )0.23 0.26 0.21 )0.04
Cath. (%) – )0.47 )0.32 0.32 0.14 )0.05
Other (%) – )0.22 )0.06 )0.11 )0.27
MENA – )0.19 0.21 0.39
Democ. – 0.51 )0.18
GNI ⁄ pc – 0.12
Oil Exp. –
Summary statistic
Variable Observation Mean SD Min. Max.

Political terror scale 2,567 2.4 1.1 1 5

Muslim (%) 2,567 24.7 36.1 0 99.8
Protestant (%) 2,567 12.1 20.2 0 97.8
Catholic (%) 2,567 30.3 35.4 0 96.9
Other (%) 2,567 32.8 31.8 0.2 100
Oil exporter 2,567 0.16 0.4 0 1
GNI per capita 2,567 8.1 1.1 5.6 10.4
Growth per capita 2,567 0.64 6.0 )52.1 37.1
Democracy 2,567 0.43 0.5 0 1
British legal system 2,567 0.33 0.47 0 1
Socialist legal system 2,567 0.12 0.32 0 1
Ethnic fractionalization 2,567 0.42 0.28 0.004 0.93
Population size 2,567 16.1 1.5 12.8 30
Civil war 2,567 0.19 0.39 0 1
Civil peace years 2,567 20.3 17.1 0 54
Interstate war 2,567 0.04 0.19 0 1
Interstate peace years 2,567 24.2 15.4 0 54

List of countries:
Western Hemisphere—United States, Canada, Cuba, Haiti, Dominican Repub-
lic, Jamaica, Trinidad & Tobago, Mexico, Guatemala, Honduras, Nicaragua,
Costa Rica, Panama, Colombia, Venezuela, Guyana, Ecuador, Peru, Brazil, Boli-
via, Paraguay, Chile, Argentina, Uruguay.
Europe & Central Asia—The United Kingdom, Ireland, the Netherlands, Bel-
gium, France, Switzerland, Spain, Finland, Sweden, Norway, Denmark, Portugal,
Poland, Austria, Hungary, Italy, Albania, Greece, Cyprus, Bulgaria, Moldova,
Romania, Russian Federation, Estonia, Latvia, Lithuania, Ukraine, Belarus, Arme-
nia, Georgia, Azerbaijan, Turkmenistan, Tajikistan, Kyrgyz Republic, Uzbekistan,
Kazakhstan.
Sub-Saharan Africa—Guinea Bissau, Gambia, Mali, Senegal, Benin, Mauritania,
Niger, Cote d’Ivoire, Guinea, Burkina Faso, Liberia, Sierra Leone, Ghana, Togo,
Cameroon, Nigeria, Gabon, Central African Republic, Chad, Congo Rep., Congo
Dem. Rep., Uganda, Kenya, Tanzania, Burundi, Djibouti, Somalia, Rwanda, Ethi-
opia, Angola, Mozambique, Zambia, Zimbabwe, Malawi, South Africa, Namibia,
Lesotho, Botswana, Swaziland, Madagascar, Mauritius.
1 940 Islam’s Bloody Innards?

North Africa & Middle East—Morocco, Algeria, Tunisia, Sudan, Libya, Iran,
Turkey, Egypt, Syria, Jordan, Iraq, Israel, Oman, Saudi Arabia, Kuwait, Bahrain,
United Arab Emirates.
Asia—Afghanistan, China, Mongolia, South Korea, Japan, India, Bhutan, Paki-
stan, Bangladesh, Sri Lanka, Myanmar, Nepal, Thailand, Cambodia, Laos, Malay-
sia, Singapore, Philippines, Indonesia.
Oceania—Australia, Papua New Guinea, New Zealand, Fiji (n = 141).

