Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
1
The divergent impact of Islamic religiosity on ethno-religion social tolerance
in countries from the Middle East and North Africa (2001-2014)
Niels Spierings
Radboud University
Paper prepared for the ECPR joint session 2017 – v 20170328
Please do not cite without permission of the author
Introduction
Recent work shows that the Arab Uprisings and their aftermath dealt a heavy blow to ethno-religious
social tolerance across the region (Spierings, 2017). This is worrisome not just because of the decline
in tolerance itself, but also because a many have theorized and shown the importance of tolerance
for sustainable democracies (Lipset, 1959, Mill, 2007; Putnam, Leonardi & Nanetti, 1994; Rawls,
2003; Sullivan & Transue, 1999), also for the MENA particularly (see Ciftci, 2010; Rizzo, Abdel-Latif &
Meyer, 2007; Sarkissian, 2012; Tessler, 2002 - IJCS).
As a cause for democratic support, Islam and Islamic religiosity has gotten much scholarly
attention (Fish, 2002; 2011; Hoffman, 2004; Jamal & Tessler, 2008; Moaddel, 2002; Posusney, 2004;
Rizzo, Abdel-Latif & Meyer, 2007; Spierings, Smits & Verloo, 2009; Spierings, 2014; Stepan &
Robertson, 2003; Tessler, 2002); however, in explaining ethno-religious tolerance in the MENA, the
impact of Islamic religiosity is largely absent from the literature. Consequently we do not know
whether insights from the Western-based literature on the religiosity-tolerance linkage (e.g. Allport
& Ross, 1967; Bloom & Arikan, 2012; Cigler & Joslyn, 2002; Scheepers, Gijsberts & Hello, 2002) is
generalizable to the MENA context as well, let alone that we understand how the impact of
religiosity differs across the MENA, given the substantial differences in non-democratic regime types,
state-religion relations, and degrees of religious conflict and fractionalization. In the words of
Ceobanu & Escandell in their annual review of sociology overview study: we need to enlarge the
analytical focus of the field and “research has yet to provide a clear picture of the institutional and
sociopolitical macro-level factor that affect [these attitudes]” (2010: 310).
Nevertheless religion can be expected be important in the MENA. Based on in-depth
interviews in Oman Al Sadi & Basit conclude that youngster “use religion to define their relations
with, and determine their tolerance of, those who hold different religious beliefs.” (2013:447)
Moreover, the only comparative public opinion study so far finds that religious identification has
2
different effects in five countries (2001-2007), but finds few effects of doctrinal orthodoxy (after
control for a measure that taps political tolerance) (Spierings, 2014).1 Neither of these studies clearly
distinguished between the different behavioural and attitudinal dimension of religions that are at the
core of the existing studies of religion in political sociology (e.g. Stark & Glock, 1968; Kellstedt, Green,
Guth & Smidt, 1996; Scheepers, Gijsberts & Hello, 2002) and theorize how these might impact social
ethno-religious tolerance differently. Moreover, these studies provide no explanations for why and
how the impact of religiosity might differ between MENA countries and years.
In this study, I will therefore follow the 3B approach in studying the impact of religiosity on
social tolerance, by distinguishing religious belonging, beliefs and behaviour, and I will theorize the
impact of these three dimension in the MENA context, acknowledging not only the differences in
context between the MENA and the West, but also the differences among MENA countries, in order
to avoid the Orientalist trap of simplifying and essentializing the MENA region (Owen, 2004; Said,
1978; Spierings, 2015). As such this study (a) provides unique empirical analyses of the applicability
of existing theories linking religiosity to social tolerance to the MENA region, (b) helps to further
translate and refine our theoretical understanding of this linking to different a different socio-
political context and religion, and (c) gives insight into the degree to which these mechanisms are at
work in the same way across the MENA region and explores theoretical explanations for possible
differences among MENA countries.
These contributions to the existing literature are based on the analyses of 32 synchronized
Arab Barometer and World Value surveys for 13 countries, with the years ranging from 2001 to 2014.
These data will first be analysed using pooled multilevel regression models to estimate the general
relationships between religiosity dimensions and social tolerance, which help to assess the unique
impact of each dimension of religiosity across the MENA. Next country-disaggregated OLS regression
models and multilevel models with cross-level interactions are estimated to assess the internal
generalizability of religiosity’s impact among MENA countries and explain the differences in this
impact.
Theoretical background and expectations
Ethno-religious social tolerance
1 In her online master’s thesis Siegel (2015) does focus on explaining social ethno-religious tolerance, but does not including religiosity. Also Sarkissian’s unpublished APSA paper (2011) focusses on explaining tolerance in the Arab MENA, but it has political, not social tolerance as explanandum; moreover the two religion indicators as explanation do not align with existing conceptualization of religiosity and no controls for the other dimensions are included.
3
The literature on the relationship between people’s religion and ethno-religious tolerance is
positioned at the crossroads of two subfields: the more sociological literature on ethnocentrism and
prejudice against ethnic minorities (e.g. Allport & Ross, 1967; Ceobanu & Escandell, 2010; Scheepers,
Gijsberts & Hello, 2002; Strabac & Listhaug, 2008) and the more political science oriented literature
on political and social tolerance (e.g. Gibson, 1992; Bloom & Arikan, 2012; Bloom & Bagno-
Moldavsky, 2015). The first generally focusses on people’s attitudes regarding ethno-religious
outgroups, whereas in the latter social tolerance is defined as “the preparedness to co-exist in the
private realm … [which] involves the willingness to maintain personal contact with the member of
the disliked group, such as by having him or her as a neighbor or a dating partner” (Bloom & Bagno-
Moldavsky, 2015: 626; see also Gibson 1992; Sullivan et.al 1985)). So where the first explicitly
focusses on ethnic minorities and ethno-religious otherness, the second talks more generally about
disliked groups. It is exactly at this crossroad where ethno-religious social tolerance is found: the
preparedness to co-exist in the private realm with people that have a different ethno-religious
background or worldview then oneself.
While ethno-religious social tolerance – as of here social tolerance - has been shown to be an
important influence of people’s support for democracy in the MENA region (Rizzo, Abdel-Latif &
Meyer 2007; Ciftci 2010), hardly any study has focus on explaining individual level social tolerance in
the MENA (cf. Al Sadi & Basit, 2013) let alone studying the impact of (the multiple dimensions of)
religiosity on social tolerance in the MENA.2 This is the more surprising given the large literature on
linking religion to social tolerance in the US and western democratic countries more generally.
Particularly the 3B multidimensional concept of religion – belonging, beliefs, behavior (Bloom &
Arikan, 2012, 2013; Kellstedt etl.al, 1996)3 – combined with social identity theory (see Brown, 2000;
Al Sadi & Basit, 2013; Tajfel, 1981) and socialization theory (see Camilleri & Malewska-Peyre, 1997)
have proven useful to explain the complex impact religion has on tolerance in the Western
democratic context. Below, I will discuss in more detail how belonging, beliefs, and behaviour are
expected to influence social tolerance and the first contribution of this study is therefore largely
empirical: to test whether the results derived from the western literature hold up in the context of
the MENA and help to explain how religion impacts social tolerance there.
In this paper, I thus follow, for instance, Glock and Stark (1966) and Kellstedt et.al (1996) as
well as Islamic religiosity scholar Hassan (2007) in initially understanding the underlying dimension of
religion to be rather universal. At the same time, as Hassan also stresses (2007: 439), the specific
2 Even though social tolerance has been included an explanatory factor in the public opinion MEN literature, the only study known by the author studying the impact of religion on social tolerance in the MENA is Spierings (2014), which excludes contextual factors, the behavior dimension and only covers 5 MENA countries. 3 Evidently the 3B approach is not the only one. Other contributions largely distinguish similar (sub)dimensions but use different labels. Stark & Glock (1968) distinguish more, but the additional one are less relevant with respect to the linkage to tolerance; in that sense Kellstedt et.al’s conceptualization is more developed towards this relationship with tolerance.
4
expressions of the different dimension do differ between religions, and, I would add, the context in
which the dimensions manifest themselves can also shape their relevance and influence, which aligns
with more recent studies on western countries that the context conditions the impact of religion on
tolerance (Verkuyten, Maliepaard, Martinovic & Khoudja, 2014; Bloom & Bagno-Moldavsky, 2015;
Schedler and Sarsfield 2007). More concretely, in the MENA, for instance, attendance is much more
linked to custom and social pressure (Hassan 2007: 442) and the setting is one of various degrees of
lacking religious freedoms and religious tensions or conflicts are much more part of daily live in
several countries. Consequently, the impact of the belonging, beliefs and behaviour need not only be
tested, but also translated and conditioned to the MENA context, with its rather diverse positions of
religion across countries (Moghadam, 2013; Owen, 2004; Spierings, 2015). This theoretical
translation and adaption is the second, and theoretical, contribution of this study, which might
provide important insights into the role religion plays in influencing social tolerance in MENA as well
as the more general mechanisms relevant for our overall understanding of the religion-tolerance
connection.
Belonging, Beliefs and Behaviour
As others have shown, religiosity is a container concept harbouring a multitude of interrelated but
distinguishable dimension (Stark & Glock, 1968; Hassan, 2007; Guveli, 2015) and these dimension
have different relationship with tolerance (e.g. Bloom & Arikan, 2012, 2013; Eisenstein, 2006;
Kellstedt et.al 1996; Spierings, 2014). Particular the distinction, between belonging, beliefs and
behaviour has proven useful in theorizing these differential influences (Smidt, Kellsttedt & Gudt,
2009; Wald & Smidt, 1993; Wald & Wilcox, 2006), and I would go as far as arguing that most
deviations found in the Western literature on religion and tolerance are actually due to collapsing
dimension.4 In other words, the relationships discussed below are formulated under the condition
that the other dimensions are taken into account.
