INTRODUCTION
The term “ambivalence of the sacred” refers to the inherent duality or contradictory nature of religious beliefs and their potential effects on violence (Appleby Reference Appleby2000). It suggests that religious factors can simultaneously contribute to both peaceful and violent outcomes, leading to conflicting interpretations and understandings. Religion, as a complex and multifaceted phenomenon, can be both a source of inspiration for harmony, compassion, and social cohesion and a catalyst for division, conflict, and violence. Yet, such ambivalence arises from the fact that religious beliefs and institutions can be interpreted and employed in various ways (e.g., Appleby Reference Appleby2000; Philpott Reference Philpott2007).
The al-Qaeda attacks on September 11, 2001, the Islamic State in Iraq and Syria (ISIS)'s series of brutal violence in Iraq and Syria, as well as in Europe in 2015 and 2016, and the kidnappings of hundreds of schoolgirls by Boko Haram in Nigeria in 2014 did not only shake the world with their immense human suffering and destruction but are also considered pivotal incidents in the global landscape of jihadist violence. The attacks carried out by Salafi-jihadist extremists brought the issue of religiously motivated violence to the forefront of international consciousness (Cubukcu and Forst Reference Cubukcu and Forst2018). Alongside the spikes in jihadist violence, there has also been a growing recognition of the importance of understanding sympathy among certain segments of the society for terrorist tactics and groups, as it plays a crucial role in perpetuating and sustaining extremism and providing recruitment pools and safe havens for groups (Ozeren, Cubukcu, and Bastug Reference Ozeren, Cubukcu, Bastug, Goldstone, Alimi, Ozeren and Cubukcu2021).
As the world has been grappling with an extensive number of religiously motivated violence, there is a need for nuanced analysis and contextual understanding of the role of religion in relation to violence, recognizing that religious beliefs and practices can have diverse and sometimes contradictory implications for peace and conflict. A key challenge faced by proponents of both viewpoints lies in understanding the individual-level relationship between religion and violence.
Empirical studies have similarly yielded mixed results, with some highlighting the active involvement of religious adherents, leaders, and institutions in promoting violence. In contrast, others emphasize the moderating influence of religious beliefs and institutions. This study aims to address the paradoxical nature of religion’s impact on violence by distinguishing religion as a belief from religion as an ideology. Specifically, we examine individual sympathy for al-Qaeda and justification for suicide bombings by providing a more nuanced understanding of the role of religion by distinguishing religiosity and political Islam from each other. As we do so, we take into account the role of political and economic grievances on public attitudes towards terrorism and examine how these relationships vary across different Muslim-majority countries.
BACKGROUND AND LITERATURE REVIEW
The radicalization approach considers terrorism an identity-building process that extends from sympathy for terrorism to involvement in terrorist actions. The literature on radicalization argues that sympathy for terrorist activities and terrorist groups sometimes continues with involvement in terrorist activities (United Kingdom Government 2009, 2018). This approach has caused a shift in terrorism studies and turned the attention from terrorists to a broader population of those who sympathize with or justify terrorism. However, the number of studies investigating the factors leading to public support for terrorism is limited. Also, the literature tends to fail to balance macro- and micro-level causes leading to a radicalization pathway (Cinoglu Reference Cinoglu2010).
McCauley (Reference McCauley2012) identifies the specific elements of the “global-jihad” frame that can explain individual sympathy for “jihadist” terrorism. He provides an overview of three prevalent notions regarding the origins of terrorism: discrimination, Radical Islam, and grievance. His analysis, based on polls conducted among Muslim populations in the United States, aims to shed light on the extent to which these factors play a role in shaping respondents’ global-jihad perspective. However, McCauley’s study (McCauley Reference McCauley2012) presents two major issues concerning the conceptualization of outcome and predictor variables.
Firstly, McCauley (Reference McCauley2012) used negative attitudes towards the “War on Terror” as a proxy for sympathy for terrorism. To measure individuals’ sympathy with terrorism, he utilized questions like “Is the War on Terror a real War on Islam?” and “Is the War on Terror an insincere effort to combat terrorism?” (McCauley Reference McCauley2012:317). However, this is a problematic measurement because expressing opinions about the United States counter-terrorism policies does not inherently indicate direct support for terrorism.
McGarty, Thomas, and Louis (Reference McGarty, Thomas and Louis2012) initially criticized McCauley for a weak conceptualization of sympathy for terrorism. They argue that the data presented by McCauley do not address the predictors of sympathy for terrorism in a comprehensive manner. Instead, certain aspects of McCauley’s findings inadvertently conflate opposition to the “War on Terror” with sympathy for terrorism, while these concepts are distinct phenomena. Individuals who express opposition to the War on Terror may not necessarily harbor sympathetic inclinations toward acts of terrorism. While a significant proportion of individuals may express disagreement with the government’s foreign and military policies, the number of those genuinely sympathizing with terrorism is minimal. Instead, their perceptions are shaped by several other factors.
