ANGER FROM WITHIN: THE ROLE OF EMOTIONS IN DISENGAGEMENT FROM VIOLENT EXTREMISM

There is growing recognition about the similarities between generic criminality and violent extremism. Using data derived from a unique set of in-depth life history interviews with 40 former U.S. white supremacists, as well as previous studies of criminal desistance, we examine the emotional valence that characterizes actors' descriptions of the disengagement process. More specifically, results suggest that negative emotions (i.e., anger and frustration) directed toward the extremist group and oneself function as a catalyst for disengagement. Negative emotions become a source of motivation in re-evaluating the relative importance of the group as it relates to the individual. Ultimately, the reevaluation of the group is essential to the decision to disengage from violent extremism.


INTRODUCTION
Despite claims that violent extremism is fundamentally different from generic criminal offending (Hirschi & Gottfredson, 2001;Silke, 2014), other observers point to important similarities between the two phenomena (Clarke & Newman, 2006;Fahey & LaFree, 2015;Mullins, 2009;Rice, 2009).While various criminological frameworks have recently been utilized to study violent extremism vis-a`-vis subcultural theory (Pisoiu, 2015), rational choice (Perry & Hasisi, 2015), displacement and diffusion (Hsu & Apel, 2015), social disorganization (Fahey & LaFree, 2015), routine activities (Parkin & Freilich, 2015), and deterrence (Argomaniz & Vidai-Diez, 2015), few studies employ a life course criminological approach (see for exceptions Hamm, 2013;Simi, Sporer, & Bubolz, 2016).This is an unfortunate omission as violent extremism involves a range of issues life course criminology is well suited to examine such as onset and persistence as well as disengagement and desistance.In addition, a life course approach provides an opportunity to assess the commonalties and differences between violent extremism and generic criminality and potentially expand the scope of a framework that according to some scholars represents the central and most pervasive theory within criminology (Cullen, 2010).
Violent extremism is a growing field of study with a substantial focus on how individuals and groups become mobilized for ideological violence and, more recently, what types of intervention can be used to counteract such processes (Altier, Thoroughgood, & Horgan, 2014;Bjørgo, 2011;LaFree & Miller, 2008;Sageman, 2014).Although the process of disengagement from violent extremism has begun to receive more attention in recent years (Bjørgo & Horgan, 2009;Gadd, 2006;Horgan, 2009;Kruglanski et al., 2014), no study to date has examined disengagement from violent extremism by relying on life course criminology and studies of criminal desistance.Horgan (2009), for example, concludes the ideological dimension of violent extremism limits the conceptual value of criminal desistance to explain disengagement.Yet, the field lacks systematic empirical evidence to support this assessment.Further, existing studies of disengagement from violent extremism are typically focused on international samples of extremists, with less attention directed toward domestic extremists in the United States (see for exception Aho, 1994;Blazak, 2001;Bubolz & Simi, 2015a) and thus the idea that extremism and generic crime are completely distinct may be unwarranted (Simi, Sporer, & Bubolz, 2016).
As such, this article builds on previous studies of extremist disengagement by drawing from the criminal desistance literature to examine a sample of former U.S. domestic extremists (Giordano, Cernkovich, & Rudolph, 2002;Giordano, Schroeder, & Cernkovich, 2007;Paternoster & Bushway, 2009).The comparison of generic criminal desistance to disengagement from violent extremism is important because, unlike individual offending, criminality related to violent extremism, at times, involves attachment to a highly emotive collective or group identity with varying levels of ideological commitment.Caution should be taken, however, about assuming one point of distinction means violent extremism and generic criminals share nothing else in common or that disengagement and desistance are fundamentally different.
In fact, recent empirical studies point to considerable similarities between the two populations in terms of the presence of childhood risk factors (Simi, Sporer, & Bubolz, 2016); the disproportionate rates of criminal activity committed by young males (McCauley & Segal, 1987;Russell & Miller, 1983), participation in non-ideological, generic offending (Ezekiel, 1995;Hamm, 1993) and adherence to an ongoing organizational structure (Maguire & Pastore, 1996;Short, 1997).In addition, violent extremists and various serial offenders are not defined by a single act but rather by the amalgamation of multiple violent crimes over the course of that individual's criminal career (LaFree & Dugan, 2004).In the next section, we discuss several key concepts that guide our analysis.