Belonging refers to whether people consider themselves religious, with a particular focus on
religion being part of one’s social identity (Bloom, Arikan & Courtemanche, 2015; Gibson, 2006;
Kellstedt et.al, 1996). This identification leads to intergroup tensions as a positive view of the in-
group is brought about by contrasting the in-group to outgroup. Consequently, a stronger or
activated in-group identification relates to feeling threated by other groups and intolerant views
becoming more likely (Bloom, Arikan & Courtemanche, 2015; Djupe & Calfano, 2013 (PRQ);
Scheepers , Gijsberts & Hello, 2002; Wilcox & Jelen, 1990; cf. Gibson, 2006). This negative
4 For instance Eistenstein (2006) finds no impact of religious commitment on political tolerance, but her operationalization of commitment collapses communal behaviour (attendance), individual behaviour (prayer), and religious salience, thus collapsing differen elements of belonging, socializing behaviour and ritualistic behaviour, which as is discussed here can be expected to have opposing effects.
5
relationship has thus been widely been confirmed in the literature. Studies that seem to show
otherwise either collapse this dimension in their operationalization with other dimensions for which
generally positive relationship are found (e.g. Eisenstein) or do not control for important other
dimensions (e.g. Verkuyten et.al. 2014).
H1: Religious identification is negatively related to ethno-religious social tolerance.
Beliefs refer not so much to the degree of identification, but to the fundamental values and
understanding of one’s religion. Generally, this is linked to religious compassion with the argument
that all major religious worldview have benevolence, care for the needy at their core. So, it is argued
that the more doctrinally orthodox one is, the more likely that that person think in terms of their
group and helping each other and the less individualistic the believer is expected to be. The main
empirically shown caveat, however, is that this benevolence is generally only directed towards the
ingroup (Bloom, Arikan & Courtemanche, 2015; Djupe & Calfano, 2013; Norenzayan, 2014;
Scheepers, Gijsberts & Hello, 2002). Eisenstein (2006), however, finds a negative impact of doctrinal
orthodoxy as it also increased perceived threat. An explanation for this might be that the
conceptualization of doctrinal orthodoxy as beliefs in the 3B approach might actually harbours two
causal mechanisms (see Scheepers, Gijsberts & Hello, 2002). On the one hand there is the
compassion argument; on the other there is one of extremism or particularism (Glock & Stark, 1966;
Scheepers, Gijsberts & Hello, 2002): the more doctrinal orthodox, the more people are at a religious
extreme and see their religion as the only one true religion, which leads to increased negative
orientations towards outgroups. Translating this back to a singular concept of beliefs as doctrinal
orthodoxy, it would mean that the more inwards oriented a person is the more likely that doctrinal
orthodoxy relates negatively to tolerance, whereas among less inwards-oriented people, the
doctrinally orthodox might actually more tolerant. A recent study by Bloom & Bagno-Moldavsky
(2015) seems to lend some support for this formulation, as they find that network diversity feeds
tolerance if people are open-minded but not otherwise.
H2: Doctrinal orthodoxy is negatively related to ethno-religious social tolerance among ingroup
oriented people and positively related to ethno-religious social tolerance among less ingroup
oriented people.
Last, behaviour is arguable given least attention of the 3B in the tolerance literature, partly
because it is generally seen as an antecedent of belonging or beliefs (e.g. Bloom, Arikan &
Courtemanche, 2015), but it actually has shown to be an independent influence too generally
decreasing tolerance (Scheepers, Gijsberts & Hello, 2002; Bloom & Arikan 2012; cf. Eisenstein, 2006).
This additional negative effect can be explained in terms of socialization and social norms.
6
Attendance not just leads to stronger feelings of belonging or instils orthodox views, but also feeds
unconscious negative attitudes towards other (religious) groups. The assumption in this argument
here is that at least in some sermons or gathering this message is conveyed explicitly or implicitly.
H3: Communal attendance is negatively related to ethno-religious social tolerance.
Religion and tolerance in the MENA context
As said, I follow Glock & Stark (1966), Hassan (2007), and others in expecting the dimension of
religion being rather universal, while at the same time acknowledging that their expression might
differ between religions (Hassan, 2007) and their impact might be contextually shaped. Below I will
therefore reflect on the expectations formulated above by explicitly paying attention to Islam as the
dominant religion in the MENA region and to how the different and diverse societal and institutional
context in terms of religion-state relations and religious conflict matters for the impact belonging,
belief and behaviour might have.
The mechanism of social identity often links belonging in terms of identifying as a group
member to tolerance. In the MENA, not being Muslim or not being religious is often not even an
option. For instance, in several surveys people do not even have the option to say they are atheist.
Contrasting membership of different groups is hardly a solution in this respect as the social identity
mechanism does not distinguish between different types of groups. Moreover, contrasting for
instance Muslim citizens to Christian citizens collapses doctrinal categories (which could be linked to
beliefs) with majority and minority status, which are highly relevant in the context of religious
tensions or discrimination in some MENA countries. Therefore, in terms of belonging it is crucial to
which degree people consider themselves religious. Thus the expression of Hypothesis 1 as
formulated above should be that the more people identify as religious the lower their tolerance is
expected to be.
As already alluded to above, in several MENA countries threat is more real than it is in
Western countries in general or some particular stable, safe and religiously homogenous MENA
countries. First of all, while all countries are affected by region-wide politics such as the Arab-Israeli
conflict and the Arab Uprisings (see Dalacoura, 2012; Netterstrøm, 2015; Spierings, 2017), in some
countries are more religiously diverse with salient religious cleavages (e.g. Bahrain and Lebanon) or
event ethno-religious violent conflict within their borders at least in some years (e.g. Syria and
Yemen). Under these circumstances the threat perception linking religious social identity to tolerance
can be expected to be activated more strongly. Moreover, the perceived threat linking religious
belonging and tolerance can also be expected to be stronger among non-Muslim citizens in the
MENA context. First of all, Islam is the dominant religion in all Arab MENA countries and, although
this is theologically contested (Al Sadi & Basit 2013; Abou El Fadl 2002; Baderin, 2003), the distinction
7
between Muslim believers, Christians and Jews (also referred to as People of the Book, Ahl al-Kitab,
or Dhimmi), and non-believers (also kafir or kufr) is regularly interpreted as a hierarchical distinction,
distinguishing perceived better from lesser people who should have respectively more and fewer
rights and status (e.g. Geaves, 2006). Citizens that have a stronger belonging to one of the non-
Muslim groups can than logically be expected to perceive more threat in case of equal belonging and
thus less tolerance.
H1a: The stronger the religious fractionalization and conflict in a country, the stronger the negative
relation between religious identification and ethno-religious social tolerance.
H1b: Among non-Muslims the negative relation between religious identification and ethno-religious
social tolerance is stronger than among Muslims
While Zakat (almsgiving) is one of five pillars in (almost all) Islam and prescribes all believers
to show compassion (Geaves, 2006: 121), much societal and academic discussion exists on what the
Qu’ran and hadith say about the compassion towards non-Muslims, as already referred to above
already. This seems to provide further theoretical support for the differential hypothesis on the
impact of doctrinal orthodoxy as formulated in the previous section.
At the same time we can expect the variation in state-religion relations to influence the
relationship between doctrinal orthodoxy and social tolerance as in several but certainly not all
MENA countries it is particularly the orthodox interpretations of Islam that have been suppressed.
For instance in Ben Ali’s Tunisia and Mubarak’s Egypt social and political expressions of religion were
banned and in Iraq Hussein interfered negatively with the Hajj (pilgrimage to Mecca and Medina) (El-
Ghobashy, 2005; Owen 2004; Spierings, 2014). From a Social Identity Theory perspective this threat
to more orthodox people particularly can be expected to active their orthodox identity and lead to
intolerance. At the same time others have argued from a more beliefs-centred perspective that this
might actually increase tolerance, bit little evidence have been found for this (cf. Djupe & Calfano
2013; Meyer, Rizzo & Ali 2007; Spierings, 2014). At the core of this argument is the religious notion
“Do not do unto others what you would not have them do unto you” which is not literally found in
Islamic scripture, but analogous Shura include 24:22 and 83:1-6. Particularly orthodox believers
experiencing discrimination are expected to fall back on this commandment, leading to a heightened
realization that the tolerance is an important virtue. In sum two additional, contradictory hypothesis
on doctrinal orthodoxy are formulated:
H2a1: If orthodox Islam is suppressed in a country, doctrinal orthodoxy in negatively related to
ethno-religious social tolerance.
H2a2: If orthodox Islam is suppressed in a country, doctrinal orthodoxy in positively related to ethno-
religious social tolerance.
8
Regarding the dimension of Behaviour, it is crucial to understand that attendance of prayer in
the MENA is considered a social activity strongly subject to social norms and custom of one’s direct
environment (e.g. Hassan, 2007: 442). In other words, one can easily attend mosque or prayer while
at the individual level distance oneself from what is preached and consciously disagree. Through
attendance one could therefore be socialized into an unconscious belonging and ideas that other
religions should be regarded as less, lending further supports for the idea of an additional impact of
attendance on tolerance as hypothesized in the previous section. At the same time, it should be
acknowledged that the messages preached by the cleric (e.g. imam) can differ strongly in the degree
to which it denounces other religions and other religious groups in society. Accordingly, a study on
educational programs on tolerance in Oman has shown that the actual message communicated by
figures of authority does matter in shaping children’s ethno-religious social tolerance (Al Sadi & Basit,
2013). Consequently, we can expect differences between people attending different mosques, which
can unfortunately not be tested at the mosque-level given existing MENA wide survey data, as well
as differences between countries depending on the state-religion relations. In several MENA
countries religious matters are strictly regulated by governments and in some of these cases they
prescribed a rather exclusionary or Islamist form of Islam (e.g. Testas 2005). For instance, in Saudi
Arabia the government enforces Salafi or Wahhabi Islam and central government councils of clerics
issue the fatwas that are communicated by imams to the people during attending services (see
Boucek 2010). Particularly when this state controls the sermons and backs a particularist or Islamist
version of Islam, mosque attendance will imply communicating intolerance and (unconsciously)
socializing people accordingly. In line with this the last additional hypothesis is:
H3a: If a country is more Islamist and regulates sermon, attendance has a stronger negative
relationship with ethno-religious social tolerance.