In line with the arguments presented by Agnew (Reference Agnew2010), it is crucial to recognize that political grievances, such as negative attitudes toward foreign and domestic policies, play a significant role in driving acts of terrorism. Therefore, opposition to counter-terrorism policies should not be misconstrued as support for terrorism, but rather as an expression of political grievances, or even a disagreement with the violent tactics employed.
The second problematic conceptualization in McCauley’s article (Reference McCauley2012) is that he considers religiosity as an indicator of an “extremist form of Islam” (McCauley Reference McCauley2012:300). He refers to military terms and argues that “Radical Islam” is the primary source of terrorism. He mentions that Radical Islam is about politics and individuals’ religiosity, public morality, and “religious expression in a life-dominating Puritan Islam” (McCauley Reference McCauley2012:300). According to McCauley (Reference McCauley2012), the ones who support the ideology of Political Islam “show high religiosity, including reporting religion as an important part of their life and frequent participation in prayers at their mosque or Islamic center” (McCauley Reference McCauley2012:300).
However, this represents an erroneous and narrow conceptualization of religiosity. Religiosity encompasses the depth and extent of individuals’ adherence to religious beliefs and their engagement in religious practices. On the other hand, “political Islam” refers to the integration of the Islamic religion into the secular domain of politics, with the strategic use of Islam as a means to achieve political goals (Ozeren, Bastug, and Cubukcu Reference Ozeren, Bastug, Cubukcu, Orofino and Allchorn2023). This includes the involvement of Islam in political and governmental activities, as well as opposition movements. It is important to establish a clear distinction between religiosity, which encompasses personal expressions of piety and the commitment to religious beliefs and rituals (Hirschkind Reference Hirschkind1997), and “political Islam,” which involves the instrumental utilization of Islam for political ends. This instrumental use can be exemplified by political parties such as Hizb al-'Amal in Egypt, the Islamic Salvation Front (FIS) in Algeria, or the Justice and Development Party (AKP) in Turkey.
Studies that adopt a narrow and limited perspective of Islam primarily as a political system (e.g., McCauley Reference McCauley2012) are problematic. Islam is a religion that encompasses a comprehensive set of values, beliefs, and practices. It is essential to distinguish between “the abuse of Islam in politics,” which treats Islam as a political ideology (Yilmaz Reference Yilmaz2005), and the genuine practice of devout Muslims. Yilmaz (Reference Yilmaz2005) emphasizes this distinction, highlighting the contrast between “political Islam” and “mainstream Islam,” the latter of which emphasizes individual and societal responsibility for beliefs and actions while advocating for the establishment of State rules based on social consensus and shared values.
In this perspective, Islam and universal values such as justice, freedom, the rule of law, human rights, and democracy are not perceived as opposing forces. And, importantly, acts of violence carried out by Salafi-jihadist extremists against Westerners are inherently against the fundamental principles of Islam. Religious involvement and spirituality make individuals refrain from violent acts against innocent people, even though Muslims involved in terrorist activities were influenced mainly by religious ideologies (Esposito Reference Esposito2002). Many terrorist acts are undertaken by groups or individuals who claim to be religiously motivated and inspired by political interpretations of Islam (Atran Reference Atran2002; Berman Reference Berman2003). However, it is important to note that these motivations may not solely be religious in nature. Various social, economic, political, cultural, or psychological factors can also serve as driving forces for individuals to join extremist organizations and engage in acts of violence. In such cases, religion may be utilized as a prominent predictor variable, but without considering the influence of other variables, the significance of religion may be overestimated.
Individuals’ motivations for joining religiously oriented terrorist organizations can vary. For instance, research conducted by the International Center for Terrorism and Transnational Crime reveals that some individuals who joined the Kurdish Hizbullah, a religiously motivated terrorist organization, were driven not by religion, but by fear and the desire for protection from the Kurdistan Workers’ Party (PKK) (Demir, Basibuyuk, and Karakus Reference Demir, Basibuyuk, Karakus, Bal, Ozeren and Sozer2011). In fact, 76% of the members of Hizbullah’s military wing reported having a low or no level of religiosity at the time of joining the organization (Demir et al. Reference Demir, Basibuyuk, Karakus, Bal, Ozeren and Sozer2011).
