Abstract
This research examines how social networks contribute to the process of radicalization, building on work showing that obsessive (vs. harmonious) passion for a cause is linked to greater support for political violence. Study 1 (N = 331) shows that obsessive (vs. harmonious) passion is related to affiliating with radical (vs. moderate) social networks, which in turn is associated with support for political violence. Study 2 (N = 381) provides experimental evidence for this phenomenon, by showing that inducing an obsessive mindset produces a greater proclivity to connect with radical activists, which in turn is associated with greater support for political violence. Drawing from social network analysis, Study 3 (N = 366) shows that network density intensifies obsessively passionate individuals’ affiliation to radical networks. The results offer insight into the group processes behind radicalization across different cultural contexts and ideologies.
Social networks play an undeniable role in radicalization leading to violence (Sageman, 2004; Weinberg & Eubank, 1987). They “serve as conduits through which individuals get acquainted with, and embrace, the ideological narrative that the network espouses” (Kruglanski, Bélanger, & Gunaratna, 2019, p. 51). One robust finding in the violent extremism literature is that, across vastly different ideologies, about two thirds of people who join a radical group are friends with at least one other group member before joining (Della Porta, 1988; Sageman, 2008). Moreover, strong social ties between members of a radical group produce trust and organizational loyalty, making it easier for extremists to conduct their operations clandestinely (Krebs, 2002; Lin, 2001; Qin, Xu, Hu, Sageman, & Chen, 2005). Despite these insights, research has thus far been silent about individual differences that might spur affiliation with radical social networks. Another gap in the literature is whether certain types of network structures increase affiliation with such groups.
To examine these questions, we build on prior work linking passion for a cause to violent political activism (Bélanger, Schumpe, Nociti, et al., 2019). Consistent with the 3N model of radicalization (Bélanger, Moyano, et al., 2019; Kruglanski Bélanger, & Gunaratna, 2019; Webber & Kruglanski, 2017) and uncertainty-identity theory (Hogg, 2000, 2007, 2009), we propose a trajectory whereby individuals struggling to fulfill their need for personal significance (i.e., obsessively passionate individuals), seek solace in radical networks that provide self-certainty, which predisposes them to embrace ideological narratives that condone political violence. Integrating social network analysis to these principles, we postulate that individuals’ affiliation to radical groups is amplified when members of the network are densely interconnected, because these network structures are conducive to creating strong social bonds (Podolny & Baron, 1997).
The Dualistic Model of Passion
Passion for a cause is defined as “a strong inclination toward a self-defining cause that is loved and valued, and in which people invest a significant amount of time and energy” (St-Louis, Carbonneau, & Vallerand, 2016, p. 263). Vallerand et al. (2003) proposed two distinct types of passion: harmonious passion (HP) and obsessive passion (OP), which both share similar levels of valuation and commitment to the cause, but differ with respect to how the passionate activity is regulated with other life domains (Bélanger, Schumpe, & Nisa, 2019).
HP is associated with a strong commitment to engage in the passionate activity and it remains well balanced with other activities in one’s life (Vallerand, 2015). Moreover, the activity is not associated with contingencies of self-worth (Mageau, Carpentier, & Vallerand, 2011), allowing individuals to have a secure sense of self (Bélanger, Lafrenière, Vallerand, & Kruglanski, 2013a; Lafrenière, Bélanger, Sedikides, & Vallerand, 2011). Consequently, involvement with the activity remains under the person’s control and does not conflict with other life domains. Thus, people feel free (vs. pressured) to pursue their activity (Séguin-Lévesque, Laliberté, Pelletier, Vallerand, & Blanchard, 2003).
In contrast, OP consists of an uncontrollable urge to engage in the activity; a state of motivational imbalance where the activity dominates attentional resources (Bélanger, Lafrenière, Vallerand, & Kruglanski, 2013b; Bélanger, Schumpe, & Nisa, 2019). The activity occupies an overpowering place in the person’s identity and is associated with contingencies of self-worth, leading to rigid (vs. flexible) engagement (e.g., Mageau et al., 2011). When the activity is regulated in an obsessive way, individuals’ self-worth revolves exclusively around their activity, which produces self-uncertainty (Lafrenière et al., 2011), ego-defensiveness (Bélanger et al., 2013a; Donahue, Rip, & Vallerand, 2009; Rip, Vallerand, & Lafrenière, 2012), and a high need for closure (Bélanger, Kis, Schumpe, Nisa, & Al Kindi, 2020).