Data & Method
Data sources
To test both the generalizability and the context-dependency of religion’s impact in the MENA, I
combine all 32 World Value Surveys (WVS, 1981-2014) and Arab Barometer surveys (AB, 2006-2014)
that have at least one valid item for each of the core theoretical concepts. This leads me to include
32 nationally representative surveys from 13 Arab MENA countries: Algeria, Bahrain, Egypt, Iraq,
Jordan, Lebanon, Libya, Morocco, Palestine, Saudi Arabia, Sudan, Tunisia, and Yemen. The first
included surveys were held in 2001 and the last in 2014, with some countries being represented only
9
once (Libya) and others four times (Algeria, Iraq, and Jordan). A list of surveys included and more
detailed information per surveys is found in Appendix A.
By combining pooled models based on standardized values with country- and survey-
disaggregated analyses it will be prevented that differences in the exact items present for each
theoretical concept per survey bias the substantive conclusions, as it is assessed whether the results
systematically differ according to this slight differences in phrasing of the questions. At the same
time, combining as many MENA surveys as possible allows transcending both dominant approaches
in MENA survey studies: (a) single-country studies, in which the external generalizability of the
results cannot be assessed, and (b) pooled ‘civilizationist’ studies that generalize results across the
MENA by comparing the region as a whole to the Western democracy while ignoring that the
dynamics might differ considerable within the MENA as well. The approach followed here, including
many MENA countries and years is best, despite data limitations, to grasp to what extent results are
generalizable within the MENA and what shapes the different relationship across the region.
To estimate a pooled model, the standardized values of all items are calculated per survey.
This creates similar within country-year metrics across items. Setting the country-level averages to
zero is a consequence of this, but as this study does not focus on explaining differences in the
country-year level of ethno-religious tolerance, these 0-averages are not problematic here. Still,
random-intercept multilevel models with survey and country as higher levels allows for testing the
general impact of the different religiosity items taking into account that errors are not uncorrelated
(i.e. cases are nested) and allows testing Hypotheses 1a, 2a1, 2a2, and 3a on the context-dependent
impact of the different religiosity variables without deflating the standard errors. The interaction
models are estimated with random slopes of the religiosity variables.
Additionally, country-disaggregated analyses will be modelled too, which is always advisable
when estimating models on pooled data. First of all, this shows to what extent the pooled models on
this issue actually represent the relationship per country well. Second, while these disaggregated
models do not allow for formal tests of differences in effects, they are important to see in which
countries and years certain dimensions of religiosity have a positive, negative or insignificant effect –
the core difference in effects focussed on here5. This allows for assessing whether these patterns are
related to differences in the operationalization of the dimensions, the contextual-differences
hypothesized or other substantive differences between countries (see Spierings 2016).
Ethno-religious social tolerance
5 More subtle differences between the size of, for instance, a negative effect of belonging cannot be reliably be assessed with these data.
10
Following the literature I defined ethno-religious social tolerance as the preparedness to co-exist in
the private realm with people that have a different ethno-religious background or worldview than
oneself. A common operationalization in non-Western countries focuses on accepting people of a
different race, colour or ethnicity or with a migration background as neighbour (Ciftci, 2010; Dixon,
2008; Rizzo, Abdel-Latif & Meyer, 2007; Spierings, 2014). As the referred to studies show, items on
these different groups load on the same factor, implying one underlying concept. All datasets used
here include one or more of these items asking which groups a respondent objects to having as
neighbours. Of these, all items are selected that refer to religious, ethnic, national or racial groups
and thus refer to ethno-religious social tolerance. A full overview of the groups and which was
available per survey is given in Appendix B1. Reliability test and factor analysis of different sets of
items all show that combining these items leads to a valid and reliable indicator.6 While the items
thus tap one concept, we do know that particular groups can be less or more popular in the different
countries. As the groups included differs somewhat per survey, I therefor standardized all scores by
calculating z-scores per item per survey. A higher score indicates more tolerant attitudes. In all
surveys at least two items are available. Per respondent the average of all available scores is used as
the dependent variable in this study.7
To test the robustness of the results a second dependent variable was calculated on the two
items presents in most surveys and most directly linked to the ethno-religious social tolerance: not
objecting to neighbours of a different religion and of a different race. Using the summative score of
these two items as alternative did not lead to markedly different results for the available data, but
did lead to the loss of 6 datasets and 10,000 respondents. In the results section the more
comprehensive analyses are presented.
Dimensions of religiosity
As discussed in the theory section, for the MENA it is crucial to focus on the degree of religious
identification as belonging. All used surveys include at least one item of the following three: To what
extent do you consider you self as religious (not, somewhat, religious); Are you a religious person?
(no, including atheists; yes); How important is religion in your life? (0 to 3). In multiple WVS surveys
the latter two are included and factor analysis on them shows they load on one singular dimension. A
third item - How important is god in your life? (0 to 9) – that was initially selected showed a factor
6 The Cronbach’s alpha on the different (most presents) sets of items all are well above 0.6: (1) other race, migrant background, other religion, other language 0.718; (2) other race, migrant background, other religion, refugees 0.771; (3) other race, migrant background, other religion 0.650. Factor analyses for each of these sets of neighbor indicators show they load on one dimension, with all loadings above 0.7, with one exception which is still 0.69. 7 The Arab Barometer round 3 surveys and three WVS 5th round surveys did not include neighbor items but did included items on citizens’ rights. Additional analyses showed that these items do tap a form of tolerance, particularly at the macro-level (see also Spierings 2017) but that they related differently to individual-level religiosity and therefore these surveys could not be included in this study on ethno-religious social tolerance.
11
loading well below 0.4, actually reproducing Stark & Glock’s (1968) conceptualization, and was not
included. The average survey-specific z-score over the available items per respondent was calculated
(higher score indicating a stronger religious identification). Appendix B2 presents the items included
per survey.
In addition, for H1b denomination is included as a dichotomous variable indicating whether a
respondent is Muslim or not. In five AB and WVS surveys8 no specific question on denomination was
included. In three of these Islam is makes up 98 to almost 100% of the population so I assumed the
respondents to be Muslim, not to lose these surveys.9 Surveys of these countries in other years
support this assumption: only 15 of over 10,000 respondents indicate not to be Muslim. The two
other surveys could not be included as this assumption does not hold there.10 As discussed in the
theory section the main effect of this denominational dummy in the interaction model testing H1b
should be interpreted with care as religious denomination and majority-minority status fully overlap.
The hypotheses on the beliefs focus on doctrinal orthodoxy and links this to in-group
orientation. Doctrinal orthodoxy is included in each survey in the form of people’s views on a set of
haram acts. The more a respondent thinks they should be forbidden or are not justified; the more
they are orthodox in sticking to the doctrine regarding the scriptures. The surveys issues include
suicide, alcohol, euthanasia, charging interest, divorce, and participation in a lottery. Appendix B2
indicates which items were available in which survey. While different, each act is forbidden. Again z-
scores are used to standardized the scores. Religious in-group orientation is expected to moderate
the effect of doctrinal orthodoxy. Empirically it could only be operationalized somewhat more
indirectly, by using six items on the extent to which religion is important in selection one’s closest
network: the nearest and dearest, the first-degree relative (see Appendix B2). Of these items the
within-survey z-scores are calculated and the average per respondent is used for the interaction
term.
Regarding the dimension of behaviour, attendance of communal services is at the core of the
expectations. Again each survey includes slightly different items, but all ask directly about religious
attendance: how regularly a respondent attends Friday prayer (or Sunday service) or whether the
respondent prays in church or mosque, and how often (Appendix B2). The higher the within survey z-
score per respondent the more a respondent attends communal services compared to the other
respondent in that country and year.
8 Algeria 2006, Bahrain 2014, Egypt 2012, Morocco 2006, Yemen 2011 9 Algeria 2006, Morocco 2006, Yemen 2011 10 Bahrain 2014, Egypt 2012
12
Micro-level control variables
Core socio-economic and demographic variables are included to control for confounding effect.
Available across surveys are employment status (employed; retired; housewife; student; not
employed other), education (no education, completed primary education, completed secondary
education, at least some tertiary education), sex (female, male), age in 7 categories11 and age
squared (to account for curvilinear effects), and marital status (married, single, and other). After
selecting all respondents with valid score on the main dependent variable, all religion dimensions,
and these control variables 37,413 remain. A last additional control variable is the respondents’
household income. This control variable is only added in additional analyses, as it has a relatively
large number of missing values across survey (4,563 of the remaining 37,413 cases). It is included as
standardized z-scores of 10-point variables (which use somewhat different metrics across the
surveys). Descriptive information on all micro-level variables is given in Appendix C1.
Contextual variables
To test the context-dependency hypothesis, macro-level scores on the relevant concepts are
obtained from existing data sets and studies or coded specifically for this study. An overview of
scores can be found in Appendix C2.