References
Acemoglu, Daron, and James A. Robinson. (2006) Economic Backwardness in Political Perspective.
American Political Science Review 100(1): 115–131.
Acemoglu, Daron, Simon Johnson, and James A. Robinson. (2001) The Colonial Origins of
Comparative Development: An Empirical Investigation. American Economic Review 91(5): 1369–
1401.
Achen, Christopher. (2000) Why Lagged Dependent Variables Can Suppress the Explanatory
Power of Other Independent Variables. Paper presented at the annual meeting of the Political
Methodology Section of the American Political Science Association, UCLA, July 20–22.
Alesina, Alberto, Reza Baqir, and William Easterly. (1999) Public Goods and Ethnic Divisions.
Quarterly Journal of Economics 114(4): 1243–1284.
Auty, Richard M., ed. (2001) Resource Abundance and Economic Development. Oxford: Oxford Univer-
sity Press.
Barber, Benjamin R. (1995) Jihad vs McWorld. New York: Ballantine Books.
Beck, Nathaniel, Jonathan N. Katz, and Richard Tucker. (1998) Taking Time Seriously: Time-
Series-Cross-Section Analysis With a Binary Dependent Variable. American Journal of Political
Science 42(4): 1260–1288.
Beunderman, Mark. (2006) ‘Islamic Terrorism’ Phrase to be Banned From EU Lexicon. Available at
http://www.euobserver.com. (Accessed April 13, 2006)
Bolt, Katherine, Mampite Matete, and Michael Clemens. (2002) Manual for Calculating Adjusted
Net Savings. Washington, D.C.: World Bank, Environment Department.
Booth, John A., and Mitchell A. Seligson. (2004) Paths to Democracy and the Political Culture
of Costa Rica, Mexico, and Nicaragua. In Political Culture and Democracy in the Developing Countries,
edited by Larry Diamond. Boulder, CO: Lynne Rienner.
Chehabi, Houchang E., and Juan Linz, EDS. (1998) Sultanistic Regimes. Baltimore, MD: Johns Hop-
4 kins University Press.
Cingranelli, David L., and David L. Richards. (1999) Respect for Human Rights After the End of
the Cold War. Journal of Peace Research 36(5): 511–534.
Collier, Paul, and Anke Hoeffler. (2004) Greed and Grievance in Civil War. Oxford Economic
Papers 56(4): 563–595.
Davenport, Christian, and David A. Armstrong II. (2004) Democracy and the Violation of
Human Rights: A Statistical Analysis From 1976 to 1996. American Journal of Political Science
48(3): 538–554.
de Soysa, Indra. (2002) Paradise Is a Bazaar? Greed, Creed, and Governance in Civil War, 1989–
1999. Journal of Peace Research 39(4): 395–416.
de Soysa, Indra, and Helga Malmin Binningsbø. (2006) Devil’s Excrement or Social Cement? Oil
Wealth and Repression, 1980–2001. Paper presented at the International Studies Association
Annual Meeting, San Diego, CA, March 21–27.
Diamond, Larry, ed. (1994) Political Culture and Democracy in Developing Countries. Boulder, CO:
Lynne Rienner.
Donno, Daniela, and Bruce Russett. (2004) Islam, Authoritarianism, and Female Empowerment:
What Are the Linkages? World Politics 56(4): 582–607.
Easterly, William. (2001) Can Institutions Resolve Ethnic Conflict? Economic Development and Cul-
tural Change 49(4): 687–706.
Esposito, John L. (1999) The Islamic Threat: Myth or Reality? New York: Oxford University Press.
Fearon, James D., and David D. Laitin. (2003) Ethnicity, Insurgency, and Civil War. American Politi-
cal Science Review 97(1): 1–16.
Fein, Helen. (1995) More Murder in the Middle: Life-Integrity Violations and Democracy in the
World, 1987. Human Rights Quarterly 17(1): 170–191.
Fish, Steven M. (2002) Islam and Authoritarianism. World Politics 55: 4–37.
 INDRA DE SOYSA AND RAGNHILD NORDA˚S 941