Tessler and Robbins (Reference Tessler and Robbins2007) studied Muslim populations in Jordan and Algeria and investigated individuals’ support for terrorist activities affiliated with Radical Islam and their grievances. They found no effect of gender, economic grievance, religiosity, and Radical Islam on public support for terrorism. However, they found that individuals’ negative perceptions about local and foreign politics increased the likelihood of approval of terrorist actions. In a similar vein, several empirical studies suggest no significant relationship between religiosity and support for Islamist militancy and violence (e.g., Fair, Littman, and Nugent Reference Fair, Littman and Nugent2018; Fair, Malhotra, and Shapiro Reference Fair, Malhotra and Shapiro2012; Kaltenthaler et al. Reference Kaltenthaler, Miller, Ceccoli and Gelleny2010; Tessler and Nachtwey Reference Tessler and Nachtwey1998). Accordingly, the vast majorities of Muslims are against terrorism and are appalled by using of violence in the name of Islam (Matthes et al. Reference Matthes, Kaskeleviciute, Schmuck, von Sikorski, Klobasa, Knupfer and Saumer2020).
The literature on religiously motivated extremism presents a range of pull and push factors that have been identified as influential in the process of individuals’ radicalization. One major significant push factor is the presence of economic and political grievances. Agnew (Reference Agnew2010) argues that collective strains serve as a central intervening mechanism that links strains to criminal behavior. These strains give rise to negative emotions, including anger, hopelessness, and humiliation. Such emotions create pressure on individuals to address the impact of collective strains through violent means (Hoffman Reference Hoffman and Hoffman2006; Mogdaham Reference Moghadam2006; Pape Reference Pape2005). Individuals’ willingness to seek revenge for collective strains serves as a strong motivation for engaging in violent actions, even when they are aware of the undesirable consequences that may result from their acts of revenge and vengeance (Araj Reference Araj2008).
In addition to economic and political grievances, such as inequality and relative deprivation, negative attitudes toward foreign and domestic policies are widely acknowledged as underlying factors contributing to the process of radicalization. These factors can result in an individual’s isolation from mainstream society while simultaneously providing a motive for engagement with terrorist organizations (Hudson Reference Hudson1999). Moreover, Crenshaw (Reference Crenshaw1985) proposed that “the motivation needed for individuals to join terrorist organizations can be broken down into four categories: (1) opportunity for action; (2) need for a sense of belonging; (3) need for social status; (4) prospect of financial income” (as cited in Ozeren et al. Reference Ozeren, Sever, Yilmaz and Sozer2014:324).
These negative attitudes significantly influence individuals’ perceptions and can contribute to their inclination toward embracing extremist ideologies and actions (Muibu & Cubukcu Reference Muibu and Cubukcu2021). Consequently, in studies on radicalization, political grievances related to counter-terrorism policies should not be viewed as an indicator of an individual’s sympathy for terrorism. Instead, they should be considered explanatory (independent) variables that help elucidate the process of radicalization.
This study examines the public sympathy for terrorism and terrorist groups, taking into consideration the perspective of religiosity and religious ideology, as well as economic and political grievances. This study addresses the issues inherent in McCauley’s (Reference McCauley2012) conceptualization of the outcome variable, “sympathy with terrorism,” and the explanatory variable, “Radical Islam.” We establish a clear distinction between the concepts of religiosity and political Islam and investigate their respective associations with public sympathy for terrorism. Moreover, this research broadens the scope of analysis by utilizing representative survey data from five countries with majority Muslim populations. By encompassing a diverse range of countries, the study explores the factors that may contribute to public sympathy for terrorism across various cultural and social contexts.
We conducted an independent examination of the hypotheses concerning two dependent variables: individuals’ justification of suicide bombings and sympathy for al-Qaeda. The hypotheses under investigation are as follows:
H1a-b: The level of individuals’ ideology on political Islam is related to the probability of justification of suicide bombings/sympathy with al-Qaeda.
H2a-b: The level of religiosity is related to the probability of justification of suicide bombings/sympathy with al-Qaeda.
H3a-b: Economic grievances are related to the probability of justification of suicide bombings/sympathy with al-Qaeda.
H4a-b: Political grievances against Western societies are related to the probability of justification of suicide bombings/sympathy with al-Qaeda.
DATA AND MEASUREMENTS
The study utilized a subset of the 2011 Pew Global Attitudes Survey, specifically focusing on countries with significant Muslim populations. The subsample consisted of Muslims residing in Egypt (N=699), Indonesia (N=562), Jordan (N=729), Lebanon (N=474), and Turkey (N=536). The data collected for this research were cross-sectional in nature and obtained through face-to-face surveys. The sampling process employed a randomized multi-stage cluster technique. The dataset used in the study was obtained from the Pew research website.
Dependent Variables
The dependent variables, namely individuals’ justification of suicide bombings and support for al-Qaeda, are categorized as ordinal-level variables with four distinct categories. To address the first research question, the association between these two variables at the ordinal level is examined using Spearman’s rho correlation coefficient. The results of Spearman’s rho test indicate that there is no significant association between individuals’ justification of suicide bombings and their support for al-Qaeda (Spearman’s rho = –0.02, p = 0.27). It is important to note that these two concepts are different concepts. Justification of suicide bombings refers to the endorsement of a strategy and tactic employed by terrorist organizations, whereas sympathy for al-Qaeda represents a favorable view specifically towards this particular terrorist organization. Given the notable distinction between these concepts, the relationships between the predictor variables and each of these dependent variables were estimated separately.