To address these deficiencies, several studies focused on the cognitive and emotional components related to desistance (Maruna, Lebel, Mitchell, & Naples, 2006).Giordano and colleagues (2007) argued that life course transitions (e.g., marriage, joining the military) are embedded with emotion.More specifically, Giordano and colleagues (2007) proposed that anger and frustration increase the likelihood of experiencing setbacks when individuals attempt to leave a criminal lifestyle.As such, an important part of moving away from offending involves developing the ability to cope with these negative emotions.Emotional changes play an important role in the desistance process and provide needed details about how life transitions are connected to decreases in antisocial behavior.
Along these lines, Giordano and colleagues (2007) stressed the reduction of anger helps individuals restructure the importance of their priorities toward personal aspirations and interests unrelated to antisocial behavior.This restructuring sparks the change process and reduces the likelihood of future criminal offending.Giordano and colleagues' (2007) research, however, focuses on individual-level offending as opposed to disengagement from a group-level context.Leaving an extremist group, as opposed to desisting from generic criminality, may involve a different process because the person must shed his or her attachment to the ideology and group in order to disengage.
Relying on a rational choice model, Paternoster and Bushway (2009) developed a theory of desistance by incorporating the concept of a "feared self" (p.1119).Within this framework, fear and anxiety related to the individual's future possible self-motivate desistance from antisocial behaviors.While Paternoster and Bushway's (2009; also see Bushway & Paternoster, 2013) model offers significant insight regarding the criminal desistance process, their theory pivots on the notion of a "crystallization of discontent" but stops short of specifying how and under what circumstances discontent emerges.In addition, Paternoster and Bushway (2009) do not link discontent to emotional processes such as anger in terms of generating and clarifying a person's sense of disillusionment.
In light of this oversight, Bubolz and Simi (2015b) recently argued that in terms of disengagement from street gangs, the presence of anger provides members with important motivation to seek disaffiliation and begin the change process.The experience of anger helps members solidify their feelings of discontent with the incongruence between expectations and realities of gang life.We elaborate on Bubolz and Simi's (2015b) theory of gang exit by presenting an empirical analysis of disengagement from violent extremism focused on the role of emotion.Given the role emotions may play within the disengagement process, a relevant theory to explore disengagement is the circumplex model of affect.
Historically, research focused on emotion has viewed affective states as discrete, such as anxiety, sadness, tension and elation (Chipperfield, Perry, & Weiner, 2003;Izard, 1972;Tomkin, 1963).Recently, however, the idea that affective states are independent of one another has been challenged (Posner, Russell, & Peterson, 2005;Russell, 1980).An emerging consensus suggests that emotions are overlapping experiences that lack distinct boundaries (Erbas, Ceulemans, Koval, & Kuppens, 2015;Posner et al., 2005;Russell & Fehr, 1994;Saarni, 1999).Individuals do not experience or recognize emotions as isolated events, but rather report their emotions as overlapping experiences (Saarni, 1999).Individuals rarely report feeling a specific emotion without also reporting other emotions, so one possible approach to understanding the overlap of emotional affect is through the application of the circumplex model of affect (CMA).
The CMA has become one of the most widely used models of emotional affect (Huelsman, Furr, & Nemanick Jr., 2003;Remington, Rabrigar, & Visser, 2000) and was originated by Schlosberg (1952) and later elaborated by Russel (1980).In the CMA, emotions are arranged in a circular structure characterized by two poles: 1) intensity (i.e., activation or deactivation); and 2) valence (e.g., pleasant or unpleasant).The CMA assumes that affective states are related by their distance from one another.For instance, excitement, pleasure, and contentment should have positive correlations between one another, whereas pleasure should be negatively correlated with misery, distress, and depression.The latter emotions may be especially prominent during periods of high stress or disillusionment, such as disengaging from violent extremism.
We rely on a systematic content coding system derived from the CMA to identify overarching themes, specific events, and markers of emotion expressed by participants during intensive life history interviews.Specifically, we examine the emotional valence that characterizes actors' descriptions of the exit process.Our central argument is that negative emotions (i.e., anger, frustration) directed toward the extremist group and oneself function as a catalyst for disengagement.These negative emotions become a source of motivation in reevaluating the relative importance of the group as it relates to the individual.Ultimately, the reevaluation of the group is essential to the decision to disengage from extremism.Our study builds on previous studies of criminal desistance and extremist disengagement by utilizing a theoretical framework that offers greater precision for understanding the impact of cognitive and emotional processes as they relate to a person's decision to leave these types of groups.

