Reading I NEED HELP ON MY HOMEWORK

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=uter20

Studies in Conflict & Terrorism

ISSN: 1057-610X (Print) 1521-0731 (Online) Journal homepage: https://www.tandfonline.com/loi/uter20

Terrorist Transformations: The Link between Terrorist Roles and Terrorist Disengagement

Mary Beth Altier, Emma Leonard Boyle & John G. Horgan

To cite this article: Mary Beth Altier, Emma Leonard Boyle & John G. Horgan (2020): Terrorist Transformations: The Link between Terrorist Roles and Terrorist Disengagement, Studies in Conflict & Terrorism, DOI: 10.1080/1057610X.2019.1700038

To link to this article: https://doi.org/10.1080/1057610X.2019.1700038

Published online: 21 Jan 2020.

Submit your article to this journal

Article views: 906

View related articles

View Crossmark data

Citing articles: 2 View citing articles

Terrorist Transformations: The Link between Terrorist Roles and Terrorist Disengagement

Mary Beth Altiera , Emma Leonard Boyleb, and John G. Horganc

aCenter for Global Affairs, New York University, New York, New York, USA; bDepartment of Political Science, La Salle University, Philadelphia, Pennsylvania, USA; cDepartment of Psychology, Georgia State University, Atlanta, Georgia, USA

ABSTRACT Research pays little attention to the diverse roles individuals hold within terrorism. This limits our understanding of the varied experi- ences of the terrorist and their implications. This study examines how a terrorist’s role(s) influence the likelihood of and reasons for disengagement. Using data from autobiographies and in-person interviews with former terrorists, we find that role conflict and role strain increase the probability of disengagement. We show those in certain roles, especially leadership and violent roles, incur greater sunk costs and possess fewer alternatives making exit less likely. Finally, certain roles are associated with the experience of different push/pull factors for disengagement.

ARTICLE HISTORY Received 27 September 2019 Accepted 25 November 2019

To hold a gun is to hold power, but even as I felt the excitement of firing live bullets, I knew that I could never use a gun on another human being. I did not mind setting up operations in which people would be shot, but I knew my limitations: I would never be a shooter.

Eamon Collins1

Unearthing why and how, as well as which, individuals choose to leave terrorism behind has important implications for the design of policies intended to facilitate or trigger defection. It also informs the risk assessment of terrorist detainees and the design of disengagement or “de-radicalization” programs. While recent studies examine why and how terrorists disengage,2 they pay little attention to the different roles individuals hold within terrorism. Evidence, however, suggests that terrorists’ roles are important for dis- cerning profiles and pathways into terrorism3 and developing more valid instruments for individual risk assessment.4 As such, greater attention to terrorists’ roles may also offer important insights for our understanding of pathways out of terrorism. Given the variety of roles within a terrorist group, it is plausible that individuals in

certain roles may incur greater sunk costs and/or possess fewer alternatives as a result of involvement making disengagement more difficult. Individuals in certain roles may also be more prone to experience specific push or pull factors for disengagement (e.g. difficulty adapting to the clandestine lifestyle, difficulty carrying out attacks, interactions with moderates). Further, as the opening quotation by Collins highlights, terrorists often

CONTACT Mary Beth Altier [email protected] Center for Global Affairs, New York University, New York, NY, USA. Supplemental data for this article is available online at https://doi.org/10.1080/1057610X.2019.1700038.

� 2020 Taylor & Francis Group, LLC

STUDIES IN CONFLICT & TERRORISM https://doi.org/10.1080/1057610X.2019.1700038

possess expectations about what their involvement will entail. Assignment to tasks incompatible with one’s expectations or skillset may result in disillusionment or dissatis- faction that then drives disengagement decisions. Altier et al.5 found experiencing disillusionment with day-to-day tasks increased the likelihood of voluntary disengage- ment by 55 percentage points. A terrorist’s role, therefore, may have important implica- tions for understanding whether an individual is likely to disengage, why the individual might choose to leave and what factors may be most effective in precipitating and sus- taining disengagement. To investigate these claims, we use detailed qualitative and quantitative data gleaned