Independent Variables
In the dataset, religiosity, economic grievances, and political grievances are the concepts that are assessed using multiple variables. Confirmatory factor analysis is used to calculate the factor loadings, and a factor-based scale is used to construct indexes of variables that measure similar concepts.
The predictor variables are derived from the radicalization framework utilized by Tessler and Robbins (Reference Tessler and Robbins2007) as well as McCauley (Reference McCauley2012). As discussed in the literature review, items related to religion are divided into two distinct concepts: individuals’ ideology of political Islam and their level of religiosity. Political Islam is gauged based on respondents’ perspectives on whether laws should strictly adhere to the teachings of the Quran or not.
Religiosity, on the other hand, pertains to one’s involvement and dedication to religious practices rather than their ideology of political Islam. It is measured using three indicators: individual engagement in prayer and fasting, and the significance of religion in their daily lives. These indicators are combined to form a scale representing religiosity (Tessler and Robbins Reference Tessler and Robbins2007).
The variables associated with grievances can be categorized into two groups: economic grievances and political grievances. Economic grievances are assessed by measuring the extent of deprivation concerning food, medication, and clothing, with high factor loadings (>0.77). Political grievances are evaluated using four indicators: negative attitudes towards the war in Afghanistan, perception of the United States as a threat, and perceived hostility from Europeans and Americans (McCauley Reference McCauley2012; Tessler and Robbins Reference Tessler and Robbins2007). Confirmatory factor analysis indicates that these four variables load onto a single factor representing political grievances against Western societies and politics.
Control Variables
The literature on terrorism posits that individuals have the potential to shape their perspectives on terrorism under the influence of various social factors, including friends, family members, school environments, community members, religious figures, and numerous media sources (Forst Reference Forst2008). To account for the relationship between dependent and independent variables, country dummy variables are utilized as control variables. Furthermore, gender, age, and sectarian groups are included as additional controls in the analysis. The sectarian groups are represented using dummy variables, with Sunnis coded as 1 and other groups coded as 0.
ANALYTIC STRATEGY
The quantitative analyses were conducted in two stages, with the first stage involving descriptive statistics and the second stage involving multivariate analyses. Cases with missing values were excluded from the sample. Descriptive statistics for the study variables are presented in Table 1.
Table 1. Descriptive Statistics for Study Variables (N = 3000, Listwise)
Given that the dependent variables in this study have ordinal scales, the ordered logistic method was employed for multivariate estimation. The assumption of parallel regression, which assumes a similar relationship between each pair of outcome groups, was tested using the “oparallel” command in Stata 12. The Brant test indicated a violation of the parallel regression assumption in the models.
To address potential bias in the ordered logit model, the dependent variables were dichotomized. For the first dependent variable, justification of suicide bombings, two categories were created. Categories 1 (often justified) and 2 (sometimes justified) were combined and coded as 1, indicating “justified.” Categories 3 (rarely justified) and 4 (never justified) were combined and coded as 0, indicating “not justified.”
Similarly, for the second dependent variable, individuals’ sympathy with al-Qaeda, two categories were formed. Categories 1 (very favorable) and 2 (somewhat favorable) were combined and coded as 1, indicating “favorable.” Categories 3 (somewhat unfavorable) and 4 (very unfavorable) were combined and coded as 0, indicating “unfavorable.”
To examine the research hypotheses using aggregated data from six different countries, multivariate regression models will be utilized as formulated below:
$${{\rm{Y}}_{\rm{i}}} = {\rm \beta _0} + {\rm \beta _{\rm{1}}}{{\rm{R}}_{\rm{i}}} + {\rm \beta _{\rm{2}}}{{\rm{G}}_{\rm{i}}} + {\rm \beta _{\rm{3}}}{{\rm{Z}}_{\rm{i}}} + \alpha + {\varepsilon _{\rm{i}}},$$
where Yi is the justification of suicide bombing and sympathy to al-Qaeda of individual i, Ri is a vector of individuals’ ideology about political Islam and their religiosity, Gi is a vector of economic and political grievances, Z is a vector for demographic characteristics (e.g., gender, age, and sect), α is a vector of country fixed effects to control for the differential association within countries, and ϵ is an error term.
Using data from multiple countries simultaneously in a multiple regression analysis assumes a constant relationship between outcome and predictor variables across all countries, which may not be valid due to variations in ethnic, social, and geographic contexts. Recognizing this potential heterogeneity, we posit that the relationship between predictor and outcome variables differs across countries. To examine this assumption, we also conducted separate regression models for each country, allowing regression slope coefficients to vary between countries. We employed a series of binary logistic regression models to estimate the probability of two dependent variables: individuals’ justification of suicide bombings and support for al-Qaeda.