Fox, Jonathan. (2000) Is Islam More Conflict Prone Than Other Religions? A Cross-Sectional Study
of Ethno-Religious Conflict. Nationalism and Ethnic Politics 6(2): 1–23.
Fox, Jonathan. (2004) The Rise of Religious Nationalism and Conflict: Ethnic Conflict and Revolu-
tionary Wars, 1945–2001. Journal of Peace Research 41(6): 715–731.
Fox, Jonathan. (2006) World Separation of Religion and State Into the 21st Century. Comparative
Political Studies 39(5): 537–569.
Friedman, Thomas L. (2005) ‘Why Us?’ Sunni Arabs Should Ask, ‘Why Anyone?’ New York Times,
November 16.
Gartner, Scott Sigmund, and Patrick M. Regan. (1996) Threat and Repression: The Non-Linear
Relationship Between Government and Opposition Violence. Journal of Peace Research 33(3):
273–287.
Gibney, Mark, and Matthew Dalton. (1996) The Political Terror Scale. In Human Rights and Devel-
oping Countries, edited by David Cingranelli. Greenwich, CT: JAI press.
Gibson, James L. (2006) Enigmas of Intolerance: Fifty Years After Stouffer’s Communism, Confor-
mity, and Civil Liberties. Perspectives on Politics 4(1): 21–34.
Gleditsch, Kristian Skrede. (2002) Expanded Trade and GDP Data. Journal of Conflict Resolution
46(5): 712–724.
Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and
Håvard Strand. (2002) Armed Conflict 1946–2001: A New Dataset. Journal of Peace Research
5 39(5): 615–637.
Gurr, Ted Robert. (1986) The Political Origins of State Violence and Terror: A Theoretical Analy-
sis. In Government Violence and Repression: An Agenda for Research, edited by Michael Stohl, and
G. A. Lopez. New York: Greenwood.
Hafez, Mohammed M. (2003) Why Muslims Rebel: Repression and Resistance in the Islamic World.
London: Lynne Rienner.
Hafner-Burton, Emilie M. (2005) Right or Robust? The Sensitive Nature of Repression to Global-
ization. Journal of Peace Research 42(6): 679–698.
Hamilton, Lawrence C. (1992) Regression With Graphics. Belmont, CA: Wadsworth.
Harrison, Lawrence, and Samuel P. Huntington, eds. (2000) Culture Matters: How Values Shape
Human Progress. New York: Basic Books.
Hasenclever, Andreas, and Volker Rittberger. (2000) Does Religion Make a Difference? Theo-
retical Approaches to the Impact of Faith on Political Conflict. Millennium: Journal of Interna-
tional Studies 29(3): 641–674.
Henderson, Conway E. (1991) Conditions Affecting the Use of Political Repression. Journal of
Conflict Resolution 35(1): 120–142.
Holmes, Stephen. (2005) Al Qaeda, September 11, 2001. In Making Sense of Suicide Missions, edited
by Diego Gambetta. Oxford: Oxford University Press.
Huntington, Samuel P. (1996) The Clash of Civilizations and the Remaking of World Order. New York:
Touchstone Books.
Hutchinson, John. (2005) Nations as Zones of Conflict. London: Sage.
Jackman, Robert W., and Ross A. Miller. (1998) Social Capital and Politics. Annual Review of Politi-
cal Science 1: 47–73.
Jensen, Nathan, and Leonard Wantchekon. (2004) Resource Wealth and Political Regimes in
Africa. Comparative Political Studies 37(7): 816–841.
Juergensmeyer, Mark. (2000) Terror in the Mind of God: The Global Rise of Religious Violence. Berkeley,
CA: University of California Press.
Karatnycky, Adrian. (2002) Muslim Countries and the Democracy Gap. Journal of Democracy 13(1):
99–112.
Keck, Margaret E., and Kathryn Sikkink. (1998) Activists Beyond Borders: Advocacy Networks in Inter-
national Politics. Ithaca, NY: Cornell University Press.
Keith, Linda C. (1999) The United Nations International Convention on Civil and Political Rights:
Does It Make a Difference in Human Rights Behavior? Journal of Peace Research 36(1): 95–118.
La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Schleifer, and Robert Vishny. (1998)
The Quality of Government. Cambridge, MA: NBER. Working paper no. 6727.
Landes, David S. (1999) The Wealth and Poverty of Nations: Why Some Are So Rich and Some So Poor. New
York: W. W. Norton.
Landman, Todd. (2005) Protecting Human Rights: A Comparative Study. Washington, DC: Georgetown
University Press.
Lee, Chris, Ronny Lindström, Will H. Moore, and Kürsad Turan. (2004) Ethnicity and Repres-
sion: The Ethnic Composition of Countries and Human Rights Violations. In Understanding
1 942 Islam’s Bloody Innards?