PRIOR RESEARCH ON EXTREMIST DISENGAGEMENT
Disengagement is "the process whereby an individual no longer accepts as appropriate the socially defined rights and obligations that accompany a given role in society" (Ebaugh, 1988, p. 3).In this sense, disengagement is a type of role exit that typically occurs as individuals transition from one stage in life to another (Simi, Blee, DeMichele, & Windisch, 2017).For instance, disengaging from a street gang typically accompanies a shift in behavior as these individuals internalize a new role in society (Decker, Pyrooz, & Moule, Jr., 2013).For these individuals, disengagement from the gang lifestyle requires them to shed their "gangster" role and adopt the new role of "former gang member."As part of the disengagement process, these individuals may decrease their level of "embeddedness" within the gang by de-identifying as a gang member (e.g., altering their appearance or changing the way they respond to conflict) (Sweeten, Pyrooz, & Piquero, 2013).
Similar to criminal desistance, disengagement from violent groups occurs in two forms.First, the individual may alter his/her level of participation in the group such as avoiding violence or reducing the amount of time they spend with other members (Horgan & Braddock, 2010).In these situations, the individual remains a member but reduces his/her level of investment (e.g., time, energy and risk of injury/arrest) with the group.Second, disengagement may entail the individual leave the group entirely.This is the typical scenario envisioned and often involves a complete separation from extremist activities.In terms of the current study, we emphasize the latter form of disengagement in which an individual completely severs his/her involvement with violent extremism.
In terms of psychological factors, disengagement may be the result of burnout or disillusionment that stems from differences between expectations and reality (Aho, 1994;Bubolz & Simi, 2015b;della Porta, 2009;Horgan & Braddock, 2010;Kimmel, 2007).Disillusionment results from dissatisfaction with the activities of the group, a lack of loyalty among members of the group, and the way that younger members are manipulated by veterans (Bjørgo, 2011).Gadd's (2006) study of a British far-right extremist found that identification with different individuals (i.e., children, romantic partner, and community) led to a recognition of dissatisfaction with far-right extremism.Part of the dissatisfaction may also result from a moral uneasiness with movement ideology and participation (Bjørgo, 1997).
Although scholarship on disengagement has advanced in recent years, the varied explanations do little to describe the complex interactional process by which structural, emotional, and cognitive factors interact as part of an individual's decision to exit from his/her role as a member of a violent extremist group.Further, previous studies of disengagement or desistance have not typically incorporated a structured theoretical and methodological approach to assess the emotional markers embedded within individual narratives (for an exception see Latif, Blee, DeMichele, and Simi, 2018).The current study addresses this gap by utilizing CMA to specifically focus on the role of anger during the disengagement process.In the following section, we introduce the methodology followed by data and theoretical implications.

METHODOLOGY Sampling Procedures
Long-term ethnographic fieldwork with far-right extremists provided the basis for initial contacts with interview participants.The study also relied on contacting former extremists who have written books about their experiences, shared their stories on websites, or have spoken publicly about their extremist involvement.Each of the initial participants was asked to provide referrals to other former extremists who might also be willing to participate in an interview.This snowball sampling process produced contacts that otherwise would not be accessible using traditional means of sampling, such as mailing lists (Wright, Decker, Redfern, & Smith, 1992).Multiple individuals were used to generate unique snowballs, and thus participants were often not acquainted with each other.Substantial rapport was established and maintained through regular contact via telephone, email, and social media.

Participants
Our sampling method resulted in life history interviews with 40 former members of U.S. white supremacist groups.The current sample included 5 female and 35 male participants whose aged ranged from 24 years to 63 years of age.Regarding socioeconomic status, 6 participants described themselves as lower-class, 13 as working-class, 17 middle-class, and 4 described themselves as upper-class.The wide distribution of socioeconomic status is consistent with previous studies of white supremacists (Aho, 1990;Blee, 2002).In terms of education, 9 individuals earned less than a high school diploma, 8 earned a high school diploma, 11 attended college, and 12 earned some form of college degree.The level of group involvement for members included 4 individuals who founded a white supremacist group and 36 participants who were either core or peripheral members.Participants' length of involvement ranged from 1-22 years.