To test Hypothesis H1a the degree of religious fractionalization and violence was measured,
by two variables. The first is the standard measure of religious fractionalization as operationalization
in the seminal article by Alesina, Devleeschauwer, Easterly, Kurlat & Wacziarg (2003). As religious
fractionalization is fairly stable over time (certainly over a period of at most 14 year) and
(consequently) data are not available for every year, I use the scores presented by Alesina et.al. as
time-invariant data at the country level. Second, I created an indicator to measure a broad religious-
based threat of violence in society. The Global Restrictions on Religion Data (GRRD [ARDA, 2017])12
offers, to my knowledge, the broadest range of acts of religious-based (threats of) violence. Per
country-year the average score on 17 items13 (all running from 0 to 1) was calculated for the year of
11 For one Moroccan survey only the categories are available. Rerunning the models with age as an interval variables excluding this survey does not lead to substantively different results. 12 The data were downloaded from the Association of Religion Data Archives, www.TheARDA.com, and were collected by the Pew Research Center. 13 (i) mob violence related to religion; (ii) acts of sectarian or communal violence between religious groups; (iii) activity of religion-related terrorist groups; (iv) extensiveness of a religion-related war or armed conflict; (v) violence resulting from tensions between religious groups; (vi) organized groups using force or coercion in an attempt to dominate public life with their perspective on religion; (vii) religious groups attempting to prevent other groups to operate; (viii) use or threat of violence to try to enforce religious norms; (ix) assaults and displacements from homes in retaliation for religious activities; (x) incidents of hostility over proselytizing; (xi) hostility over conversations from one religion to another; (xii) harassment or intimidation of religious groups by social groups motivated by religious hatred or bias; (xiii) destruction of personal or religious property motivated by religious hatred or bias; (xiv) detentions or abductions motivated by religious hatred or bias; (xv) any displacement of individuals from their homes motivated by religious hatred or bias; (xvi) physical assaults motivated by religious hatred or bias; (xvii) deaths motivated by religious hatred or bias.
13
the survey and the year before the survey (if only one of two was available that single score is use).
This data captures a broad range of recent violent events that might influence how people translate
their religious belonging to tolerance. Because the main differences in scores are found between
countries and not between the years included in the analyses here and the GRRD data is only
available from 2007 through 2013, I subsequently calculated country averages and also this variable
is included at the country-level.14
For the impact of doctrinal orthodoxy, a possible moderation by suppressing orthodox Islam
was derived from the literature (Hypotheses H2a1 and H2a2). As a measurement of this I combined
two dimensions: (A) whether Islamist are forbidden to organize politically, in power politically15, or
neither; and (B) whether main orthodox religious groups where deprived or suppressed in society,
economy and politics, as was for instance the case with the Muslim Brotherhood being cracked down
upon in the Egypt of el-Sisi or Mubarak around 2007 (e.g. Owen, 2004).16 Subsequently, I grouped
countries in three: all cases with orthodox Islam being in political power (-1); countries in which
orthodox groups are suppressed on at least one front (1); and countries in which neither was the
case (0).
Lastly, to test Hypothesis H3a I combined two data sources in order to come to an indication
in which countries and years a fundamentalist government controls sermons, which falls apart in two
elements: whether Islamists rule and whether government has control over what is preached in
mosques. Regarding the latter, the GRRD (see above) also include an indicator of the extent to which
the government regulates or manages religious affairs. For years missing (2014 and the earliest
surveys), I consulted the literature to determine whether major changes could have been expected
compared to years for which data was available. The focus for this study is on the highest of four
scores with indicates strict and enforced regulation. Next I combined this with the Political Islamism
indicator as introduced in Spierings, Smits & Verloo (2010) and updated in Spierings (2015), which
indicates to what extent government is institutionally Islamist based on countries’ constitutions. For
the missing years the coding scheme provided in these sources was used to supplement the data
with new scores. Scores of 5 and 6 indicate countries without religious freedom, an Islamic state, and
14 Rerunning the analyses on the country-years for which country-year data was available and including this at that level does not lead to different conclusions. 15 With Hamas ruling in Gaza, the Palestinian territories are included 2006 onwards; the others are Saudi Arabia, Sudan, Tunisia in 2011, Egypt in 2012, and Yemen. 16 The other country-years considered here are: Bahrain with the Shi’a majority being discriminated against (Gengler, 2013), the Lebanese Sunni’s in 2012/2013 being politically intimidated (BBC), and to some extent (coded 0.5) the Islamist Sudanese by not having control over oil-rich Abyei. Other examples could be Tunisia under the rule of Ben Ali, but only surveys of Tunisia from later years are included.
14
sharia based law. So if in a country and year government regulated religions strictly according to the
GRRD and was Islamist (5 or 6) it was coded 1 for the purpose of this study, otherwise 0.17
Results
Religiosity and social tolerance across the MENA region
As a first step, Table 1 (Models 1 through 4) provides the results of the analyses that represent the
overall average impact of belonging, doctrinal orthodoxy and attendance. Finding an effect in these
models indicates that across all 13 MENA countries a patterns if found that is highly unlikely to be
coincidence, but it does not mean that the effect is per se present in all 13 countries (or all years);
this will be focussed on explicitly in the next section. Model 1 presents the main effects of the three
variables at the core of the hypotheses, controlled for differences in tolerance between socio-
economic and demographic societal groups as well as for survey-level differences in tolerance. Model
2 and 3 present the two micro-level interaction terms respectively (cf. H1b, H2). Model 4 includes the
results of the ‘most extreme’ robustness check: the model with the more restrictive dependent
variable, income as additional control variable, and only Muslim respondents as units of analysis. As
each of the models contributes pieces of the puzzle for each effect, I will discuss these results per
dimension of religiosity across models.
In all models people who more strongly identify as religious (belonging) are on average
somewhat less tolerant towards other ethno-religious groups, but this difference is clearly not always
statistically significant at conventional levels. There is thus considerable probability that this
relationship is not found in the population at large. This indicates that there is either no clear
relationship or under some circumstances a negative and under other positive one, which cancel
each other out. However, no indications were found that the effect is considerable stronger among
non-Muslim citizens (H1a) or weaker for that matter (see Model 2). The disaggregated analyses
might shed more light on whether and how the impact of belonging is dependent on other factors.
Regarding the denomination – the other half of the interaction terms (see Model 2) it is
important to note that, as discussed, we should be careful interpreting the results. That non-Muslims
are on average less tolerant (Model 2), might go against the expectation in some of the literature
that Muslim are generally less ethno-religiously tolerant, but we cannot disentangle denomination
17 Algeria, Bahrain, Saudi Arabia in 2011, and Yemen as of 2011 are coded 1; 5 of 32 country years. Algeria’s score was corrected from the original source from 3 to 5 [it does not have explicit freedom of religion, head of state must be Muslim, practices contrary to Islamic ethics are forbidden (implying the importance of Shari’a); and the state’s religion is Islam] which means it scores 5 to 6, both are coded 1 here.
15
from majority/minority status here. It might well be that is particularly because of belonging to a
religious minority that non-Muslim are more hesitant to have neighbours belonging to the religious
majority, have lower social tolerance.
Doctrinal orthodoxy, linked to the dimension of beliefs, shows a rather stable positive
relationship with social tolerance across the models: the more orthodox the more tolerant. It was
hypothesized that this effect might differ in direction depending on the in-group or out-group
orientation of the respondent, but no strong evidence for this was found. The interaction was only
marginally significant and substantially only indicates a variation in the degree of the positive
relationship, which refutes H2. It seems that people who attach stronger value to living by the book,
are on average more tolerant towards others, also if they do not share their religion, suggesting the
benevolence mechanisms is dominant in understanding the impact of doctrinal orthodoxy
(controlled for the other dimensions or religiosity).
Model 3 also shows a strong negative relationship of in-group focus, which mediates the
impact of belonging it seems (cf. Models 1 and 3). Again care is needed in interpreting this direct
effect of in-group orientation as the specific operationalization focuses on a preference of people
with similar religious background in one’s closest social network, with the dependent variable in the
models being a similar, but negatively formulated variable for the larger social network. As such this
direct negative effect is not surprising, but contributes little new insights in itself. Interestingly
though is that is particularly mediates the effect or religious identification, which is (marginally
significant) in Models 1, 2 and 4, but not in 3. This lends some additional support for H1.
Lastly, communal attendance has a clear negative relationship with ethno-religious social
tolerance in all pooled models. The more people attend religious services the less likely they are on
average not to object to living next to people with ethno-religious backgrounds. This preliminarily
confirmed H3.