In the last two decades, research on the dualistic model of passion has found that OP and HP are both positively correlated with commitment to the passionate activity (see Curran, Hill, Appleton, Vallerand, & Standage, 2015, for meta-analysis). However, despite similar levels of commitment, OP and HP are associated with different outcomes. Research has shown that HP is positively associated with the experience of flow (Lafrenière, Jowett, Vallerand, Donahue, & Lorimer, 2008), positive emotions (Mageau & Vallerand, 2007), and general psychological well-being (Rousseau & Vallerand, 2008). In contrast, OP is negatively related to these features (Vallerand, 2015; Wang, Khoo, Liu, & Divaharan, 2008). Furthermore, experimental research reveals that OP and HP are mindsets that can be situationally induced using a manipulation asking people to recall a time when their activity was in conflict (vs. well integrated) with other life domains (see Bélanger et al., 2013b; Schellenberg, Bailis, & Mosewich, 2016). There is also evidence that need frustration and need satisfaction (autonomy, competence, relatedness) increases OP and HP, respectively (Bélanger, Nisa, Schumpe, & Chamberland, 2019; Lalande et al., 2017). For example, individuals that feel chronically incompetent, disconnected from others, or lack complete autonomy outside of their passionate activity (e.g., at work, at school, etc.) may become OP (vs. HP) because they compensate by overinvesting themselves in the activity that satisfies their psychological needs (Ryan, Deci, Grolnick, & La Guardia, 2006; Vansteenkiste & Ryan, 2013).
Within the purview of political activism, Rip et al. (2012) found that Quebec nationalists who scored higher on HP preferred peaceful and democratic ways of furthering their ideology, whereas those scoring higher on OP showed a proclivity toward aggressive and radical methods. Extending these findings, experimental evidence suggests that the relationship between OP (vs. HP) and radical (vs. peaceful) activism is mediated by moral disengagement, that is, people’s ability to deactivate moral self-regulatory processes, allowing unethical behavior—a proclivity related exclusively to OP (Bélanger, Schumpe, Nociti, et al., 2019). Although these results shed light on the cognitive mechanisms linking OP to political violence, they do not address the social components associated with radicalization. This is an important limitation given that radicalization is not an atomistic process and people’s social networks play an important role in its development (see Kruglanski, Bélanger, & Gunaratna, 2019; Sageman, 2004, 2008).
Connecting Passion to Radical Networks
The question of what motivates individuals to identify with groups has been a long-standing issue in social psychology (Tajfel, 1972; Turner, 1975; Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). According to uncertainty-identity theory (UIT; Hogg, 2000, 2007, 2009, 2015), self-conceptual uncertainty plays an important role in group identification processes. UIT proposes that (a) feeling uncertain about oneself (one’s values, beliefs, place in society) motivates individuals to engage in behaviors geared toward reducing uncertainty; and (b) group identification is an effective means of reducing feelings of self-uncertainty “because it provides a clear sense of self that prescribes behavior and renders social interaction predictable” (Hogg, 2007, p. 88) and also “furnishes consensual validation and approval of who one is and what one should think and do” (Hogg, 2009, p. 223; see also Mullin & Hogg, 1999).
However, groups vary in the extent to which they provide a clear social identity to assuage self-uncertainty. One important feature to consider is a group’s entitativity, or the degree to which a group is perceived as a coherent social unit (Campbell, 1958). Indeed, highly entitative groups are particularly effective at reducing self-uncertainty because membership criteria are unambiguous, they have a clear internal structure, and group members share a common fate and goals (Castano, Yzerbyt, & Bourguignon, 2003; Hamilton & Sherman, 1996; Lickel et al., 2000). Moreover, members of entitative groups are perceived to be imbued with an underlying nature, an essence, that provides them with a certain set of immutable qualities (Demoulin, Leyens, & Yzerbyt, 2006). Consistent with this view, research by Hogg, Sherman, Dierselhuis, Maitner, and Moffitt (2007) has experimentally shown that individuals experiencing self-uncertainty identify more strongly with their group, especially if the group is highly entitative.