A large portion had extensive histories of criminal conduct including property offenses such as shoplifting, vandalism, and other forms of property destruction and a variety of violent offenses such as murder, attempted murder, street fights, violent initiation rituals, and bomb making.Of the 40 participants, 27 reported a history of violent offending, 34 reported a history of delinquent activity, and 18 had spent time in prison All names used in this manuscript are pseudonyms to conceal the identities of our participants.

Data Gathering
Interviews were conducted using a semi-structured interview protocol in private settings such as hotel rooms and residential homes and public settings such as restaurants and coffee shops.Participants were asked to describe their childhood experiences as an initial starting point.The interviews included questions about broad phases of the participant's extremism such as entry, involvement, and disengagement, with probes to encourage participants to elaborate on aspects of their life histories.While participants were periodically asked direct questions to focus on specific topic areas, the interviews relied on an unstructured format intended to generate unsolicited data embedded in their personal narrative.We view the elicited narratives as instructive in terms of assessing how individuals make sense of their lives (Blee, 1996;Copes, Hochstetler & Forsyth, 2013;Giordano, Johnson, Manning, Longmore, & Minter, 2015;McAdams, 2013).Each interview concluded with more structured questions and scale items to collect comparable information across interviewees in terms of risk factors (e.g., history of child abuse, mental health problems, etc.), demographic information, and criminal histories.The interviews lasted between four and eight hours and generated 4,578 pages of transcripts, which indicate the level of detail provided by the life histories.

Data Preparation and Coding
Two graduate students trained in the psychology of violence, ideological extremism, and life narratives read the interview transcripts line-by-line.The initial coding process examined all phases of our participants' life-histories including childhood, extremist involvement, and exit.The unit of analysis for the current study, however, is the 'disengagement event,' which refers to a specific episode that generated doubts and/or movement away from extremist affiliations.In situations where raters identified multiple, unique disengagement events, each extract was weighted according to prominence (i.e., fluency of discussion) and proximity to the participant's departure.If the same disengagement event was discussed at different times throughout the interview, separate extracts of text were combined for the purpose of analyses.After all the interviews were coded, raters selected each participant's most influential disengagement event for further analyses.
Once extracted, raters thematically analyzed the participants' disengagement event at a broad level by noting holistic, emergent reasons as to why participants disengaged (e.g., drugs, family, incarceration).These codes were considered 'disengagement themes,' which refers to the underlining motif (e.g., disillusionment, exposure to diversity) conveyed during the participant's departure.While raters selected the most pervasive disengagement event for each participant, there were no restrictions regarding the number of disengagement themes that may characterize a participant's exit.Disengagement themes were coded dichotomously as present or absent for each participant.The coding process was iterative, ensuring that each participant's disengagement event was reviewed whenever a new disengagement theme emerged.Raters compared and contrasted disengagement themes, noting features between them, and moving back and forth between first-level data and general categories (Glaser & Strauss, 1967;Miles & Huberman, 1994).Once all participants were coded, final ratings were discussed and reviewed among all authors for quality assurance.Overall, 12 distinct disengagement themes were identified (see Appendix for description disengagement themes).
In addition to the disengagement themes, raters also relied on a content coding system derived from the CMA to identify markers of emotion expressed by participants during their disengagement event (see Table 1 for a complete list of the circumplex affect variables).Raters coded each disengagement event and identified as many affective markers that could be coded using the circumplex affect codebook (Posner et al., 2005;Russell, 1980;Yik, Russell, & Steiger, 2011).While the retrospective nature of the life history interviews raises questions about the temporal order related to emotions and disengagement, raters were careful to note subtle expressions (e.g., "was"; "started getting"; "became") that participants used to describe disengagement events.As such, raters tried to capture what participants said they were feeling at the time of disengagement.After researchers completed the CMA coding process for each participant, frequencies for all twelve circumplex affect variables were summed and recorded.Because the narratives were embedded with emotions like hatred, anger, and frustration, the CMA offered a structured coding system to examine each participant's behavioral, cognitive, and affective markers as described in their disengagement events.The CMA also allowed us to determine whether disengagement events evoked positive or negative affect as well as the intensity in which the participant experienced these emotions.