Table 1. Multilevel regression models estimating the impact of religion on ethno-religious tolerance
Model 1: Direct effects
Model 2: Interaction H1b
Model 3: Interaction H2
Model 4: Robustness – different DV, incl. income and only Muslim respondents
FIXED RELIGION Belonging -0.0084# -0.0077# -0.0047 -0.0193*** Non-Muslims (ref=Muslim) -0.0493** Interaction Belonging* non-Muslim -0.0130 Doctrinal orthodoxy 0.0150** 0.0143** 0.0164*** 0.0216*** Religious in-group orientation -0.0265*** Interaction Orthodoxy*in-group 0.0077# Attendance -0.0232*** -0.0226*** -0.0219*** -0.0181** CONTROLS
16
Employment (ref = employed) Retired -0.0316 -0.0334 -0.0310 -0.0141 Housewife -0.0268* -0.0274* -0.0247* -0.0300# Student 0.0199 0.0196 0.0193 0.0464* Not employed other -0.0037 -0.0047 -0.0036 0.0070 Education (ref = no) Primary completed 0.0786*** 0.0797*** 0.0767*** 0.0552*** Secondary completed 0.1158*** 0.1173*** 0.1136*** 0.1142*** At least some tertiary 0.1832*** 0.1847*** 0.1817*** 0.1964*** Sex (Ref = male) -.0091 -0.0082 -0.0085 -0.0111 Age (0-7) 0.0195* 0.0202* 0.0192* 0.0417*** Age sqaured 0.0010 0.0010 0.0009 -0.0041#
Marital status (ref = married) Single 0.0274* 0.0283** 0.0267* 0.0307* Other (incl. divorced, widowed) 0.0140 0.0134 0.0139 0.0568* Income 0.0175*** Intercept -0.1335*** -0.1334*** -0.1316*** 1.2808*** RANDOM Survey level Intercept 0.000123 0.000128 0.000098 0.036442* Belonging Doctrinal Orthodoxy Attendance RANDOM Country level Intercept 0.000322 0.000286 0.000317 0.024714 MODEL STATISTICS BIC 84285.259 84289.814 84260.780 52044.771 Nind 37,413 37,413 37,413 23,533 Nsrvy 32 32 32 28 Nctry 13 13 13 13 *** p<0.001 ** p<0.01 * p<0.05 # p<0.10 Data source: AB and WVS surveys
How the impact of religiosity differs across contexts
Based on the pooled analyses represented in Table 1, the impact of religiosity seems rather
straightforward, confirming some of the expectations derived from the Western-based literature
while refuting others, particularly the restriction of orthodoxy’s benevolence towards people from
other religions. However, there is some more nuance to reality, as illustrated by Table 2 which shows
the coefficients of the country-disaggregated models. For doctrinal orthodoxy and attendance the
results from the pooled model are largely reproduced for all countries but some exceptions are laid
bare. In respectively 10 and 11 of 13 countries a positive coefficient for doctrinal orthodoxy and a
negative one for attendance is found. For both, the coefficients in five countries are large enough to
reach statistical significance (p<0.10). At the same time, we find a few exceptions: in Bahrain and
Sudan the relationship between doctrinal orthodoxy and social tolerance is negative and statistically
17
significant, and in Libya a positive and significant effect is found for attendance. However,
orthodoxy’s positive impact and attendance’s negative one by and large represent a relatively
general pattern that helps to understand the impact of religiosity on social tolerance in the MENA.
For the impact of identifying less or more strongly as religious, a whole different conclusion should
be drawn based on Table 2: there does not seem to be a rather general effect. The pool of 13
countries is split across the middle, with half showing a negative relationship and half a positive one
– regardless of whether we focus on all coefficients or the statistically significant ones only. In other
words: under some circumstances a stronger religious identity leads to less social tolerance, while
under other circumstances it is associated with more tolerance. Additional models with random
slopes for the three core religiosity variables confirm this picture: the variance of the coefficient
differs significantly at the contextual levels. Moreover, although the effect of orthodoxy and
attendance are mainly in the same direction also those effects are significantly different across
context statistically speaking. The pooled regression models thus suggest a universality or internal
generalizability of effect across the MENA that might turn out an unwarranted simplification after
more detailed analysis.
Table 2. Impact of Religious belonging, orthodoxy and attendance on ethno-religious social tolerance in 13 MENA countries
Belonging Doctrinal Orthodoxy Attendance
Algeria (3811) 0.082*** 0.002 -0.080*** Bahrain (340) -0.038 -0.176** 0.037 Egypt (4032) -0.003 0.031# -0.018 Iraq (6422) 0.004 0.010 -0.012 Jordan (4330) -0.002 0.017 -0.010 Lebanon (3085) 0.031* -0.013 -0.039** Libya (1712) 0.009 0.010 0.035* Morocco (3290) 0.038* 0.020 -0.007 Palestine (1934) -0.104*** 0.048* 0.019 Saudi Arabia (2236) -0.079*** 0.061** -0.048** Sudan (1270) -0.060** -0.044* -0.045* Tunisia (2253) -0.044* 0.035# -0.002 Yemen (4546) 0.016 0.055** -0.062**
*** p<0.001 ** p<0.001 * p<0.05 # p<0.10 All models are controlled for employment status, education, marital status, sex, and age. Positive coefficient, statistically significant Positive coefficient, statistically insignificant Negative coefficient, statistically significant Negative coefficient, statistically insignificant
The context-dependency as discussed and shown above thus asks for more attention. In
Table 3 the results are given of the models testing the hypotheses formulated above. Each of these
models is estimated with random slopes which means the threshold to reach statistical significance
as very high as the degrees of freedom for these cross-level interaction coefficients are close to the
number of countries or surveys included. In addition, for the relationships that are considered
statistically significant, I plotted the coefficients of identification, orthodoxy, or attendance per
18
country or survey (see e.g. Table 2) against the values of the contextual interaction variable in order
to provide an additional, visual assessment of the relationships. This will lay bare the impact of
outliers as well as provide more nuance to what actually makes the difference in terms of cut of
points or non-linear relationships.
Table 3. Multilevel models estimating the impact of contextual factors on the relationship between religion and tolerance.
Interaction with
Belonging Doctrinal orthodoxy
Attendance
FIXED RELIGION Belonging 0.0052 -0.0075# -0.0078# Interaction with Religious Fractionalization 0.0082 Interaction with Religious-based Violence -0.0444 Doctrinal orthodoxy 0.0151** 0.0090** 0.0152** Interaction with Orthodox Islam being suppressed
-0.0474*
Attendance -0.0236*** -0.0219*** -0.0156 Interaction with Islamist regulated sermons -0.0706* CONTROLS Employment (ref = employed) Retired -0.0298 -0.0299 -0.0317 Housewife -0.0272* -0.0292* -0.0303* Student 0.0200 0.0207 0.0241 Not employed other -0.0051 -0.0055 -0.0024 Education (ref = no) Primary completed 0.0813*** 0.0800*** 0.0834*** Secondary completed 0.1186*** 0.1215*** 0.1237*** At least some tertiary 0.1860*** 0.1888*** 0.1945*** Sex (Ref = male) -0.0083 -0.0080 -0.0135 Age (0-7) 0.0209* 0.0217* 0.0214* Age squared 0.0007 0.0006 0.0007 Marital status (ref = married) Single 0.0277* 0.0292** 0.0286* Other (incl. divorced, widowed) 0.0150 0.0129 0.0131 CONTEXTUAL FACTORS Religious Fractionalization (country) -0.0328 Religious-based violence (country) -0.0181 Orthodox Islam being suppressed (country year) -0.0017 Islamist regulated sermons (country year) 0.0133 Intercept -0.1409*** -0.1412*** -0.1440*** RANDOM Survey level Intercept 0.000095 0.000734 0.001556 Belonging 0.002059* Doctrinal Orthodoxy 0.004411* Attendance 0.005334 RANDOM Country level Intercept 0.000402 0.000889 0.001906
19
Belonging 0.001480 Doctrinal Orthodoxy 0.000267 Attendance 0.000052 MODEL STATISTICS BIC 84260.746 84252.023 84215.342 Nind 37,413 37,413 37,413 Nsrvy 32 32 32 Nctry 13 13 13 *** p<0.001 ** p<0.01 * p<0.05 # p<0.10 Data source: AB and WVS surveys; GRRD data; Alesina et.al (2003); Spierings, Smits & Verloo (2009); Spierings, 2015; own codings
Model 1 (Table 3) shows that there does not seem to be a systematic relationship between
the religious fractionalization in a country or the degree of violence and the impact one’s religious
identification has on their ethno-religious tolerance. A further inspection of the relationships,
including models with alternative operationalizations, supports the conclusion that the expected
relationships are unlikely. At the same time it is noteworthy that Table 2 not only illustrates the
context-dependent impact of religious identification, but also suggests a pattern: all statistically
negative relationship are found in countries in which Islamist have ruled. Moreover, if we plot against
the survey level coefficient are rather strong pattern appears (see Figure 1) that is also statistically
significant if we model it as a cross-level interaction. In the conclusion I will reflect on this more.
While the impact of doctrinal orthodoxy on tolerance was relative robustly positive across
the board, the interaction term with societal position of orthodox Islam in society is statistically
significant (Model 2, Table 3). Figure 2 adds to this that this differential effect is mainly caused by the
impact of doctrinal orthodoxy on social tolerance being actually much more likely to be (strongly)
negative in countries in which orthodox Muslim have reason to feel suppressed. As the figure shows
this effect is driven by three of thirty-two country-years. While the pattern warrants more research,
the data and models here clearly suggest that under normal circumstances more doctrinal orthodox
people are on average more tolerant, but if they are suppressed as collective this turns to a rather
strong negative impact, whereby particularly the most orthodox citizens do not want to be in social
contact with people of other groups.
Lastly, I looked at whether mosque attendance had a stronger (more likely) negative effect in
countries and years in which sermons were more controlled by Islamist governments compared to
countries and years in which this was not the case. The difference in effect of attendance between
these two groups is statistically significant in Table 3. Actually the general negative effect seems to be
driven rather strongly by the countries in which such control was present, as they include the
country-years in which the strongest negative effect was found (see Figure 3), such as Saudi Arabia
(2011) and Yemen (2014).