Integrating concepts from UIT with the dualistic model of passion is key to the present research because one of the fundamental characteristics of radical groups is that they uphold a rigid and absolutist worldview which leaves no room for dissent or nuance (Hogg, Fielding, & Darley, 2005; Hogg, Kruglanski, & Van den Bos, 2013; Kruglanski et al., 2014). Furthermore, radical groups tend to be highly entitative; they are characterized by a hierarchical structure under the central control of a leader and clearly delineated group boundaries (Milla, Hudiyana, Cahyono, & Muluk, 2020). In that sense, radical groups are a perfect vehicle to reduce self-uncertainty and should be sought after by those who crave certainty. Consistent with this proposition, research by Dugas et al. (2018) has found that the need for closure (i.e., the desire to avoid ambiguity) is associated with ingroup glorification and aggressive actions toward members of outgroups. Likewise, in a set of studies with samples of jihadists and Tamil Tigers, Webber et al. (2018) found that self-uncertainty (i.e., feeling personally ashamed and humiliated) predicts greater need for closure, which in turn predicts support for violent ideologies (Webber et al., 2018). Building on UIT, we thus predict that because of their uncertain sense of self and need for closure (e.g., Bélanger et al., 2020; Bélanger et al., 2013a; Lafrenière et al., 2011), OP individuals should be motivated to anchor themselves in radical groups, which in turn should be associated with their support for political violence. In contrast, individuals whose sense of self is more secure (i.e., HP) are more likely to connect with moderate individuals, and thus oppose political violence.
Amplifying Affiliation With Radical Groups: The Role of Network Density
In addition to these principles, the present research examines whether certain network structures increase affiliation with radical groups. In the social network analysis literature, the concept of network density—measured by the proportion of possible social ties in a network that are actually present—represents the extent to which members in a social network are connected to one another (i.e., network cohesion). Depending on the type of social tie, groups with high density tend to be highly entitative social units (Igarashi & Kashima, 2011; Sohn, 2014) that provide a sense of community and “we-ness”; a clear social identity with well-defined norms and expectations (Brown & Wade, 1987; Podolny & Baron, 1997). Furthermore, research indicates that densely interconnected networks facilitate communication, trust, and social support between individuals (Coleman, 1988; Feld & Carter, 1998), which can motivate individuals to adopt the attitudes and beliefs of those around them (Bott, 1977; Erickson, 1988; Krackhardt & Porter, 1985). For these reasons, we predict that when people are part of a radical group, the extent to which people report being friends with members of such groups increases with network cohesion. More formally, individuals’ affiliation to radical groups should be moderated by network density.
Radicalization Leading to Violence: A Trajectory
The present research postulates that individual passion and radical networks are dynamically interrelated in a process that increases people’s support for violent political actions. These concepts map well onto the 3N model of radicalization (Kruglanski et al., 2014; Kruglanski, Bélanger, & Gunaratna, 2019), which proposes three major determinants of radicalization that progresses into violent extremism. The first determinant is the need: individuals’ universal desire for personal significance, the sense that one’s life has meaning (Bélanger, Kruglanski, & Kessels, 2019; Dugas et al., 2016; Kruglanski, Gelfand, Bélanger, Hetiarachchi, & Gunaratna, 2015). This corresponds to the feeling of self-uncertainty that plagues individuals with an obsessive (but not harmonious) passion (e.g., Bélanger et al., 2013a; Lafrenière et al., 2011). The second determinant is the ideological narrative that elevates violence as a legitimate way of attaining significance—the reasoning that guides individuals in the quest for significance. The third determinant is the network, membership of one’s group who validate the narrative and who dispense rewards (respect and veneration) to members who follow its dictates. The 3N model of radicalization posits that individuals might be exposed to these ingredients in different temporal orders, thus creating different radicalization trajectories. However, echoing uncertainty-identity theory (Hogg, 2000, 2007), prior research has shown that the lack of personal significance is related to a “collectivistic shift” orienting individuals toward joining and identifying with a group (Kruglanski et al., 2013; Orehek, Sasota, Kruglanski, Dechesne, & Ridgeway, 2014). For this reason, we hypothesize that the radicalization trajectory should begin with OP (need), which should be related to affiliating with a radical group (network), which in turn should be associated with supporting political violence (narrative).