Circumplex Affect Results
Two affective markers were identified as the most common throughout our sample: activated displeasure and displeasure.Both of these affective markers tap into an overall emotional state related to anger (Berkowitz & Heimer, 1989;Yik et al., 2011), which supports our central argument that negative emotions (i.e., anger and frustration) directed toward the extremist group and oneself function as a catalyst for disengagement.These negative emotions are a source of motivation to re-evaluate the relative importance of the group vis-à-vis the individual's lived experience within the group.The most common circumplex affect variable identified was activated displeasure.Activated displeasure represents agitation, annoyance, fearfulness, frustration, hostility, irritability, and tension (e.g., "Knowing the fact that they [fellow group members] are not there for you… They want you to take a bullet, but they don't want to support you." -Mark, Interview 29, 2014).In terms of activated displeasure, participants exhibited an average of 4.05 (SD = 4.46) expressions, ranging from 0-18.Participants attributed their anger and frustration to numerous factors such as inadequate organization and leadership and interpersonal relationships characterized by substantial conflict.
The second most common emotion identified among our participants was displeasure, which encompasses several related emotional states such as dissatisfaction, unhappiness, troublesomeness, and discomfort.For instance, one participant described the moment he realized displaying a swastika was offensive: "I knew it was offending people, but for the first time I felt bad for offending someone with it" (Dillon, Interview 12, 2013).In terms of displeasure, participants exhibited an average of 3.53 (SD = 2.19) expressions, ranging from 0-11.Overall, disengagement events that led participants away from an extremist lifestyle were characterized by general unhappiness, frustration, and dissatisfaction with group behaviors.These affective states create interruptions in the identity cycle, which, in turn, produces social stress and, a high degree of anger toward the extremist group (Stets & Tsushima, 2001).Individuals who leave violent extremism respond to these negative emotions by reducing the relative importance of their extremist identity, which begins the process of disengagement.

Disengagement Theme Results
Each participant exhibited an average of 2.88 (SD = 1.18) disengagement themes ranging from 1-6 themes in his/her central disengagement event.These results broadly suggest that disengagement from ideological extremism can occur for a variety of reasons and that people disengage based on a complex constellation of situational and personal factors.Figure 1 shows the number of participants exhibiting each disengagement theme.Although not exhaustive, the following section illustrates the three most common disengagement themes: 1) disillusionment, 2) violence, and 3) familial and peer relationships.In doing so, we focus particularly on how anger directed toward the self and the extremist group contributed to disengagement processes.Disillusionment.There are numerous motivations for participation in extremism.Broadly speaking, people join extremist groups for acceptance and belonging as well as a desire to fulfill personal achievements (Bjørgo, 2011;Schafer, Mullins, & Box, 2014).In terms of disengagement, by far the most common theme expressed across the sample involved disillusionment, which is best understood as the realization that a consistent incongruence exists between idealized expectations and the everyday realities associated with those same expectations (Casserly & Megginson, 2008;Ebaugh, 1988).In this sample, 26 participants (65%) were identified as experiencing disillusionment.For instance, the following participants described their dissatisfaction with the lack of loyalty and integrity among group members.
It's a whole bunch of hypocrites, back stabbing, and the whole movement is kind of a joke.There's no such thing as fucking white power; some of these dudes might fully believe it's "white pride, white power."You learn at some point, it's all just a joke.It's a fucking scam.(Bradly, Interview 7, 2014) I was starting to have doubts … Over the years, you start looking at the people who are preaching white power and doing drugs.Okay, I drank a lot.[But] they are doing hard drugs.You watch a guy with a needle in his arm and he was out there preaching today about drugs and then in here with a fucking needle in [his] arm.You start seeing things and you just start getting really discouraged… you are looking at a cause that is full of fakes.I felt I was a fake.(Seth, Interview 3, 2014) Each example illustrates individuals who experienced disillusionment in response to hypocrisy and "back stabbing."Disillusionment often leads to a disconnection between prior expectations and reality, which is manifested when the individual comes to see the group as illegitimate.This disconnection can produce distrust (Windisch, Ligon, & Simi, 2018), stress (Burke, 1991) and anger (Stets & Tsushima, 2001) directed toward the extremist group and fellow members.In turn, the individual's anger toward extremism creates definitional clarity or the "crystallization of discontent" (Paternoster & Bushway, 2009) by focusing on extremist involvement as an undesirable investment of the person's time and energy.As such, the emerging definitional clarity provides an impetus for disengagement from extremism.From this perspective, anger is an essential motivational or energizing emotion (Collins, 2004;Katz, 1999).In the next section, we discuss how violence influences disengagement.