from 87 autobiographies authored by individuals involved in terrorism as well as nine semi-structured interviews with former terrorists. Using the autobiographical data, we compare 49 cases of individual, voluntary disengagement from terrorism with a com- parison group of 54 cases of individual, involuntary disengagement (usually as a result of imprisonment) where the individual immediately reengaged upon their release under the premise these individuals would have otherwise remained involved in terrorism. Our evidence suggests first, that individuals experiencing role strain – a mismatch between their abilities and assigned role – and role conflict – conflicting roles within or outside the group – are more likely to leave terrorism. Second, those in leadership and violent roles are less likely to leave terrorism behind and those in support roles are more likely, even when controlling for ideological commitment and push factors likely to influence their satisfaction with involvement. This, in combination with our qualita- tive evidence, suggests leaders and violent operators incur greater sunk costs and pos- sess fewer alternatives. Finally, we show individuals in certain roles are more likely to experience specific push and pull factors associated with disengagement. We begin by reviewing what a “role” is, highlight how greater attention to terrorists’

roles might better inform our understanding of disengagement and put forth a series of hypotheses. Second, we summarize our autobiographical and interview data and methods. Third, we test the plausibility of our hypotheses and present the results. We conclude by summarizing our findings, their limitations, the implications for counter- terrorism policy and directions for future research.

Terrorist Roles and Terrorist Disengagement

What is a Role?

Individuals involved in terrorist groups,6 like any conventional organization, may carry out different tasks or serve different functions. An individual’s role within an organiza- tion, according to Krantz and Maltz,7 who have studied traditional work organizations, is comprised of two parts: role as given and role as taken. Although leaders may assign certain roles or sets of tasks to individuals within the organization (role as given), indi- viduals interpret their roles in unique ways (role as taken).8 Consistent with Krantz and Maltz, we believe an individual’s role should be conceptualized not just as what the individual is told or assigned to do, but also how the individual understands, interprets and carries out the tasks within the normative framework and social context of the group.

2 M. B. ALTIER ET AL.

What Do We Know About Terrorist Roles?

To date, there is little systematic research on the various roles individuals occupy within terrorist groups. Although research has yet to establish a typical terrorist profile,9 schol- ars still tend to examine terrorists as if they were a homogenous entity with similar characteristics. Such approaches ignore key differences in the individuals who become involved in terrorism and how their experiences while involved shape their attitudes and behavior, including their commitment to the group and the likelihood they will remain involved. Indeed, Ross10 noted while terrorists bring certain knowledge, skills and personal beliefs to the group, they are subsequently exposed to various learning experiences that cause them to adopt different roles, orientations and behavior. Recent research demonstrates greater attention to roles and functions within a terror-

ist group may help us more systematically identify profiles and pathways into terrorism. Perlinger et al.’s study of violent Islamist terrorists across a range of networks found the disaggregation of their sample by roles and functions (violent participation, skilled/ unskilled, position and tenure, network connectivity) yielded statistically significant soci- oeconomic attributes or profiles.11 Gill and Horgan’s study of 1240 Provisional Irish Republican Army (PIRA) members similarly concluded that certain sociodemographic characteristics emerged when one disaggregated by role(s) (gunrunner, bomber, bomb- maker, gun-man, nonviolent).12 Likewise, Simcox and Dyer’s study of 171 individuals convicted of Al-Qaeda related or inspired offenses in the U.S. finds the disaggregation of the sample into five types of terrorist activity (active participants, aspirants, facilita- tors, trained aspirants and ideologues) yields different sociodemographic profiles.13

While research has considered the relationship between roles and profiles and pathways into terrorism, the purpose of this study is to consider their influence on possible path- ways out of terrorism.