The Present Research
The purpose of this research was to investigate the social processes linking passion to support for political violence. We hypothesized that OP is related to being affiliated with radical social networks, which in turn is positively associated with supporting political violence. In contrast, we expected HP to be related to affiliating with moderate social networks, which in turn should be negatively associated with supporting political violence. We conducted three studies using a mix of methodologies to test our model. Study 1 provides cross-sectional evidence for the role of passion in support for political violence through affiliation with radical (vs. moderate) social networks. Study 2 conceptually replicates Study 1 and provides causal evidence for the mediated relationship. Study 3 replicates Study 1 and extends these results by examining the role of network density in moderating the relationship between OP and affiliation with radical social networks. Of note, each study is conducted in a different cultural context involving a wide range of political and religious ideologies.
Study 1
The aim of Study 1 was to provide initial evidence for the notion that OP and HP are associated with different social network affiliations. Extending the work of Bélanger, Schumpe, Nociti, et al. (2019), we hypothesized that OP is related to support for political violence and that this relationship is explained by affiliation with a radical social network. We expect that HP, on the other hand, is related to affiliation with a moderate network, which in turn is negatively associated with supporting political violence.
Method
Participants
Assuming medium effect sizes and power set at .80, a sample of 146 people was suggested by 5,000 Monte Carlo simulations. Three hundred and thirty-one Spaniards were recruited (250 women; Mage = 24.83 years, SDage = 7.79) using snowball sampling, and completed an online survey.
Procedure and materials
Participants completed a survey measuring their passion for a cause, their ideological commitment, affiliation to moderate and radical social networks, and their support for political violence.
Passion
Participants were first instructed to indicate a cause that is important to them. People listed causes such as the antiausterity movement (Movimiento 15-M; 0.9%), Association for the Recovery of Historical Memory (Asociación para la Recuperación de la Memoria Histórica; 1.2%), movement against bullfights (Movimiento Anti-taurino; 1.5%), Spanish Nation Defense Foundation (Fundación para la Defensa de la Nación Española; 2.7%), climate change (27.8%), women’s rights (57.4%), and other causes (8.5%). Then, participants completed the Passion Scale (Vallerand et al., 2003) to measure participants’ harmonious (e.g., “My involvement in this cause allows me to live a variety of experiences,” “My cause is in harmony with the other activities in my life,” “My cause reflects the qualities I like about myself,” “My cause is well integrated in my life,” “My cause is in harmony with the other things that are part of me,” and “The new things that I discover with my cause allow me to appreciate it even more”; M = 5.44, SD = 0.90; α = .78) and obsessive passion (e.g., “I have difficulties controlling my urge to think about my cause,” “I have almost an obsessive feeling for my cause,” “My cause is the only thing I can think of,” “If I could, I would devote myself entirely to my cause,” “I have the impression that my cause controls me,” “My involvement in this cause is so exciting that I sometimes lose control over it”; M = 2.48, SD = 0.93; α = .73) for the cause they listed. Participants provided their answers on a 7-point Likert scale (1 = not agree at all, 7 = very strongly agree).
Commitment
We included four items to measure the extent to which participants are committed to their cause (i.e., “My cause is important for me,” “I spend a lot of time thinking about my cause,” “I like my cause,” and “My cause is a passion for me”; M = 4.86, SD = 0.87; α = .67). Participants provided their answers on a 7-point Likert scale (1 = not agree at all, 7 = very strongly agree).
Moderate network
Affiliation with a moderate network of people (M = 6.19, SD = 0.81; α = .74) was measured using three items (e.g., “People around me are usually hard-working and competent,” “My friends are sincere and fair people: good people,” and “In my family and at school, people transmit positive values related to helping others”; 1 = not agree at all, 7 = strongly agree). A factor analysis with maximum likelihood and oblimin rotation was performed on these items and results indicated a single factor with an eigenvalue of 1.98 explaining 66% of the variance. All factor loadings were above .65.