Human Rights Violations: New Systematic Studies, edited by Sabine C. Carey, and Steven C. Poe.
Aldershot: Ashgate.
Lewis, Bernard. (2003) What Went Wrong? The Clash Between Islam and Modernity in the Middle East.
New York: Perennial.
Long, Scott J. (1997) Regression Models for Categorical and Limited Dependent Variables. London: Sage.
Marty, Martin E., and R. Scott Appleby, eds. (1995) Fundamentalism Comprehended. Chicago, IL:
University of Chicago Press.
Mayer, Ann Elizabeth. (1988) The Dilemmas of Islamic Identity. In Human Rights and the World’s
Religions, edited by S. R. Leroy. South Bend, IN: University of Notre Dame.
Mayer, Ann Elizabeth. (1993) Islam and Human Rights: Different Issues, Different Contexts. Les-
sons From Comparisons. In Islamic Law Reform and Human Rights. Challenges and Rejoinders, edi-
ted by Tore Lindholm, and Kari Vogt. Oslo: Nordic Human Rights Publications.
McTernan, Oliver. (2003) Violence in God’s Name: Religion in an Age of Conflict. Maryknoll, NY: Orbis.
Moore, Will H. (2000) The Repression of Dissent: A Substitution Model of Government Coercion.
Journal of Conflict Resolution 44(1): 107–127.
Myers, Steven Lee. (2005) Growth of Islam in Russia Brings Soviet Response. New York Times,
November 22.
Nasr, Vali. (2005) The Rise of ‘Muslim Democracy.’ Journal of Democracy 16(2): 13–27.
Neumayer, Eric. (2005) Do International Human Rights Treaties Improve Respect for Human
Rights? Journal of Conflict Resolution 49(6): 925–953.
Nordås, Ragnhild. (2004) Regulating Religious Minorities: For Better or Worse? Paper presented at
the International Studies Association Annual Convention, Montreal, Canada, March 17–20.
Norris, Pippa, and Ronald Inglehart. (2004) Sacred and Secular. Religion and Politics Worldwide.
Cambridge: Cambridge University Press.
North, Douglass C. (1990) Institutions, Institutional Change and Economic Performance. Cambridge:
Cambridge University Press.
Pipes, Daniel. (2003) Militant Islam Reaches America. New York: W.W. Norton.
Plümper, Thomas, Vera Troeger, and Philip Manow. (2005) Panel Data Analysis in Comparative
Politics: Linking Method to Theory. European Journal of Political Research 44(2): 327–354.
Poe, Steven C. (2004) The Decision to Repress: An Integrative Theoretical Approach to the
Research on Human Rights and Repression. In Understanding Human Rights Violations: New
Systematic Studies, edited by Sabine C. Carey, and Steven C. Poe. Aldershot: Ashgate.
Poe, Steven C., and Neal Tate. (1994) Repression of Human Rights to Personal Integrity in the
1980s: A Global Analysis. American Political Science Review 88(4): 853–872.
Poe, Steven C., Neal Tate, and Linda C. Keith. (1999) Repression of the Human Rights to Per-
sonal Integrity Revisited: A Global Cross-National Study Covering the Years 1976–1993. Interna-
tional Studies Quarterly 43(2): 291–313.
Poe, Steven C., Tanya C. Vazquez, and Sabine C. Carey. (2001) How Are These Pictures Different?
A Quantitative Comparison of the U.S. State Department and Amnesty International Human
Rights Reports, 1976–1995. Human Rights Quarterly 23(3): 650–677.
Price, Daniel E. (1999) Islamic Political Culture, Democracy and Human Rights: A Comparative Study.
Westwood, CT: Greenwood.
Reynal-Querol, Marta. (2002) Ethnicity, Political Systems, and Civil Wars. Journal of Conflict Resolu-
tion 46(1): 29–54.
Richards, David L., Ronald D. Gelleny, and David H. Sacko. (2001) Money With a Mean Streak?
Foreign Economic Penetration and Government Respect for Human Rights in Developing
Countries. International Studies Quarterly 45(2): 219–239.
Ross, Michael. (2001) Does Oil Hinder Democracy? World Politics 53: 325–361.
Russett, Bruce, John R. Oneal, and Michealene Cox. (2000) Clash of Civilizations, or Realism
and Liberalism Déjà Vu? Some Evidence. Journal of Peace Research 37(5): 583–608.
Simmons, Beth A., and Zachary Elkins. (2004) The Globalization of Liberalization: Policy Diffu-
sion in the International Political Economy. American Political Science Review 98(1): 171–189.
Smock, David. (2002) Islam and Democracy. Washington, DC: US Institute of Peace.
Stata Corporation. (2003) Cross-Sectional Time-Series: Reference Manual Release 8. College Station, TX:
Stata Corporation.
Stohl, Michael, David Carleton, and Steven E. Johnson. (1984) Human Rights and U.S. For-
eign Assistance From Nixon to Carter. Journal of Peace Research 3(1): 1–11.
Tessler, Mark. (2002) Do Islamic Orientations Influence Attitudes Toward Democracy in the Arab
World? Evidence From Egypt, Jordan, Morocco, and Algeria. International Journal of Comparative
Sociology 43(3–5): 229–249.
 INDRA DE SOYSA AND RAGNHILD NORDA˚S 943

Walker, Scott, and Steven C. Poe. (2002) Does Cultural Diversity Affect Countries’ Respect for
Human Rights? Human Rights Quarterly 24(1): 237–263.
Wiggins, Vince. (1999) Comparing XTGLS with Regress Cluster. Stata Corporation. Available at
www.stata.com/support/faqs/stat/xtgls_rob.html. (Accessed September 17, 2007).
World Bank. (2005) World Development Indicators, CD-rom. Washington, DC: World Bank.