Prior to conducting the logistic regression analyses, we inspected the data for the potential problems of missing cases, multicollinearity, and heteroscedasticity. To handle missing entries in the dataset, listwise deletion was initially employed for both the dependent and independent variables. This approach resulted in a reduced sample size of 3000, which is an adequate sample size for the logistic regression analysis (Peduzzi et al. Reference Peduzzi, Concato, Kemper, Holford and Feinstein1996).
Furthermore, to assess the presence of multicollinearity problems, the predictor variables were examined. According to Menard (Reference Menard2010) and Mertler and Vanatta (Reference Mertler and Vannatta2016), tolerance values below 0.1 indicate high multicollinearity. To calculate the tolerance values for the predictor variables in this study, separate linear regressions were performed for each dependent variable. The obtained tolerance values indicated the absence of multicollinearity issues, as all values exceeded the predetermined threshold.
Lastly, heteroscedasticity, which refers to unequal variances across the predictor variables, was assessed in the logistic regression models. The assumption of homoscedasticity was met, as indicated by the lack of significant heteroscedasticity problems in the analysis. These preliminary data checks ensured the suitability of the dataset for logistic regression analysis.
FINDINGS
This section presents the descriptive statistics of the study variables and provides an overview of the findings obtained from the binary logistic regression analyses. The primary objective of these analyses was to estimate the relationship between the measures of radicalization and two outcome variables: individuals’ justification of suicide bombings and sympathy with al-Qaeda.
Descriptive Statistics
Table 1 presents the descriptive statistics, including the mean and standard deviation, as well as the minimum and maximum values of the variables. The mean scores of the dichotomous dependent variables reflect the proportion of cases falling into each category. For instance, the mean probability of justification of suicide bombing is 0.20, indicating that approximately 20% of respondents in the sample justify suicide bombings. Similarly, the proportion of respondents who express sympathy with al-Qaeda is 16%, which is comparatively lower than the proportion justifying suicide bombings. Among the respondents, 53% are male, the average age is approximately 37 years, and 75% identify as Sunni.
Table 2 presents the percentages of respondents justifying suicide bombings and expressing sympathy for al-Qaeda for each country included in the model. In Egypt, approximately 31% of respondents justify suicide bombings, while about 20% hold favorable opinions of al-Qaeda (N=699). In Indonesia, around 30% of respondents express sympathy for al-Qaeda, but only about 9% justify suicide bombings (N = 562). For Jordan, approximately 14% of respondents justify suicide bombings, and about 18% hold favorable views of al-Qaeda (N=474). In contrast to Indonesia, around 36% of respondents in Lebanon justify suicide bombings, but only about 2% express sympathy for al-Qaeda (N=474). Turkey exhibits relatively lower levels of support for both suicide bombings (8%) and al-Qaeda (5%; N=536). Given the substantial variations in the distribution of dependent variables across countries, it is crucial to conduct separate analyses on the country-specific subsamples. This approach enables a thorough examination of the theoretical framework of radicalization within each country.
Table 2. Countries and the Percentages of Predicted Variables (N=3000)
Multivariate Analysis
We used logistic regression models for quantitative analysis to examine the probabilities of both justification of suicide bombings and sympathy with al-Qaeda. Initially, the theoretical models are examined for the entire sample from five countries. Three models were utilized to predict the two dependent variables and assess the influence of religion and grievances on the probabilities of justifying suicide bombings and expressing sympathy with al-Qaeda. The first model solely examines the relationship between the ideology of political Islam and religiosity on the dependent variables. In the second model, grievance variables and control variables were included and regressed against the dependent variables. Finally, the third model was a complete model, incorporating all independent variables as well as country fixed effects. Subsequently, separate regressions are conducted for each of the five countries. For each country subset, a single model encompassing all control variables is employed.
Justification for Suicide Bombings
Table 3 presents the first set of regression models, which aim to estimate the likelihood of justifying suicide bombings. In Model 1, the outcome variable is predicted using solely religion-based variables, namely the ideology of political Islam and religiosity. The goodness-of-fit statistics indicate that the model fits the data well (model χ 2 = 57.98, p < 0.001). This model accounts for approximately 2% of the variation in individuals’ attitudes towards suicide bombings (pseudo-R² = 0.02).
Table 3. Logistic Regression of Justification of Suicide Bombings among Muslims, 2011
Note: standard error values are in parentheses.
* p < 0.05, ** p < 0.01, *** p < 0.001.
The results of Model 1 (Table 3) reveal that both religion-based variables have statistically significant effects on the outcome variable. However, as expected, these two variables demonstrate opposing relationships. The presence of a political Islam ideology significantly and positively influences the likelihood of justifying suicide bombings. Specifically, when controlling for religiosity, each increase in the level of the political Islam ideology corresponds to a 70% increase in the probability of justifying suicide bombings ([1.70 – 1] × 100 = 70%). Conversely, for each incremental increase in religiosity, the probability of justifying suicide bombings decreases by 14% ([0.86 – 1] × 100 = –14%).