Radical network
Affiliation with a radical network of people (M = 1.86, SD = 0.88; α = .67) was measured using five items: “People around me consider it appropriate to use violence for an ideology,” “People I know talk about fighting and violence all the time,” “I have several friends or family who have problems with justice or have had them at some point in their lives,” “The people I know get into a lot of trouble (theft, drug use, fighting),” “I personally know people who support violence for political change” (1 = not agree at all, 7 = strongly agree). A factor analysis was also performed on these items and results indicated a single factor with an eigenvalue of 2.31 explaining 33% of the variance. All factor loadings were above .44.
Support for political violence
Participants’ support for political violence (M = 2.06, SD = 1.32; α = .71) was measured using three items taken from Bélanger, Moyano, et al.’s scale (2019): “Violence is necessary for social change,” “I would never consider physical violence to further a just cause” (reverse-scored), and “We should never use violence as a way of trying to change society” (reverse-scored; 1 = not agree at all, 7 = strongly agree). A factor analysis conducted on these items indicated a one-factor solution with an eigenvalue of 1.91 explaining 63% of the variance, with factor loadings above .49.
Results and discussion
The predicted model (see Figure 1) was tested with AMOS (Arbuckle, 2007) using maximum likelihood estimation. We display means, standard deviations, and correlations for all measures in Table 1.
Results from path analyses (Study 1).
Means, standard deviations, and correlations involving variables from Study 1 (N = 331).
Note. *p < .05. **p < .01. ***p < .001.
Results indicated that the hypothesized model fit the data well, χ2 (df = 2, N = 331) = 5.78, p = .05, GFI = 0.99, CFI = 0.98, IFI = 0.98, RMSEA = 0.07, AIC = 43.78. Obsessive passion was positively associated with radical network (B = 0.15, SE = 0.05, t = 2.65, p = .008) and negatively related to moderate network (B = −0.19, SE = 0.05, t = −3.76, p < .001), whereas harmonious passion was negatively related to radical network (B = −0.18, SE = 0.06, t = −2.85, p = .004) and positively related to moderate network (B = 0.24, SE = 0.05, t = 4.12, p < .001). Moderate network was negatively related to support for political violence (B = −0.27, SE = 0.08, t = 3.15, p = .02), whereas radical network was positively related to it (B = 0.25, SE = 0.08, t = −3.09, p = .002). Commitment was not related to any of the variables (all ps > .15).
Bootstrapped confidence interval estimates of the indirect effects (Preacher & Hayes, 2008) were calculated to test the significance of the mediation. We found an indirect effect of (a) OP on participants’ support for political violence through radical network (B = 0.09, SE = 0.03, 95% CI [0.04, 0.16]) and of (b) HP on participants’ support for political violence through moderate network (B = −0.11, SE = 0.03, 95% CI [−0.18, −0.05]).
Thus, Study 1 extends the work of Bélanger, Schumpe, Nociti, et al. (2019) by showing that OP is positively associated with support for political violence because it is positively associated with affiliation to a radical social network; whereas HP is negatively associated with support for political violence because it is positively associated with affiliation to a moderate social network. One limitation of Study 1 is that the data are cross-sectional, and thus do not permit causal inference. We address this limitation in Study 2.
Study 2
The purpose of Study 2 was to extend Study 1 by providing experimental evidence that OP (vs. HP) produces a desire to connect with radical activists. Given that effective recruitment and radicalization leading to violence are increasingly taking place online (Koehler, 2014), we created a decision-making paradigm whereby participants evaluated different activist profiles on Facebook and indicated to what extent they would like to befriend them. We hypothesized that OP (vs. HP) would increase people’s preference for radical (vs. moderate) profiles, which in turn would be associated with greater support for political violence.
Method
Participants
Assuming small-to-medium effect sizes and power set at .80, a sample of 271 people was suggested by 5,000 Monte Carlo simulations. Three hundred and eighty-one Americans were recruited (168 women; Mage = 35.67 years, SDage = 11.41) via Mechanical Turk to complete an online survey. Participants were invited to take part in this study based on their self-identification as environmentalists in an independent prescreening survey.