20
-0,3
-0,2
-0,1
0
0,1
0,2
0,3
0 1
Co
eff
icie
nt
for
Re
ligio
us
Ide
nti
fica
tio
n
on
eth
no
-re
ligio
us
soci
al t
ole
ran
ce
Islamists ruling
Figure 1. Context-dependent impact of religious identifcation
-0,2
-0,15
-0,1
-0,05
0
0,05
0,1
0,15
-1 0 1
Co
eff
icie
nt
for
Do
ctri
nal
Ort
ho
do
xyo
n
eth
no
-re
ligio
us
soci
al t
ole
ran
ce
Suppression orthodox islam
Figure 2. Context-dependent impact of doctrinal orthodoxy
-0,3
-0,2
-0,1
0
0,1
0 1
Co
eff
icie
nt
for
relg
iou
s at
ted
ance
on
eth
no
-re
ligio
us
soci
al t
ole
ran
ce
Islamist control over mosque sermons
Figure 3. Context-dependent effect of religious attendance
21
Conclusion
This study set out to provide a systematic and nuanced understanding how MENA people’s religious
belonging, beliefs and behaviour shape their ethno-religious social tolerance. While social tolerance
is a fundamental attitude for sustainable democracies to come about and religion being an important
part of many people’s live across the region (Ciftci, 2010; Harik, 1994: 56; Rizzo, Abdel-Latif & Meyer,
2007: 1155; Spierings, 2014), the relationship between individual level religiosity and tolerance has
hardly gotten any attention in the public opinion literature on the MENA region (cf. Spierings, 2014).
Not only does this mean that little is known about the role religion plays in shaping the MENA
peoples’ tolerance, but this also implies that we should be very careful in generalizing results from
the broader and Western-dominated literature to the MENA region.
While also in the Western-based literature the debate on the impact of different dimension
of religiosity is ongoing, the 3B – belonging, beliefs, behaviour (Kellstedt et.al 1996; Leege & Kellstedt
1993; Wilson, Smidt, Kellstedt & Guth 2009) – has proven to provide a very fruitful framework for
theorizing and analysing the relationship between religion and tolerance and is applied and refined
to the MENA context in which Islam is the predominant religion and religious conflict, suppression
and rule form the context in which the individual-level relationships should be understood.
Moreover, these contextual elements not only characterize the region, but also vary considerably
among different MENA countries. This study thus contributes theoretically and empirically to our
understanding of how the religiosity-tolerance linkage is context-dependent. First, it theorizes and
empirically tests whether relationships found in Western countries are generalizable to the MENA
region. Second, acknowledging ‘“the internal dynamics and plurality” of Muslim nations’ in the
Middle East (Davis & Robinson, 2006: 167), I test and theoretically explore how the different political
and social position religion has across the region shapes the impact religious belonging, belief and
behaviour has on ethno-religious social tolerance. In sum, this study has put forward a context-
sensitive 3B approach, which turned out rather helpful in understanding the multifaceted empirical
relationship between religion and tolerance in the MENA.
For each of the three core dimensions of religion – Belonging (degree of identification), Belief
(doctrinal orthodoxy), Behaviour (communal attendance) – I found that they relate to the degree of
social ethno-religious tolerance, but that this relationship differs in strength or even direct across
countries and years. At the same time, particularly doctrinal orthodoxy and communal attendance
showed predominantly similar effects in direction across countries. Moreover, the differences in
relationships could partly be linked to differences in the socio-political and institutional position
religion has at the contextual level. As such, this study’s result not just contribute to more insight in
the MENA region, but also offers a understanding of why scholar from a multitude of disciplinary
22
background have found different results for the impact religion has on attitudes towards other ethnic
and religious groups, including immigrants (see also Bloom, Arikan, Courtemanche, 2015: 203).
First, the impact of the degree of belonging on social tolerance varied considerably in
presence, strength and direction across the MENA and was not particularly negative among Muslims
or in societies characterized by religious fractionalization and violence, thereby refuting Hypothesis 1,
1a and 1b. The underlying idea that a stronger identification leads to a higher perceived threat and
less tolerance of outgroups members, particularly under pressure of war or violence (e.g. Bloom,
Arikan & Courtemanche, 2015; Djupe & Calfano, 2013; Hutchison & Gibler, 2006; Scheepers ,
Gijsberts & Hello, 2002; Simmel, 1955; Wilcox & Jelen, 1990) finds little general support in the MENA
context of religious societies. This does however not mean we should discard the belonging
hypothesis and threat perception mechanism for the MENA altogether. On the contrary, one
conditioning clause seems to help to fit the results here with the Western-based literature: the group
deviating from the societal religious norms is most subject to the threat mechanism. In this paper
this was shown by a negative relationship between being more religious and social tolerance among
non-Muslims minority members, which echoes Rizzo, Abdel-Latif & Meyer’s (2007: 1156) that the
perception of being part of a stigmatized group affects peoples’ democratic attitudes. In Western
countries it can than actually be the more religious groups who tend to feel threatened, leading to
seemingly contradictory results.
Also, while refuting the context-dependency hypothesis for Belonging, a pattern across
country years was found: if Islamist rule, it is more likely that a stronger belonging leads to less
tolerance. This suggests that when the regime is on the side of the most religious the latter
experience more freedom to not want to mix with people from other religious groups. Future
research might shed more light on how these two mechanisms discussed here interrelate, and
whether their simultaneous presence in some contexts but not in others explains the absence of an
overall effect in some cases but not in others.
Second, the impact of doctrinal orthodoxy was positive overall and rather consistently
positive across countries, and did not seem to depend on in-group and out-group orientation, partly
confirming Hypothesis 2. Most noteworthy is that Muslims who attach more value to living by the
rules of the Qur’an and Hadith seems to be more tolerance towards people with other ethno-
religious background and not just more benevolent to people of their own group. In other words, in
empirical terms the Islamic scriptures seem to teach not to exclude others and leave the judgement
to Allah. So while all major religions share a focus on compassion (Bloom, Arikan & Courtemanche,
2015: 205; Schwarts & Huismans 1995), the people whom it encompasses might actually be more
inclusive in contemporary Islam than in other religions. The context-dependency analysis, however,
did show a pattern of major exceptions: in places and time where orthodox Islam was supressed this
23
positive relationship turns strongly negative (confirming H2a1 and refuting H2a2). This mechanism
which seems to tap the threat mechanism again, might also help to identify groups in Western
democracies for which a similar impact can be expected, such the more doctrinal orthodox Jewish
people or Muslim immigrants – which also explain possible different effects among Muslims in
European countries and Muslims in MENA countries (cf. Dixon, 2009; Fish, 2011; Verkuyten et.al,
2014).
Third, I focussed on the impact of communal attendance from a socialization background and
the overall and country specific results largely confirm Hypothesis 3 by showing a negative
relationship between attendance and tolerance (see also Bloom & Arikan 2012; Scheepers, Gijsberts
& Hello, 2002); however further inspection of the relationship does indicate it duly matters what is
preached at the mosque and in line with H3a I find that in the more Islamist countries that also
control the Mosque sermons the socialization effect is particularly strong, and seem to drive the
overall relationship.
All in all, the results of this study show that it is a combination of identity-based threat,
socialization, and following scriptural rules that link people’s multifaceted religion to their social
tolerance of people belonging to other ethno-religious groups. By taking into account, which groups
are under pressure in society and which groups control government and what is preached, the
complex impact of religiosity can be understood rather well. As such, this study not only sheds more
light on how religiosity is working in varying ways in influencing tolerance across the divers group of
MENA countries. The results here also provides important insight into the conditions under which
results derived from analysis of western democracies can be generalized to other regions. In other
words, this study suggests which additional conditioning clauses help to lift the 3B theoretical
framework to a more universal framework, and which societal-level differences in religion’s position
are at the core of such conditioning clauses. Such an approach pushes our understanding of context-
dependent influences of religion on tolerance from claiming context-dependency to understanding it
in a systematic comparative way.
24
Bibliography
AB (2006-2014) [Arab Barometer] Wave I II and III. http://www.arabbarometer.org/instruments-and-data-files
Al Sadi, F. H., & Basit, T. N. (2013). Religious tolerance in Oman: addressing religious prejudice through educational intervention. British Educational Research Journal, 39(3), 447-472.
Allport, G. W., & Ross, J. M. (1967). Personal religious orientation and prejudice. Journal of personality and social psychology, 5(4), 432.
ARDA (2017). GRRD data. Data were downloaded from the Association of Religion Data Archives, www.TheARDA.com, and were collected by Pew Research Center. Accessed http://www.thearda.com/Archive/Files/Descriptions/GRELREST.asp on 20170302.
Baderin, M. A. (2003). International human rights and Islamic law. OUP Oxford.
Bloom, P. B. N., & Arikan, G. (2012). A two-edged sword: The differential effect of religious belief and religious social context on attitudes towards democracy. Political Behavior, 34(2), 249-276.
Bloom, P. B. N., & Bagno-Moldavsky, O. (2015). The conditional effect of network diversity and values on tolerance. Political Behavior, 37(3), 623-651.
Bloom, P. B. N., Arikan, G., & Courtemanche, M. (2015). Religious social identity, religious belief, and anti-immigration sentiment. American Political Science Review, 109(02), 203-221.
Brown, R. (2000). Social identity theory: Past achievements, current problems and future challenges. European journal of social psychology, 30(6), 745-778.
Boucek, C. (2010). Saudi Fatwa Restrictions and the State-Clerical Relationship. 2010/10/27 Carnegie Endowment for International Peace. Accessed: http://carnegieendowment.org/2010/10/27/saudi-fatwa-restrictions-and-state-clerical-relationship/6b81 2017/02/28
Camilleri, C., & Malewska-Peyre, H. (1997). Socialization and identity strategies. Handbook of cross-cultural psychology, 2, 41-67.
Ceobanu, A. M., & Escandell, X. (2010). Comparative analyses of public attitudes toward immigrants and immigration using multinational survey data: A review of theories and research. Annual review of sociology, 36, 309-328.
Ciftci, S. (2010). Modernization, Islam, or social capital: what explains attitudes toward democracy in the Muslim world? Comparative Political Studies, 43 (11), 1442-1470.
Cigler, A., & Joslyn, M. R. (2002). The extensiveness of group membership and social capital: The impact on political tolerance attitudes. Political Research Quarterly, 55(1), 7-25.