In Model 2 (Table 3), the analysis includes the predictor variables of economic and political grievances, as well as control variables, to examine their relationship with the probability of justifying suicide bombings. The overall fit of the model to the data is good (model χ² = 49.12, p < 0.001). The R² value of 0.05 indicates that this model explains approximately 5% of the variance in the probability of justifying suicide bombings.
When considering the simultaneous effects of all predictor variables, the direction, and significance of the religion-based variables, political Islam and religiosity, remain consistent. However, in contrast to the ideology of political Islam, individuals who are actively engaged and committed to Islam are less likely to justify suicide bombings. Additionally, political grievances are significantly and positively associated with the outcome variable. Specifically, with each increase in the level of political grievances, the probability of justifying suicide bombings increases by 87% ([1.87 – 1] × 100 = 87%). Furthermore, being Sunni and younger are factors associated with a lower likelihood of justifying suicide bombings. Sunnis, in particular, are 35% less likely to justify suicide bombings compared to individuals affiliated with other sectarian groups.
As the data is clustered by countries, failing to account for the grouped nature of individual observations may result in biased estimations (Beck and Katz Reference Beck and Katz1995). The third model (Table 3) includes country fixed effects to address this issue. Clark and Linzer (Reference Clark and Linzer2015:399) assert that “under certain conditions, random effects models can introduce bias but reduce the variance of estimated coefficients of interest. Fixed effect estimates will be unbiased but may be subject to high variance.” The Hausman specification test is used to compare the performance of fixed and random effects. According to the Hausman test results (Prob>Chi2 = 0.2455), the coefficient difference is not systematic, and the assumption of independence of irrelevant alternatives (IIA) is not violated in the random-effects model. The overall fit of the model to the data is good (model χ² = 357.78, p < 0.001). The R² value of 0.12 indicates that this model explains approximately 12% of the variance in the predicted variable. The inclusion of country fixed effects renders the effects of religiosity and being Sunni insignificant. However, the effects of political Islam, political grievances, and age remain similar to the previous models.
The model demonstrates a good fit to the data, as evidenced by the statistically significant p-value (Prob > likelihood-ratio = 0.00). The log-likelihood for a model with only the intercept is –1481, while for the full model, it is –1302. The Akaike information criterion (AIC) and Bayesian information criterion (BIC) indicate that the full model is superior to nested models. The Wald test results reject the null hypothesis (χ² = 111.88, p < 0.001), which tests whether all coefficients except the intercept are simultaneously equal to zero. Furthermore, testing the impact of religiosity alone using the Wald test and likelihood-ratio test yielded significant results (Wald test: χ² = 5.88, p < 0.015; likelihood-ratio test: χ² = 5.73, p < 0.017), rejecting the null hypothesis that the effect of religiosity is zero. These results support the use of the full model over nested models.
Table 4 presents a series of regression models for each country subset, allowing for a comparison of predictor variable effects across countries. The dependent variable, justification of suicide bombings, is regressed on all predictor and control variables. The goodness-of-fit statistics demonstrate that these models fit the data well (p < 0.001).
Table 4. Logistic Regression of Justification of Suicide Bombings among Muslims, 2011
Note: standard error values are in parentheses.
* p < 0.05, ** p < 0.01, *** p < 0.001.
The findings highlight the significant role of political grievances against Western politics and societies in individuals’ justification of terrorism. Political grievances exhibit a positive relationship with the likelihood of justifying suicide bombings across all country subsets. However, the analysis provides partial support for the effect of political Islam on the outcome variable. In Jordan, Lebanon, and Turkey, individuals who perceive Islam as a political system are more likely to justify suicide bombings. However, in Egypt and Indonesia, political Islam does not significantly influence the outcome variable. Notably, in Lebanon, religiosity reduces the likelihood of justification of suicide bombings, aligning with expectations.
Sympathy to al-Qaeda
Table 5 presents the results of the second set of regression models, which examine the likelihood of individuals’ sympathy for al-Qaeda. The models are constructed in a similar manner to those for the first dependent variable. In Model 1, only the religion-based variables, the ideology of political Islam and religiosity, are included. The model demonstrates a good fit to the data (model χ² = 62.65, p < 0.001) and accounts for approximately 2% of the variation in individuals’ sympathy for al-Qaeda (pseudo-R² = 0.02). The results indicate that only the ideology of political Islam has a significant effect on the outcome variable, while the effect of religiosity is not statistically significant. Specifically, controlling for religiosity, each incremental increase in the ideology of political Islam leads to a 76% increase in the probability of sympathy with al-Qaeda.