Procedure and materials
Participants were randomly assigned to one of the two experimental conditions to induce different passion mindsets (see Bélanger, Schumpe, & Nisa, 2019; Schellenberg et al., 2016). In the HP condition (N = 201), participants were instructed to,
Write about a time when the environmental cause was in harmony with other things that are part of you, and you felt that the environmental cause allowed you to live a variety of experiences. Recall this event vividly and include as many details as you can to relive the experience.
Participants in the OP condition (N = 180) were assigned to a similar writing task but were instructed to,
Write about a time when you had difficulties controlling your urge to engage in the environmental cause, and you felt that the environmental cause was the only thing that really turned you on. Recall this event vividly and include as many details as you can to relive the experience.
Then, participants were told that they would be shown anonymized Facebook profiles of environmentalists and, for each profile, the person’s most salient characteristic would be presented. Participants were instructed to indicate the extent to which they would like to befriend (or not) that person on a scale ranging from −5 (would not make friends with that person) to +5 (would make friends with that person). Participants were exposed to three radical profiles (e.g., “Engages in behavior to oppose ruling powers,” “Rebels against socially fitting in,” and “Rebels against authority”; M = −0.05, SD = 2.17; α = .78) and four peaceful profiles (e.g., “Thinks the best way to save planet Earth is through peaceful protests,” “Organizes fundraising activities for an environmental organization,” “Sensitizes close relatives and colleagues about eco-friendly behaviors,” and “Organizes cultural activities to increase awareness about the environment”; M = 2.78, SD = 1.44; α = .77). After rating the profiles, participants completed the Support for Political Violence Scale as in Study 1 (M = 2.45, SD = 0.99; α = .74). A factor analysis with maximum likelihood and oblimin rotation was conducted on the three items related to the rebel profiles; results indicated a one-factor solution with an eigenvalue of 2.08 explaining 69% of the variance, with factor loadings above .66. A similar factor analysis was conducted on the four items related to the peaceful profiles; results indicated a one-factor solution with an eigenvalue of 2.48 explaining 62% of the variance and factor loadings above .54.
Results and Discussion
The predicted model (see Figure 2) was tested with AMOS (Arbuckle, 2007) using maximum likelihood estimation. We display means, standard deviations, and correlations for all measures in Table 2.
Results from path analyses (Study 2).
Means, standard deviations, and correlations involving variables in Study 2 (N = 381).
Note. *p < .05. **p < .01. ***p < .001.
Results indicated that the hypothesized model fit the data well, χ2 (df = 1, N = 381) = 0.15, p = .90, GFI = 1.00, CFI = 1.00, IFI = 1.00, RMSEA = 0.00, AIC = 10.01. The experimental condition (coded 0 = HP, 1 = OP) was positively associated with intention to befriend radical activists (B = 0.44, SE = 0.22, t = 1.98, p = .04), which in turn was positively associated with supporting political violence (B = 0.16, SE = 0.25, t = 6.72, p < .001). There was no effect on intention to befriend peaceful activists (B = 0.14, SE = 0.14, t = 0.96, p = .33).
Bootstrapped confidence interval estimates of the indirect effect (Preacher & Hayes, 2008) were calculated to test the significance of the mediation. Results indicated that the effect of the experimental condition on support for political violence was mediated by participants’ willingness to befriend radical activists (B = 0.07; 95% CI [0.00, 0.16]).
Thus, Study 2 experimentally demonstrates that inducing OP (vs. HP) produces a greater willingness to befriend radical (vs. moderate) activists, which in turn is positively associated with support for political violence. In the next study, we aim to demonstrate that the relationship between OP and affiliation to a radical social network can be amplified when people’s social network is densely interconnected.
Study 3
The purpose of Study 3 was to conceptually replicate and extend the previous studies by examining the moderating role of network density. Prior work suggests that network density increases the experience of “we-ness” and community among group members (Brown & Wade, 1987; Podolny & Baron, 1997). As in previous studies, we hypothesized that OP would be related to affiliating with a radical social network, which in turn would be positively associated with support for political violence. However, we expected that our mediation model would be moderated by network density such that, the denser the network, the more people should report being friends with radical individuals.