How Roles Relate to Disengagement

We define terrorist disengagement as the cessation of terrorist activity and note that it connotes a behavioral change and, thus, is distinct from the process of de-radicalization, which entails an attitudinal change or loss of faith in the ideology underlying a terrorist cause. Existing research tends to categorize the factors that drive terrorists’ disengage- ment decisions into a push/pull framework (Table 1).14 Push factors are those aspects of involvement in terrorism that drive individuals away from the group, while pull factors are factors outside of involvement that lure individuals to a more conventional social

Table 1. Factors for terrorist disengagement.75

Push factors Pull factors

� Unmet expectations (including disillusionment with day-to-day tasks)

� Disillusionment with strategy/actions of group � Disillusionment with personnel � Difficulty adapting to clandestine lifestyle � Inability to cope with physiological/

psychological effects of violence � Loss of faith in ideology � Burnout

� Competing loyalties � Positive interactions with moderates � Employment/educational demands or opportunities � Desire to marry/establish a family or family demands � Financial incentives � Amnesty

STUDIES IN CONFLICT & TERRORISM 3

role. Recently, Altier et al. noted that although the push/pull framework provides a use- ful heuristic listing those factors most commonly associated with disengagement deci- sions, it remains theoretically underdeveloped and lacks predictive value as certain individuals may experience one or some of these factors and never disengage.15 Further, the framework fails to specify which factors or which combinations of factors are most important in driving disengagement decisions within the terrorist population, for certain groups, or for certain types of individuals.16

In order to move beyond the push/pull framework, Altier et al. proposed Rusbult and colleagues’ investment model from social psychology.17 The model specifies one’s com- mitment to (or probability of exit from) a given organization, relationship or social role at any given point in time is a function of three factors: (1) the extent to which the individual obtains satisfaction from involvement, (2) the perceived quality of available alternatives and (3) the investments (or sunk costs) accrued (Figure 1). The satisfaction component is a function of the actual rewards and costs vis-�a-vis those an individual expects from involvement. There is substantial empirical evidence in support of Rusbult’s investment model in

explaining exit decisions across a variety of organizations18 and relationships or social roles.19 Research also suggests satisfaction may be more pivotal in predicting exit than alternative quality and that some weighting of the components may be in order.20

Nevertheless, the model provides a more dynamic and comprehensive theoretical frame- work for understanding terrorist disengagement that takes into account why certain push/pull factors matter for some individuals and not others, the interaction or cumula- tive effects of different push/pull factors, and why the experience of certain push/pull factors may not precipitate disengagement for an individual at one point in time, but may be pivotal later.21

The investment model suggests the role(s) individuals hold while involved in terrorism may influence the likelihood of disengagement and the potential causes of disengagement in several ways. First, individuals may enter a terrorist group with a clear sense of what role(s) they want to assume (e.g. bomber, martyr, assassin, recruiter, trainer) and the expected rewards or satisfaction they will derive from involvement, including their day- to-day tasks. For example, an individual who would like to take on a violent operational role might not be satisfied when ordered to work as a recruiter or fundraiser. Indeed, the inability to attain the role one desires within a terrorist group may result

in what sociologists term role strain.22 Role strain occurs when an individual lacks the requisite knowledge, skills or traits to successfully fulfill their assigned role. An individ- ual, for example, who has trouble coping with carrying out violent acts, may experience

Figure 1. Rusbult’s investment model.

4 M. B. ALTIER ET AL.

role strain if assigned to be a bomber or an assassin. Role conflict, on the other hand, occurs when an individual occupies multiple roles that place competing or conflicting demands on the individual.23 Someone, for instance, who is both a father/husband and a member of a terrorist group, may experience role conflict if tasked with traveling or remaining in hiding for long periods of time precluding him from seeing his family. Individuals may also experience role conflict as a result of multiple roles held within the same terrorist group that are difficult to balance. Role conflict and role strain may generate dissatisfaction,24 thereby depleting one’s commitment to involvement and increasing the likelihood of exit. Indeed, early research in industrial and organizational psychology suggests role conflict and role strain as well as role ambiguity (i.e. unclear or undefined expectations about what one’s tasks entail) are associated with low job satis- faction and turnover in traditional workplace settings.25 We summarize our hypotheses regarding role strain and role conflict below:

H1: Individuals experiencing role strain are more likely to disengage, all else equal.

H2: Individuals experiencing role conflict are more likely to disengage, all else equal.