Table 5. Logistic Regression of Sympathy with al-Qaeda among Muslims, 2011
Note: standard error values are in parentheses.
*** p < 0.001.
Model 2 (Table 5) represents the complete model, including the predictor variables of economic and political grievances, as well as the control variables. The overall fit of the model to the data is good (model χ² = 114.94, p < 0.001), and it explains approximately 4% of the variance in the probability of support for al-Qaeda. The direction and significance of the effects of political Islam and religiosity remain consistent with the previous model. Notably, the effect of economic grievances on the outcome variable is significant and negative. Individuals experiencing higher levels of economic grievances and poverty are less likely to sympathize with al-Qaeda, with each incremental increase in economic grievances associated with a 28% decrease in the likelihood of support for al-Qaeda. Additionally, Sunnis are 36% less likely to support al-Qaeda. However, the effects of political grievances, gender, and age are not statistically significant in this model.
The third model (Table 5) incorporates country fixed effects. The results of the Hausman test (Prob>Chi2 = 0.00) suggest that the coefficient difference is systematic, indicating that the random-effects model violates the IIA assumption. Therefore, the fixed-effects model is deemed more robust than the models with random effects. The overall fit of the model to the data is good (model χ² = 334.92, p < 0.001), and it explains approximately 13% of the variance in the outcome variable. The inclusion of country fixed effects does not significantly alter the relationships compared to the previous models. The effect of religiosity remains insignificant, but the direction of the relationship has changed from positive to negative.
The model exhibits a good fit to the data, as indicated by the statistically significant p-value (p < 0.001). Table 5 presents the log-likelihood for a model with only the intercept (–1315) and the full model (–1258). The AIC and BIC values support the full model as the best fit among the nested models. The Wald test results reject the null hypothesis, which assumes that all coefficients, except the intercept, are simultaneously equal to zero (model χ² = 104.13, p < 0.001). Additionally, we conducted tests on the impact of religiosity using the Wald test (χ² = 1.57, p = 0.21) and likelihood ratio test (χ² = 1.61, p < 0.20). Both tests failed to reject the null hypothesis, indicating that the effect of religiosity is not significantly different from zero.
Table 6 presents the regression models for sympathy with al-Qaeda separately for each country. These models include all predictor and control variables. (The goodness-of-fit statistics suggest that the models for Indonesia and Turkey require a better fit to the data.) Overall, the relationships between the outcome variable and predictor variables vary substantially across countries. The ideology of political Islam only shows significant relationships in Egypt and Turkey. Moreover, religiosity is negatively associated with sympathy for al-Qaeda in Egypt. Economic grievances are negatively correlated with the outcome variable in Egypt, Jordan, and Turkey. Surprisingly, political grievances are negatively related to the dependent variable in Jordan and Lebanon, indicating that individuals with stronger negative attitudes towards Western policies and societies are less likely to support al-Qaeda. However, individuals with higher political grievances in Turkey are more likely to express sympathy towards al-Qaeda. None of the predictor variables shows a significant relationship with the outcome variable in Indonesia. In summary, the findings in Table 6 demonstrate considerable variation in the relationship between sympathy for al-Qaeda and predictor variables across countries.
Table 6. Logistic Regression of Sympathy with al-Qaeda among Muslims, 2011
Note: standard error values are in parentheses. “Sunni” is excluded from the model of Turkey, as it fully predicts the outcome.
* p < 0.05, ** p < 0.01, *** p < 0.001.
DISCUSSION
This study aims to understand the underlying causes of public support for suicide bombings and al-Qaeda. It seeks to unravel the complex relationship between religion and attitudes toward terrorist violence by focusing on individual sympathy for al-Qaeda and justification for suicide bombings. The study differentiates between religiosity and religion as an ideology to gain a nuanced understanding of the role of religion. Additionally, it analyses five countries with majority Muslim populations – Egypt, Indonesia, Jordan, Lebanon, and Turkey – and examines the influence of political and economic grievances on public attitudes towards terrorism.
The first major finding highlights the significant differences between the concepts of justification for suicide bombings and sympathy for al-Qaeda, emphasizing the need to analyse them separately. Justification for suicide bombings pertains to individual support for a strategy or tactic of terrorism and represents attitudes towards terrorism in general. In contrast, sympathy for al-Qaeda refers specifically to support for the terrorist organization. Therefore, it is crucial to exercise caution when conceptualizing and measuring support for terrorism.
The second major finding reveals that religiosity and political Islam have distinct associations with the justification of suicide bombings, often exhibiting opposite directions of influence. These empirical findings challenge the measurement of “Radical Islam” proposed by McCauley (Reference McCauley2012), emphasizing the need to differentiate between political Islam as an ideology based on a political interpretation of Islam and religiosity as an individual commitment to religion and spirituality. It is essential to distinguish political Islam and Islam as a faith. Political Islam instrumentalizes religion to attain political goals. And it is usually less selective of means toward those political goals.