Violence.For fourteen participants (35%), negative experiences with violence were identified as an important factor leading to disengagement.Violence in relation to disengagement events can take one of two forms.First, violence typically involved interpersonal conflicts among group members over romantic relationships, money or respect; a finding consistent with the study of conventional street gangs (Decker & van Winkle, 1996;Fleisher, 1998;Jankowski, 1991).According to several participants, consistent in-fighting among members reduced the legitimacy of the group and produced frustration with the inability of leaders to manage group dynamics.
Second, violence also involved witnessing or participating in aggressive action directed toward bystanders such as children, the elderly, inter-racial couples, or members of the LGBTQ community.In some cases, participants described a general unwillingness to commit violence against people the group considered a target.These findings support previous research that examines "thresholds" for the level of violence an individual is willing to tolerate before disengaging from the group (Decker & Lauritsen, 2002;Reiner, 1992).Participants also described substantial guilt and emotional burden when violence was carried out against members of outgroups.For instance, the following participants explained how violence contributed to their exit from violent extremism: Women are thrust into conflicting roles.You're to be a mother.You're to pop out babies.That's your biggest achievement, that's the only important thing really you can do.At the same time, you are supposed to be able to fight, to survive [and] to protect.I'm supposed to cherish a child's life, but be willing to take it… when I saw the [Oklahoma City Bombing] and put that image together with what the expectations were for me.I don't know if it was seeing a child in that state, or the realization that I could be the person causing that… It was like hitting a brick wall.(Abby, Interview 22, 2013) My last hate crime happened about a month after my son was born.I'd beaten that guy with a hatchet… The butt.It was a small, pocket hatchet that I carried around...So when that incident happened, I felt guilt in a way I'd never felt it.I'd always felt it.I would commit a crime, drunk usually.It would be so bad for ten minutes.After I got away, I would be throwing up from the stress.For years that's what happened.So I experienced guilt this time differently.That was it, I was done.(Doug, Interview 31, 2014) In both instances, the individuals described an incident that crossed a threshold that became pivotal to redirecting their anger toward the group.In Doug's case, the violent incident and the ensuing guilt became a breaking point and his feelings of anger and guilt motivated him to completely sever his ties with extremism.
At other times, violence may occur internally among members of the same group.These violent interactions may spark a period of reflection where the individual questions the group's political agenda.For instance, Kevin described how he felt following an encounter with a fellow member who was beaten by the group: It was probably around the holidays, with stuff that happened with George.Yeah we hurt him bad.His buddies, right, his bros shattered his jaw.Doesn't compute.That's a wrong idea of a buddy or a bro, "My buddies just kicked my ass."I mean it left me… I see him walking down the street he's got his jaw wired shut.It was stuff like that.I think it all kind of built on each other.(Kevin, Interview 17, 2014) As Kevin's experience illustrates, anger, frustration, and guilt can be directed inward toward the participant's own behavior as well as toward group activities.Violence-both internal and external-can reduce one's commitment to the group's ideological beliefs by producing feelings of guilt, paranoia, and betrayal.Additionally, individuals may begin to burn out because of the demanding lifestyle and extreme emotions produced from engaging in violence (Bjørgo, 1997;Kimmel, 2007).Adverse emotional reactions to violence stem from discrepancies between an individual's expectations and actual lived experiences related to violence.These findings underscore the potential for violence to have a physical and emotional strain on some individuals.We now discuss another common source of disengagement identified within our sample: family and peers.
Familial and peer relationships.For eighteen participants (45%) in our sample, the presence of a positive familial or peer relationship represented a major disengagement theme in their extremist careers.Familial relationships include children, intimate partners, and distant relatives (e.g., uncles, grandparents).By peer relationships, we mean friendships and acquaintances close to the participant such as childhood friends or coworkers.Obligations to these relationships outside of the movement created a conflict between loyalty to the group and responsibilities to their families and friends.