Dalacoura, K. (2012). The 2011 uprisings in the Arab Middle East: political change and geopolitical implications. International Affairs, 88(1), 63-79.
Davis, N. J., & Robinson, R. V. (2006). The egalitarian face of Islamic orthodoxy: Support for Islamic law and economic justice in seven Muslim-majority nations. American Sociological Review, 71(2), 167-190.
Dixon, J. C. (2008). A clash of civilizations? Examining liberal-democratic values in Turkey and the European Union1. The British journal of sociology, 59(4), 681-708.
Djupe, P. A., & Calfano, B. R. (2013). Religious value priming, threat, and political tolerance. Political Research Quarterly, 66(4), 768-780.
Eisenstein, M. A. (2006). Rethinking the relationship between religion and political tolerance in the US. Political Behavior, 28(4), 327-348.
El Fadl, K. A. (2002). The Orphans of Modernity and the Clash of Civilisations. Global Dialogue, 4(2), 1.
25
El-Ghobashy, M. (2005). The metamorphosis of the Egyptian Muslim brothers. International Journal of Middle East Studies, 37(03), 373-395.
Fish, M. S. (2011). Are Muslims distinctive?: a look at the evidence. Oxford: Oxford University Press.
Fish, M. S. (2002). Islam and Authoritarianism. World Politics 55: 4–37.
Geaves, R. (2006). Key words in Islam. London: Bloomsbury Publishing.
Gengler, J. J. (2013). Royal Factionalism, the Khawalid, and the Securitization of ‘the Shīʿa Problem’in Bahrain. Journal of Arabian Studies, 3(1), 53-79
Gibson, J. L. (1992). Alternative Measures of Political Tolerance: Must Tolerance be" Least-Liked"?. American Journal of Political Science, 560-577.
Gibson, J. L. (2006). Do strong group identities fuel intolerance? Evidence from the South African case. Political Psychology, 27(5), 665-705.
Glock, C. Y., & Stark, R. (1965). Religion and society in tension. Chicago, IL: Rand McNally.
Guveli, A. (2015). Are movers more religious than stayers? Religiosity of European majority, Turks in Europe and Turkey. Review of Religious Research, 57(1), 43-62.
Harik, I. F. (1994). Pluralism in the Arab world. Journal of Democracy, 5(3), 43-56.
Hassan, R. (2007). On Being Religious: Patterns of Religious Commitment in Muslim Societies1. The Muslim World, 97(3), 437-478.
Hoffman, S. (2004). Islam and democracy: Micro-level indications of compatibility. Comparative Political Studies, 37, 652-676.
Hutchison, M. L., & Gibler, D. M. (2007). Political tolerance and territorial threat: A cross-national study. The Journal of Politics, 69(1), 128-142.
Jamal, A. A., & Tessler, M. A. (2008). Attitudes in the Arab world. Journal of Democracy,19(1), 97-110.
Kellstedt, L. A., Green, J. C., Guth, J. L., & Smidt, C. E. (1996). Grasping the essentials: The social embodiment of religion and political behavior. In: J. C. Green (1996). Religion and the culture wars: Dispatches from the front. (pp. 174-192). Lanham, MD: Rowman & Littlefield
Leege, D. C., & Kellstedt, L. A. (1993). Rediscovering the religious factor in American politics. ME Sharpe.
Lipset, S. M. (1959). Some social requisites of democracy: Economic development and political legitimacy. American political science review, 53(01), 69-105.
Mill, J. 2007. Utilitarianism, Liberty and Representative Government. Rockville, MD: Wildside Press.
Moaddel, M. (2002). The study of Islamic culture and politics: An overview and assessment. Annual Review of Sociology, 28(1), 359-386.
Moghadam, V. M. (2013). What is democracy? Promises and perils of the Arab Spring. Current Sociology, 61(4), 393-408.
Netterstrøm, K. L. (2015). The Islamists' Compromise in Tunisia. Journal of Democracy, 26(4), 110-124.
Norenzayan, A. (2014). Does religion make people moral?. Behaviour, 151(2-3), 365-384.
Owen, R. 2004. State, Power and Politics in the Making of the Modern Middle East. London: Routledge.
Posusney, M. P. (2004). Enduring authoritarianism: Middle East lessons for comparative
26
Putnam, R. D., Leonardi, R., & Nanetti, R. Y. (1994). Making democracy work: Civic traditions in modern Italy. Princeton university press.
Rawls, J. (2003). The Law of The Peoples With “The Idea Of Public Reason Revisited.” Harvard University Press.
Rizzo, H., Abdel-Latif, A. H., & Meyer, K. (2007). The relationship between gender equality and democracy: A comparison of Arab versus non-Arab Muslim societies. Sociology, 41(6), 1151-1170.
Said, E. (1978). Orientalism: Western concepts of the Orient. Delhi: Penguin, 1995, 2001.
Sarkissian, A. (2012). Religion and civic engagement in Muslim countries. Journal for the Scientific Study of Religion, 51(4), 607-622.
Schedler, A., & Sarsfield, R. (2007). Democrats with adjectives: Linking direct and indirect measures of democratic support. European Journal of Political Research, 46(5), 637-659.
Scheepers, P., Gijsberts, M., & Hello, E. (2002). Religiosity and prejudice against ethnic minorities in Europe: Cross-national tests on a controversial relationship. Review of Religious Research, 242-265.
Schwartz, S. H., & Huismans, S. (1995). Value priorities and religiosity in four western religions. Social Psychology Quterly, 88-107.
Siegel, T. D. (2015). Does Existential Security Affect Tolerance in the Middle East? (master thesis) Faculty of the Graduate School of Arts and Sciences of Georgetown University, Wastington, D.C.
Simmel, G. (1955). The web of group affiliations. Conflict and the web of group affiliations, 125-95.
Smidt, C., Kellstedt, L., & Guth, J. L. (Eds.). (2009). The Oxford handbook of religion and American politics. Oxford University Press.
Spierings, N., Smits, J., & Verloo, M. (2009). On the compatibility of Islam and gender equality. Social Indicators Research, 90(3), 503-522.
Spierings, N., Smits, J., & Verloo, M. (2010). Micro‐and Macrolevel Determinants of Women's Employment in Six Arab Countries. Journal of Marriage and Family, 72(5), 1391-1407.
Spierings, N. (2014). Islamic Attitudes and the Support for Gender Equality and Democracy in Seven Arab Countries, and the Role of Anti-Western feelings. GÉNEROS-Multidisciplinary Journal of Gender Studies 3, 423–456
Spierings, N. (2015). Women’s Employment in Muslim Countries: Patterns of Diversity. Springer.
Spierings, N. (2016). Multilevel Analysis as a Tool to Understand the Spatio-Temporality of Gender. Politics & Gender, 12(3).
Spierings, N. (2017). Trust and Tolerance across the Middle East and North Africa: A Comparative Perspective on the Impact of the Arab Uprisings. Politics and Governance, 5(2), 4-15.
Stark, R., & Glock, C. Y. (1968). American piety: The nature of religious commitment (Vol. 1). Jackson, CA: Univ of California Press.
Strabac, Z., & Listhaug, O. (2008). Anti-Muslim prejudice in Europe: A multilevel analysis of survey data from 30 countries. Social Science Research, 37(1), 268-286.
Stepan, A. C., & Robertson, G. B. (2003). An" Arab" more than a" Muslim" democracy gap. Journal of Democracy, 14(3), 30-44.
Sullivan, J., and J. Transue. 1999. “The Psychological Underpinnings of Democracy: A Selective Review of Research on Political Tolerance, Interpersonal Trust, and Social Capital.” Annual review of psychology, 50, 625–650.
27
Sullivan, J. L., Shamir, M., Walsh, P., & Roberts, N. S. (1985). Political tolerance in context. Boulder, CO: Westview, 51-76.
Tajfel, H. (1981). Human groups and social categories: Studies in social psychology. CUP Archive.
Tessler, M. (2002). Islam and democracy in the Middle East: The impact of religious orientations on attitudes toward democracy in four Arab countries. Comparative Politics, 337-354.
Testas, A. (2005). The economic costs of transition to democracy in the Maghrib: Quantitative results for Algeria and Tunisia. The Journal of North African Studies, 10(1), 43-59.
Testas, A. (2002). The roots of Algeria's religious and ethnic violence. Studies in Conflict and Terrorism, 25(3), 161-183.
Verkuyten, M., Maliepaard, M., Martinovic, B., & Khoudja, Y. (2014). Political tolerance among Muslim minorities in Western Europe: The role of denomination and religious and host national identification. Politics and Religion, 7(02), 265-286.
Wald, K. D., & Smidt, C. E. (1993). Measurement strategies in the study of religion and politics. D. C. Leege, & L. A. Kellstedt, (1993). Rediscovering the religious factor in American politics. ( pp. 26-49). ME Sharpe.
Wald, K. D., & Wilcox, C. (2006). Getting religion: has political science rediscovered the faith factor?. American Political Science Review, 100(04), 523-529.
Wilcox, C., & Jelen, T. (1990). Evangelicals and political tolerance. American Politics Quarterly, 18(1), 25-46.
Wilson, J. M., Smidt, C., Kellstedt, L., & Guth, J. (2009). The Oxford Handbook of Religion and American Politics.
WVS (1981-2014) [WORLD VALUES SURVEY] LONGITUDINAL AGGREGATE v.20150418. World Values Survey Association (www.worldvaluessurvey.org). Aggregate File Producer: JDSystems, Madrid SPAIN.