The third major finding highlights significant variations in the dynamics of justification for suicide bombings and support for al-Qaeda across countries. Political grievances and the ideology of political Islam hold substantial explanatory power regarding individuals’ justification of terrorism. In all countries, political grievance significantly relates to the outcome variable. In Jordan, Lebanon, and Turkey, individuals who believe in the ideology of political Islam are more likely to justify suicide bombings. Religiosity has minor explanatory power, reducing the likelihood of justification of suicide bombings only in Lebanon. Regarding sympathy for al-Qaeda, the analyses provide limited support for the research hypothesis. The study indicates that the ideology of political Islam significantly impacts individuals’ feelings of sympathy for al-Qaeda. However, there is no convincing evidence supporting the effects of religiosity, poverty, and political grievances on the outcome variable. Estimations substantially vary when regressions are run separately for each country. The data demonstrate limited explanatory power of the models on individuals’ support for al-Qaeda, except for Egypt, where the model exhibits substantial explanatory power. The findings suggest that individuals with a political Islam ideology are more likely to feel sympathy for al-Qaeda, while religious individuals are less likely to support al-Qaeda. Therefore, support dynamics for al-Qaeda differ across countries and necessitate in-depth assessments to identify underlying factors.
Lastly, the study indicates that political grievances significantly explain individuals’ support for suicide bombings, particularly among those who harbor grievances against Western societies and policies. These findings align with the research hypothesis and previous literature (Tessler and Robbins, Reference Tessler and Robbins2007), which suggests that terrorism is more influenced by political strains than economic strains (Agnew Reference Agnew2010; Tessler and Robbins Reference Tessler and Robbins2007). However, the survey findings do not provide statistical evidence supporting the influence of economic grievances on the justification of suicide bombings.
The study acknowledges certain caveats, including the potential issue of endogeneity that should be addressed in future research. Specifically, a possible “reverse causation” problem may exist between the dependent variable of support for al-Qaeda and the independent variables of political grievance and ideology of political Islam. It is plausible that individuals supporting al-Qaeda, as a result of exposure to extremist propaganda, possess higher levels of negative attitudes toward Western societies, which may also influence their ideology of political Islam.
Researchers and practitioners should examine organizational and individual factors to comprehensively understand the recruitment process, including leadership characteristics, propaganda and recruitment activities, the social, economic, and political characteristics of potential sympathizers, and the radicalization process. These factors play a crucial role in understanding how organizations attract new members who may be driven by a “quest for significance” (Ozeren, Cubukcu, and Cash Reference Ozeren, Cubukcu and Cash2021).
Specifically, it is imperative to make a clear distinction between religiosity and the ideology of political Islam, particularly in studies pertaining to terrorism. There is a pressing need for thorough and comprehensive research that delves into the viewpoints of individuals who perceive religion as a political ideology, taking into account the specific contextual factors within each country. Overall, religiosity and mainstream Islam should not be perceived as a threat; instead, fostering a better understanding and practicing of Islam can serve as a remedy to prevent violent extremism.
Suat Cubukcu, PhD, is an assistant professor in the Department of Sociology, Anthropology, and Criminal Justice, and also holds an affiliated faculty position in the Integrated Homeland Security Program at Towson University. Additionally, he serves as a senior fellow at the Orion Policy Institute in Washington, DC. His research primarily focuses on terrorism, extremism, and policing. His work has appeared in Justice Quarterly, Studies in Conflict and Terrorism, Homicide Studies, The Journal of Policing, Intelligence and Counterterrorism, and Current Trends in Islamist Ideology (Hudson Institute), and been published by IOS Press and the National Bureau of Economic Research.
Huseyin Cinoglu, PhD, is an associate professor of Criminal Justice in the Department of Social Sciences at Texas A&M International University. Dr Cinoglu has published and presented extensively in the areas of (countering) violent extremism, (countering) terrorism, (de)radicalization, immigration, crime and criminality, identity formation, and terrorist identity formation. He has authored or co-authored several books. His articles have appeared in respected journals, such as International Journal on Criminology, European Scientific Journal, International Journal of Human Sciences, and in publications of IOS Press.
Suleyman Ozeren, PhD, is a professorial lecturer at the School of Public Affairs at American University. His areas of expertise are political violence, terrorism, cyber policy, the Kurdish issue, and Turkish foreign policy. He has worked collaboratively with NATO, the US State Department, the US Congress, and the United Nations Global Counterterrorism Forum. His recent publications include a co-edited book titled From Territorial Defeat to Global ISIS: Lessons Learned (2021), which NATO published. His opinion pieces and commentaries have appeared in several media outlets, including The National Interest, Orion Policy Institute, Cambridge Middle East and North Africa Forum (MENAF), Jerusalem Post, and Voice of America.