In line with previous findings (see Bjørgo, 2011), children were identified as the most common type of familial relationship (n = 7) related to disengagement from extremist activities.Most research that focuses on parenting and desistance has found mixed results.For instance, several studies have found that maternal and paternal roles acted as motivators that changed the individual's future orientation, outlook and sense of responsibility (Brown & Bloom, 2009;Edin, Nelson & Paranal, 2004).Alternatively, parenting has also been identified as a source of stress as the individual balances their role as a parent with their prior criminal identity (Farrington & West, 1995;Massoglia & Uggen, 2007;Rand, 1987).
The current analysis found evidence in support of the positive outcomes related to parenting.Specifically, parental roles functioned as a turning point away from extremist behavior because these individuals had fewer opportunities to participate (see also, Moloney, MacKenzie, Hunt, & Joe-Laidler, 2009).In fact, several participants indicated that their extremist lifestyle was no longer compatible with their new role as a parent.Disengagement events involving familial relationships were less likely to involve anger than other sources of disengagement.Anger, however, did play a role in a small portion of these events.In these cases, participants experienced anger which was directed internally.For these individuals, the presence of family and peer relationships provided an opportunity to redefine their orientation toward the group by crystalizing discontent in terms of their group affiliation (Paternoster & Bushway, 2009).
Previous research also finds significant others outside of the extremist group provide important reference points and sources of motivation during the change process (Aho, 1994;Gadd, 2006).For example, significant others may teach an individual that a hateful ideology is counter-productive (Blazak, 2004).For example, I was still living a fucked up life as far as the ugly things but my heart was coming back.I started to care about others… So, me and my mother were never good… I think she found out she was alone after my dad died.And I was alone the minute he died.But like I said she found out I was in Iowa, she got me to come over, eat some food and we started talking.I started drying out a little bit.She had to move me [into her house].I still kept the beliefs and finally one day she said, "can you come here a second?" and I'm like "what?"She said, "Your dad, taught you better than this.Time to get it together son." (Blake, Interview 35, 2014) As Blake's feelings of antipathy toward himself grew, he relied on family support until all ties with extremism were severed.In this way, the family functioned as an escape from the "fucked up life" that encouraged his anti-social behavior.Indeed, evidence suggests that the role of exhaustion as a consequence of engaging in violent behavior is an important factor of the exit process (Gallant, 2014).
Peer relationships also contributed to several (n = 9) of our participants' disengagement.In some cases, participants reported developing new friendships outside of the group.Over time, these relationships became more prominent than those centered on extremism.At other times, participants reported a desire to leave once their friends in the group left.In the following example, Brian described experiencing doubts about the movement, but he did not leave until his friends also decided to quit.In this respect, Brian's exit reflected a "block defection" (McCauley, 2008) where multiple members of a group disengage simultaneously: I had doubts in the back of my head, but I was staying because I got all my friends involved.Then, when they were out, I was like, "Well, I'm not staying.I was just staying in because I got all of you guys in.So, I'm out, too."We all stayed friends for a while… Within three or four months, they ended up joining a similar crew, which I wanted nothing to do with… They were drawing a line saying, "If you go to their concerts and all this stuff, then you're written off."I was like, "Don't tell me what to do."That's how everything fell apart.(Bryan, Interview 15, 2014) Brian's experience underscores our central argument in which anger directed towards the group functions as a push factor away from group activities.Similar to Brian, several participants (n = 13) indicated they had previously experienced doubts regarding their involvement in extremist activities but continued to "go along with it."In these situations, family and peer relationships were a source of escape from their extremist lifestyle.
Overall, participants who reported familial and peer relationships as their main source of disengagement from violent extremism shifted their focus away from extremist activities toward those relationships.As a result, this shift diminished previous commitments to violent extremism and replaced them with commitments to children, romantic partners, and peers.For the majority of our participants, anger directed toward the group allowed them to create distance between themselves and their collective identity by recognizing the potential dangers and contradiction of their actions.
Generally, participants encountered a broad range of experiences that contributed to their eventual disengagement from violent extremism.The themes we identified represent environmental and organizational obstacles toward achieving goals related to extremist involvement.At least 12 broad themes related to disengagement were identified across our sample, suggesting there is no single reason as to why white supremacists disengage.Across these broad themes, however, we found the presence of anger and frustration consistently directed toward the group and oneself.Disillusionment represents the most common theme expressed among the participants in our sample followed by familial and peer relationships.Violence directed internally or externally was the third most common theme.