Method
Participants
Assuming medium effect sizes and power set at .80, a sample of 215 people was suggested by 5,000 Monte Carlo simulations. Three hundred and sixty-six Muslim students in Pakistan were recruited (137 women; Mage = 21.05 years, SDage = 1.59) to complete a survey in the classroom.
Procedure and materials
Akin to Study 1, participants first completed the passion and commitment scales. Then we measured people’s egocentric networks (Perry, Pescosolido, & Borgatti, 2018), which represent the connections (links, ties) between the people (alters) that a participant (ego) knows. Specifically, we used a name generator approach to elicit a list of up to 10 alters with whom egos were familiar and have talked to most within the past month (Laumann, 1966). Then, using an adjacency matrix, we asked egos to indicate the connections between alters (e.g., does Alter A know Alter B, does Alter A know Alter C, and so on). This generated an alter–alter network structure for each ego.
Passion
As in Study 1, participants completed the Passion Scale (adapted to religion) to measure their harmonious (e.g., “Islam is in harmony with the other activities in my life”; M = 4.88, SD = 1.06; α = .79) and obsessive (e.g., “Islam is the only thing that I can think of”; M = 4.43, SD = 1.04; α = .70) passion.
Commitment
As in Study 1, we measured participants’ ideological commitment with four items (M = 5.44, SD = 1.09; α = .72).
Network density
We use density (Wasserman & Faust, 1994) to measure the degree of connectedness of alters in an ego’s alter–alter network (M = 0.41, SD = 0.30). The density of an ego network is the number of ties among the alters divided by the number of possible ties, such that Network density =
Radical network
Participants’ affiliation with a radical social network (M = 2.44, SD = 1.24; α = .74) was measured with the following four items: “My friends usually speak about fighting and violence all the time,” “My friends get into too much trouble (theft, drug use, fights, etc.),” “Sometimes I have been criticized for not fighting to defend my country,” and “People around me say it is appropriate to use violence for an ideology.” Participants rated these items on a 7-point scale (1 = not agree at all, 7 = very strongly agree). A factor analysis with maximum likelihood and oblimin rotation was performed on these items and results indicated a single factor with an eigenvalue of 2.28 explaining 57% of the variance and factor loadings above .61.
Support for political violence
Participants’ support for political violence was measured with the following two items: “Violence is necessary for social change” and “When using violence to further a just cause, anybody can be a target.” Participants rated these items on a 7-point scale (1 = not agree at all, 7 = very strongly agree). Items were correlated (rs = .55) and were thus averaged (M = 2.41, SD = 1.37).
Results and Discussion
We tested the proposed moderated mediation model using Hayes’s (2017) PROCESS macro (Model 7) to determine whether the indirect effect between OP and support for political violence through affiliation with a radical social network is moderated by network density, while controlling for HP and ideological commitment. Following Aiken and West’s (1991) procedures, we standardized the independent variables. We display means, standard deviations, and correlations for all measures in Table 3.
Means, standard deviations, and correlations involving variables in Study 3 (N = 366).
Note. *p < .05. **p < .01. ***p < .001.
Results indicated that the path from OP to radical social network was significant (β = .27, p = .002), so was HP (β = −.26, p = .001), but not commitment (β = .06, p = .46). The path from radical network to support for violence was also significant (β = .49, p < .001). Moreover, the path from OP to radical network was moderated by network density (β = .12, p = .04). To further probe the nature of this effect, we computed the conditional effect of the interaction for low and high levels of network density (Hayes, 2017). The effect of OP on radical network was significant for high (1 SD above the mean; β = .40, p < .001), but not low (1 SD below the mean; β = .14, p = .18) levels of network density (see Figure 3).
Affiliation with radical network as a function of obsessive passion and social network density (Study 3).
Furthermore, bootstrapped confidence interval estimates of the indirect effect (Preacher & Hayes, 2008) were calculated to test the significance of the mediation. Results indicated that the effect of OP on support for political violence through affiliation with radical network was significant for high levels (β = .16, 95% CI [0.06, 0.27]), but not low levels (β = .05, 95% CI [−0.02, 0.14]), of network density (see Figure 4). The model explained 16% of the variance in support for political violence.