28
Appendix A1. Survey information Country Year Survey
organisation Wav
e1
N N in analyses
Survey weight
Algeria 2002 WVS 4 1,282 1,164 N 2006 AB 1 1,300 667 N 2011 AB 2 1,216 911 Y 2013 WVS 6 1,200 1,066 N Bahrain 2009 AB 1 435 340 N Egypt 2001 WVS 4 3,000 2,864 N 2011 AB 2 1,219 1,172 Y Iraq 2004 WVS 4 2,325 2,176 N 2006 WVS 5 2,701 2,078 N 2011 AB 2 1,234 1,064 Y 2013 WVS 6 1,215 1,123 N Jordan 2001 WVS 4 1,223 1,189 N 2006 AB 1 1,143 911 N 2010 AB 2 1,188 1,026 Y 2014 WVS 6 1,200 1,199 N Lebanon 2007 AB 1 1,195 954 N 2011 AB 2 1,387 1,308 Y 2013 WVS 6 1,200 1,002 N Libya 2014 WVS 6 2,131 1,751 Y Morocco 2001 WVS 4 1,251 1,190 Y 2006 AB 1 1,277 942 N 2007 WVS 5 1,200 1,157 N Palestine 2010 AB 2 1,200 1,053 Y 2013 WVS 6 1,000 889 N Saudi Arabia 2003 WVS 4 1,502 1,337 N 2011 AB 2 1,404 1,068 Y Sudan 2011 AB 2 1,538 1,245 Y Tunisia 2011 AB 2 1,196 1,102 Y 2013 WVS 6 1,205 1,156 N Yemen 2007 AB 1 717 514 N 2011 AB 2 1,200 886 Y 2014 WVS 6 1,000 929 N 43,484 37,413
(86%)
29
Appendix B1. Items included in operationalization of Tolerance
Does (not) object to mentioned type of neighbour
Country year Oth
er r
ace
or
colo
r
Mig
ran
t o
r
qu
est/
exp
atri
ate
wo
rker
Jew
Oth
er r
elig
ion
Oth
er c
ou
ntr
y
Ch
rist
ian
s
Ku
rds
Oth
er la
ngu
age
Sun
nis
Shiit
es
Ref
uge
s
Imm
igra
nts
Oth
er r
elig
iou
s se
ct
Algeria 2002 X X X Algeria 2006 X X X Algeria 2011 X X X X Algeria 2013 X X X X Bahrain 2009 X X X Egypt 2001 X X X Egypt 2011 X X X X Iraq 2004 X X X X X X Iraq 2006 X X X X X X Iraq 2011 X X X X X X Iraq 2013 X X X X Jordan 2001 X X X X Jordan 2006 X X X Jordan 2010 X X X X Jordan 2014 X X X X Lebanon 2007 X X X Lebanon 2011 X X X X Lebanon 2013 X X X X Libya 2014 X X X X Morocco 2001 X X Morocco 2006 X X X Morocco 2007 X X X X Palestine 2010 X X X Palestine 2013 X X X X Saudi Arabia 2003 X X X Saudi Arabia 2011 X X X X Sudan 2011 X X X X Tunisia 2011 X X X X Tunisia 2013 X X X X Yemen 2007 X X X Yemen 2011 X X X X Yemen 2014 X X X X
30
Appendix B2. Items included in operationalization of Religiosity dimensions
Belonging Doctrinal orthodoxy Religious ingroups orientation Attendance
Country year To w
hat
ext
ent
do
yo
u c
on
sid
ers
you
rsel
f as
relig
iou
s (n
ot,
so
mew
hat
, rel
igio
us)
Are
yo
u a
rel
igio
us
per
son
? Ye
s, n
o (
incl
co
nvi
nce
d
ath
eist
)
Ho
w im
po
rtan
t is
re
ligio
n in
yo
ur
life?
(0
-3)
Div
orc
e
Suic
ide
Dri
nki
ng
alco
ho
l
Euth
anas
ia
Par
tici
pat
e in
a lo
tter
y
Ch
argi
ng
inte
rest
co
ntr
adic
ts Is
lam
Ch
argi
ng
inte
rest
mu
st b
e fo
rbid
den
no
t p
rayi
ng
is a
n o
bst
acle
in a
ccep
tin
g th
e m
arri
age
of
firs
t d
egre
e r
elat
ive
(0-3
)
fro
m d
iffe
ren
t re
ligio
n/d
eno
min
atio
n is
ob
sta
cle
in
acce
pti
ng
mar
riag
e o
f fi
rst
deg
ree
re
lati
ve
(0-3
)
Imp
ort
ant
child
qu
alit
ies:
rel
igio
us
(no
/yes
)
Imp
ort
ant
for
con
sid
erin
g w
ho
is a
su
itab
le s
po
use
for
son
or
dau
ghte
r: p
rayi
ng
(0-3
)
Imp
ort
ant
for
con
sid
erin
g w
ho
is a
su
itab
le s
po
use
for
son
or
dau
ghte
r: f
asti
ng
(0
-3)
Mo
st im
po
rtan
t fo
r co
nsi
der
ing
wh
o is
a s
uit
able
spo
use
fo
r so
n/d
augh
ter
(fas
tin
g o
r p
rayi
ng
)
Do
yo
u a
tten
d F
rid
ay p
raye
r? (
0-4
)
Ho
w o
ften
do
yo
u a
tten
d r
elig
iou
s se
rvic
es?
(0-4
)
Do
yo
u p
ray
in C
hu
rch
or
Mo
squ
e (n
o, y
es)
Ho
w o
ften
do
yo
u a
tten
d F
rid
ay p
raye
r o
r Su
nd
ay
serv
ices
(0
-3)
Algeria 2002 X X X X X X X X Algeria 2006 X X X X X X X Algeria 2011 X X X X X Algeria 2013 X X X X X X X Bahrain 2009 X X X X X X X Egypt 2001 X X X X X X X X Egypt 2011 X X X X Iraq 2004 X X X X X X Iraq 2006 X X X X X Iraq 2011 X X X X X Iraq 2013 X X X X X X X X Jordan 2001 X X X X X X X X Jordan 2006 X X X X X X X Jordan 2010 X X X X X Jordan 2014 X X X X X X Lebanon 2007 X X X X X X X Lebanon 2011 X X X X X Lebanon 2013 X X X X X X X Libya 2014 X X X X X X X Morocco 2001 X X X X X X Morocco 2006 X X X X X X X Morocco 2007 X X X X X X X Palestine 2010 X X X X X Palestine 2013 X X X X X Saudi Arabia 2003 X X X X X X X Saudi Arabia 2011 X X X X X Sudan 2011 X X X X X Tunisia 2011 X X X X X Tunisia 2013 X X X X X X X Yemen 2007 X X X X X X X Yemen 2011 X X X X X Yemen 2014 X X X X X X X
31
Appendix C1. Descriptive statistics of micro-level variables
Variable Mean s.d. Min Max Frequency
TOLERANCE
Ethno-religious social tolerance 0.00 0.73 -3.62 2.05
RELIGION
Belonging (Religious identification) 0.02 0.86 -11.77 1.60
Majority status
Muslim 94.3
Non-Muslims 5.7
Doctrinal orthodoxy 0.01 0.83 -2.94 6.13
Religious in-group orientation 0.02 0.91 -4.28 2.04
Attendance 0.00 1.00 -5.74 1.68
CONTROLS
Employment
Employed 62.6
Retired 3.7
Housewife 20.1
Student 6.6
Not employed other 7.0
Education
No education 27.7
Primary completed 22.3
Secondary completed 27.8
At least some tertiary 22.2
Sex
Male 51.2
Female 48.8
Age (in 7 cat.) 1.77 1.42 0 6
Marital status
Married 64.5
Single 30.4
Other (incl. divorced, widowed) 5.1
32
Appendix C2. Macro-level scores per country or country year
Country year
Religious Fractionalizationa
Religious-based violenceb
Orthodox Islam being suppressedc
Islamist regulated sermonsd
Algeria 0.0091 0.54 Algeria 2002 0 0 Algeria 2006 0 0 Algeria 2011 0 1 Algeria 2013 0 1
Bahrain 0.5528 0.39 Bahrain 2009 1 1
Egypt 0.1979 0.77 Egypt 2001 1 0 Egypt 2011 0 0
Iraq 0.4844 0.80 Iraq 2004 0 0 Iraq 2006 0 0 Iraq 2011 0 0 Iraq 2013 0 0
Jordan 0.0659 0.44 Jordan 2001 0 0 Jordan 2006 0 0 Jordan 2010 0 0 Jordan 2014 0 0
Lebanon 0.7886 0.63 Lebanon 2007 0 0 Lebanon 2011 0 0 Lebanon 2013 1 0
Libya 0.0570 0.82 Libya 2014 0 0
Morocco 0.0035 0.33 Morocco 2001 1 0 Morocco 2006 1 0 Morocco 2007 1 0
Palestine 0.1719 0.74 Palestine 2010 -1 0 Palestine 2013 -1 0
Saudi Arabia 0.1270 0.59 Saudi Arabia 2003 -1 0 Saudi Arabia 2011 -1 1
Sudan 0.43071 0.58 Sudan 2011 -1 0
Tunisia 0.0104 0.43 Tunisia 2011 -1 0 Tunisia 2013 0 0
Yemen 0.0023 0.66 Yemen 2007 -1 0 Yemen 2011 -1 1 Yemen 2014 -1 1
A) Source: Alesina et.al (2003); B) Source: Own calculations based on GRRD data; C) Source: coded by author; D) Source: based on GRRD, Spierings, Smits & Verloo (2009) and Spierings (2015) Notes: 1) This figures is from before South Sudan became an independent country; however, taking this into account when assessing the macro-level plots does not suggest this has caused refuting the hypothesis on the moderation impact of religious fractionalization.