Additionally, certain roles may be associated with higher investments or fewer alter- natives. A terrorist trained, for example, in a very specialized role such as a bomber may possess few transferable skills, have forensic evidence tying him/her to crimes, and a reputation with law enforcement and even the community at large. This may hinder the individual’s exit, despite the presence of substantial dissatisfaction. Individuals involved in more peripheral or less criminal roles, on the other hand, may face far fewer barriers to exit. Thus, we hypothesize:

H3: Individuals in leadership roles incur greater investments and possess fewer alternatives and are less likely to disengage, all else equal.

H4: Individuals in violent, operational roles incur greater investments and possess fewer alternatives and are less likely to disengage, all else equal.

Finally, individuals in particular roles may be more likely to experience certain push/ pull factors associated with disengagement. Looking at Table 1, it seems, at least on the surface, that push/pull factors such as loss of faith in the ideology, or the desire to marry as one ages are equally likely to occur regardless of one’s role(s). Other push/pull factors, such as the inability to cope with the physiological and/or psychological effects of violence, difficulty adapting to the clandestine lifestyle, or positive interactions with moderates, seem more or less associated with certain roles. For example, individuals who carry out violent attacks (e.g. bombers, snipers) or who play an integral part in violent operations (e.g. getaway drivers, lookouts) may be more likely to experience dif- ficulty coping with the physiological or psychological effects of violence given their proximity than those solely involved in nonviolent operations (e.g. publicity/propa- ganda, fundraising). Likewise, individuals involved in certain nonviolent roles may lead more open lifestyles and, as such, may experience less difficulty adapting to a clandes- tine lifestyle and may be more likely to experience frequent and sustained interactions with moderates. Thus, we hypothesize:

H5: Individuals in certain roles may be more (less) likely to experience certain push/pull factors commonly associated with disengagement.

STUDIES IN CONFLICT & TERRORISM 5

Data and Methods

Terrorists and even many former terrorists are clandestine populations that wish to evade detection, rendering methods such as random sampling and surveying impossible. To explore the possible relationships between one’s role(s) within a terrorist group and disengagement, we therefore turned to qualitative and quantitative data collected from 87 English-language autobiographies authored by members of terrorist groups. We also conducted semi-structured interviews with nine individuals formerly involved in a ter- rorist group. The dataset obtained from the autobiographical accounts provides system- atic information on terrorist roles and how they might relate to disengagement over an extended period of time for a diverse sample of terrorists across a range of groups – information that would be time consuming and difficult to obtain through in-depth interviews. The in-depth interviews, conversely, allow us to probe more deeply and ask specific questions relevant to our research. Recent studies highlight the potential insights autobiographies26 and in-depth inter-

views27 offer into the experiences and decision-making processes of terrorists. There are, of course, limitations. There may be something systematically different about those who choose to give an interview or publish a memoir.28 Further, their accounts may be subject to hindsight bias. They also may have incentives to portray their involvement in terrorism in a certain light and may misrepresent, embellish or omit key facts.29

Interviews are further subject to reactivity, in that the behavior of or questioning by the researcher may inadvertently influence the account an interviewee provides. Autobiographies, on the other hand, restrict the researcher to an account of only what the author or editor deems interesting. With regard to roles, the autobiographical data therefore may provide an incomplete picture of the number and type of roles held by individuals, as they may not discuss certain roles or key tasks associated with those roles. The omission may be deliberate, for instance, to protect themselves or other group members or may be inadvertent, in which case omission arguably would occur with shorter-held or less encompassing roles. Despite these limitations, we believe the triangulation of autobiographical and interview data offers important insights into this clandestine population, which is not easily studied by other means. Indeed, these accounts are often more revealing than intended.30

Autobiographical Data

The autobiographical accounts represent the life histories of 85 unique terrorists31 and cover individuals engaged in a range of terrorist groups including those motivated by nationalist, extreme right-wing, extreme left-wing, religious and single-issue (e.g. envir- onmental, antiabortion) ideologies. Since the purpose is to understand disengagement from terrorism, the codebook and dataset are arranged by engagement event, rather than by autobiography or individual terrorist. An engagement event begins when an individual engages in terrorist activity32 on behalf of the group and ends when they stop, whether voluntarily or involuntarily. Thus, an individual terrorist may experience multiple engagement (and disengagement and reengagement) events over his/her life course and hold different roles during each of these periods of involvement. The final sample consists of 185 engagement events.33 Forty-nine culminated in voluntary,