CONCLUSION Theoretical and Practical Implications
This study takes one of the first steps to underscore the relative importance of anger related to an individual's decision to leave extremism.While Giordano and colleagues (2007) found the reduction in anger contributes to an individual's desistance from crime, our study of disengagement from violent extremism discovered alternative findings: the emergence of anger, specifically active anger, can promote disengagement from a violent group.This finding refines Giordano and colleagues' ( 2007) original argument and points to one potential point of distinction between generic criminal offending and involvement in violent extremism.That is the presence of group membership may result in slightly different mechanisms and alter how social processes unfold.Alternatively, Giordano's findings may point to a different temporal phase of the desistance process.The reduction of anger may occur at a later point that is initially preceded by a crystallization of anger directed toward an existing criminal lifestyle including criminal peers or in our case an extremist group.
The identification of how people feel about certain factors related to exit has important implications in terms of interventions aimed at accelerating disengagement from extremism.As we have illustrated with our empirical data, disillusionment, familial and peer relationships, and violence are important factors contributing to the naturalistic unfolding of disengagement processes.We suggest that these findings may also be beneficial for intervention programs, specifically initiatives aimed at developing counter-messaging strategies.As an alternative to narratives designed to challenge the accuracy of a group's ideological belief system-narratives that may unintentionally reinforce extremists' attachment to the ideology rather than highlight inaccuracies (Aly, Weimann-Saks, & Weimann, 2014)-we suggest focusing on techniques to amplify the various sources of disengagement illustrated in our findings.In particular, countermessaging strategies could underscore the hypocrisy among group leaders and other members.In addition, these intervention strategies could accelerate negative emotions toward the group by highlighting the high rates of intra-group violence as well as the negative impact involvement in violent extremism is likely to have on family and peer relationships.

Limitations
There are several limitations of this study worth noting.First, the retrospective nature of the life history interviews raise questions about validity and reliability due to memory erosion, distortion, and selective recall (Baddeley, 1979).The practice of remembering is a reconstructive process where memories of events are typically reinterpreted during each recall (Bridge & Paller, 2012).Despite these concerns, the rich life history accounts provide important insight from the participants' perspectives.Second, due to the relatively hidden nature of this population, the sample was derived through snowball techniques and, as a result, is not representative.Although snowball sampling minimizes the generalizability of the results of this study, the goal of qualitative research is the identification of social processes, conceptual elaboration, and describing causal mechanisms.Third, the sample was predominantly male, which may have yielded data subject to gender biases.Finally, our sample primarily included individuals who joined white supremacy during adolescence, potentially neglecting differences in disengagement among individuals who entered during pre-adolescence or adulthood.

Future Research and Conclusion
The results of this study provide further insight into the disengagement process in several ways.First, the methodology offers one of the first studies in criminology to examine the overlap of emotions by employing the CMA to study the disengagement processes.Although there are limitations in the ability to generalize these findings, the life history data gathered from former US extremists provides information inaccessible through other methodological techniques.Second, the conceptual framework we used elaborates on previous studies of criminal desistance and disengagement from extremism by offering greater precision in terms of understanding the impact of cognitive and emotional processes as they relate to a person's decision to leave extremism.
Although the results highlight the importance of disillusionment, future research should focus in greater detail on the different types of disillusionment and whether different types of disillusionment represent unique pathways toward to disengagement.Future research should also consider applying Giordano and colleagues (2015) concept of "learning curves" to the issue of disengagement from violent extremism.Traditionally, criminal desistance has been treated as singular experiences such as getting married, gaining employment, and enlisting in the military.Recently, however, Giordano and colleagues (2015) have focused on the processual dimensions of desistance.Along these lines, the authors argue that specific events are unlikely to sustain desistance without proper "anchors" that help develop fundamental relationship adjustments (Giordano et al., 2015, p. 9).While the current study highlights several anchors related to disengagement from extremist groups, future research should examine the unique conditions in which these anchors fail to curb extremist activities.By understanding more about these mechanisms, and others as well, we may not only understand disengagement more deeply, but we can also appreciate how different phenomena may derive from common underlying processes.

Figure 1 .
Figure 1.Number of Former White Supremacists Who Expressed Each Disengagement Theme