Indirect effect of harmonious and obsessive passion on support for political violence through affiliation with radical network for high (vs. low) levels of social network density (Study 3).
Taken together, Study 3 provides evidence that the indirect effect of obsessive passion on support for political violence through affiliation to radical social network is moderated by the density of the network to which people belong. Results indicated that OP predicts affiliation to a radical network and that such affiliation increases when individuals belong to a highly interconnected network. The greater the affiliation with the radical group, the greater the support for political violence.
General Discussion
Building on prior work linking OP (vs. HP) to political violence (Bélanger, Schumpe, Nociti, et al., 2019), this research focused on the social mechanisms related to violent extremism. In Study 1, OP was positively related to support for political violence, while HP was negatively related to it. We also found that both effects were mediated by people’s affiliation with radical and moderate networks, respectively. In Study 2, experimental evidence showed that individuals with an OP (vs. HP) for the environment preferred radical (vs. moderate) online profiles of fellow activists, which in turn predicted their support for political violence. Last, in Study 3, we replicated Study 1 and showed that OP (but not HP) for religion is related to affiliating with a radical social network, which in turn is positively associated with support for political violence. Extending these notions, we also found that network density increased the magnitude of the relationship between OP and affiliation with radical networks. Taken together, the present research provides strong empirical support for our hypotheses through a set of findings that were consistent across multiple studies involving a wide range of ideologies and cultural contexts (Spain, USA, Pakistan).
The first contribution of this work is to demonstrate that the relationship between passion for a cause and support for political violence is mediated by people’s affiliation with different social networks. This is the first piece of evidence showing that passion influences the type of people with whom we interact—a contribution meaningful to both the dualistic model of passion and the study of radicalization. Indeed, we show that people who pursue their ideologies obsessively increase their risk of adhering to radical ideologies because they gravitate toward radical individuals. We find that HP, on the other hand, is protective—individuals who pursue their ideologies harmoniously are more likely to connect with moderate peers. These results are consistent with UIT (Hogg, 2000, 2007) and the 3N model of radicalization (Bélanger, Moyano, et al., 2019; Kruglanski, Bélanger, & Gunaratna, 2019), which predict that self-uncertainty (i.e., lack of personal significance) produces a desire to identify with groups, especially those that provide a clear social identity, such as radical groups. Understanding these dynamics is crucial to developing preventive measures that redirect individuals toward more peaceful social networks and away from radical social networks that may accelerate their journey toward violent behaviors.
The second contribution consists in offering evidence that network structure (i.e., density) amplifies the relationship between OP and affiliation with radical social networks. Our discoveries corroborate previous findings that richly interconnected networks provide strong bonding social capital and a sense of community (e.g., Putnam, 2001), and extend these principles to the context of violent extremism. This is, to our knowledge, the first empirical evidence showing that deeply interconnected groups with radical ideologies are more prone to supporting violence. Future research should further examine how dense networks evolve over time using longitudinal designs. According to Sunstein’s (2002, 2009) research on group polarization, one would expect like-minded individuals socializing in tightly knit groups to progressively hold more polarized beliefs, values, and ideologies. Yet, the literature is replete with case studies of extremists who have successfully exited radical groups (Kruglanski, Bélanger, & Gunaratna, 2019; Kruglanski, Webber, & Koehler, 2019). Investigating the conditions in which people leave radical groups and transition to connecting with more moderate ones would be a worthy endeavor for future research. Furthermore, given that the network density of online radical groups can be measured via social media, the present findings could be useful to law enforcement agencies by allowing them to prioritize the surveillance of dense groups over loosely interconnected ones.
Conclusion
Understanding the social psychological principles associated with radicalization leading to violence is imperative to improve public safety. The present research sheds light on these mechanisms by examining the roles that passion, affiliation with radical social networks, and network density play in supporting political violence. Our results extend prior work by showing that obsessive passion creates a proclivity to affiliate with radical individuals, which in turn facilitates adhering to ideologies in support of political violence. We also demonstrate that the relationship between OP and radical networks is amplified by network density. Future research should investigate how to reverse these dynamics.