6 M. B. ALTIER ET AL.

individual disengagement (i.e. the individual choosing to leave an active terrorist group). For the purpose of quantitative analysis, up to five roles held by the individual during a particular engagement event and the role held at the time of engagement and disengagement are coded. When individuals held multiple roles at the time of their engagement or disengagement, the most salient roles are recorded. As the exact or even rough time frame in which individuals held a particular role is unclear in a large number of accounts, the data do not include measures of how long one served in a particular role. Often, individuals did not have formally defined roles. Some would carry out disparate

assigned tasks and were simultaneously involved in multiple “roles”. For example, Omar Cabezas described his experience with the Sandinista Nationalist Liberation Front:

Finding houses for compa~neros in the underground, or for meetings, or storing things, or for mail drops; coming up with cars, car repair shops; getting information about who Somoza’s informants were, spying on houses of girl friends of the Guard. So I started to work, doing everything they asked me to do and everything I thought needed to be done.34

In other instances, individuals possessed clearly defined, specialized roles and often an associated title. In the PIRA, Sean O’Callaghan noted he was “Operating Commander of the IRA Southern Command, and a member of the GHQ staff and the Sinn Fein National Executive”.35 Describing an earlier role he held, O’Callaghan explained his “job was to clean them [guns] and sort ammunition into the correct cali- ber to suit the guns before the bulk of them was moved to Northern Ireland a few days later”,36 a rather specialized task. However, even within highly structured organizations such as the PIRA, there seemed, at times, to be an obscuring of roles. Maria McGuire noted “under the pressure of events the division between people’s roles in Kevin Street [PIRA headquarters] was frequently blurred”.37

Given that organizational, operational and contextual differences largely determined the specific roles or tasks assigned to individuals or the titles given, we classified the roles held by those in the autobiographical sample into 15 general categories relevant across a range of groups. Our role categories emerged organically from the roles identi- fied in the autobiographies. The 15 role categories are listed in Table 2 as well as the role held at the time of one’s disengagement, and the further collapsed category (sup- port, violent, leadership) that we use for our logistic regression analyses.38 We compare the role held at disengagement for those engagement events that ended with an individ- ual, voluntary disengagement (n¼ 49) where the individual chose to leave, with our comparison group of engagement events that ended with an involuntary disengagement from terrorism (usually as a result of imprisonment or detention) where the individual immediately reengaged upon their release (n¼ 54) under the premise these individuals would have otherwise remained involved. We have data on the role held at disengage- ment for 48 of the 49 cases of individual, voluntary disengagement and for 51 of the 54 cases in our comparison group.

Interview Data

Our nine semi-structured interviews were conducted by one, and in some cases, two of the authors in the U.S., Canada, and the U.K. over a two-year period between 2011 and

STUDIES IN CONFLICT & TERRORISM 7

2013. Two universities’ Institutional Review Boards (IRBs) approved the research. Our inclusion criteria consisted of individuals 18 and older who could confirm they were former members of a terrorist group. We used a combination of purposive and snow- ball sampling, reaching out to individuals via email, telephone or a gatekeeper. Although our aim was to obtain an equal number of individuals from nationalist, Islamist, left-wing, right-wing and single-issue terrorist groups, our resulting interview- ees consisted of two nationalists, one Islamist and six right-wing former terrorists. All of the former left-wing terrorists that we contacted refused to be interviewed either for personal reasons or because our research was government-funded. No two individuals in our interview sample were members of the same terrorist group. Clearly, our interview sample is small and likely unrepresentative. However, these

interviews allow for probing and follow-up questions and therefore offer additional insights on the importance of one’s role(s) for disengagement that cannot be obtained from the autobiographies. Further, the triangulation of the interviews with the autobio- graphical data, which draw on a rich and fairly large sample of individuals representing over 70 terrorist groups, should help circumvent these limitations.

Analysis

Role Selection and Role Satisfaction

As we conjecture that role strain (H1) or role conflict (H2) may increase dissatisfaction and the probability of disengagement, it is important to first consider how individuals come to hold certain roles, formal or informal, within a terrorist group. A systematic review of the qualitative data obtained from the autobiographies suggests that, despite some exceptions, roles are largely assigned rather than selected.39 Individuals across a range of terrorist groups reported being told what their duties or specific role would be. Leila Khaled, a former member of the Popular Front for the Liberation of Palestine rec- ollected she “was assigned to go to Schular camp near Zarqa, to give public lectures”,

Table 2. Role held at time of disengagement. Total number/percentage

Role

Individual voluntary disengagement

(n¼ 48)

Individual involuntary disengagementþ

re-engagement (n¼ 51) Collapsed role

category

Executive leader 4/8.3% 10/19.6% Leadership Low-level leader 3/6.3% 7/13.7% Leadership Violent operator 5/10.4% 11/21.5% Violent Bomb-maker 0/0.0% 1/2.0% Violent Operational planner 1/2.1% 1/2.0% Violent Publicity officer 1/2.1% 3/5.9% Support/logistical Intelligence officer 1/2.1% 1/2.0% Support/logistical Security officer 0/0.0% 0/0.0% Support/logistical Political officer 1/2.1% 0/0.0% Support/logistical Trainer 2/4.2% 0/0.0% Support/logistical Financier 2/4.2% 3/5.9% Support/logistical Transporter/courier 4/8.3% 2/3.9% Support/logistical Materials acquisition/storage coordinator 4/8.3% 1/2.0% Support/logistical Recruiter 3/6.3% 2/3.9% Support/logistical General member 15/35.4% 9/17.7% Support/logistical Total 48/100.0% 51/100.0%

8 M. B. ALTIER ET AL.

was later “ordered to go and give another public lecture at Jabel Al-Taj school”, and then, “Dr. Habash ordered us to prepare petrol bombs to use against Hussein’s sol- diers.”40 A member of the Italian Red Brigades, who uses the alias Giorgio, commented, “But the ‘work’ itself was grueling. I had been assigned, and it had not been my choice, to the ‘manufacturing sector’.”41

Once individuals hold a high-ranking position, they seem to have more discretion over their tasks and role(s). For example, Abu Iyad, a high-ranking member of the Palestinian Liberation Organization and Fatah, recounts, “We took advantage of the intervening few days to launch a fund-raising drive through the ‘popular committees’ set up by our representatives in the early sixties …We created a counterespionage ser- vice … .”42 Nevertheless, Iyad was still assigned certain duties by his superiors. On not- ing the leadership’s decision to create an underground organization in Jordan, Iyad writes he was “designated as the head of the organization.”43

Individuals who have very little say in the role(s) they hold are more likely to experi- ence role strain, which may have important implications for their satisfaction with involvement. Abdul Salam Zaeef, a former member of the Taliban recounted: “I came to Afghanistan to take part in jihad, but found myself carrying out mundane tasks for other people.”44 Shane Paul O’Doherty similarly noted when he joined the PIRA:

But oh, the boredom of being a new IRA volunteer in 1970! It was excruciating … I was firstly attached to a “section” there where the entire emphasis was on defense. There were the endless “parades” in our Section Leader’s house, where we did little but learn to respond to march commands given in Irish and where we stood at attention, turned on the spot and stood at ease.45

O’Doherty, however, was able to initiate a change in his role. After continually expressing his desire to work in explosives, O’Doherty’s decided to independently carry out a successful, but unsanctioned, bombing that caused the group to:

Take me seriously when I again asked to specialize in explosives, and they promised to arrange for me to meet and work with the best man they had in the field. I had been catapulted into a minor prominence a year or two before I might otherwise have earned it.46

Others reported satisfaction with their initial roles. Raymond Gilmour, a former member of the Irish National Liberation Army (INLA) and a police informant explained, “As a junior member my INLA jobs were mostly menial duties, picking up guns and ammunition for the real hard men who were going to do the actual shooting. This suited me very well, however, for I’d not wish to point a gun at anyone.”47 Omar Nasiri, a former member of the Armed Islamic Group (GIA, Algeria), commented on his gunrunning, “I knew what I was doing, and it didn’t bother me. For me this was business. I was making good money, and the work was exciting. Of course, I knew where all these weapons were going.”48

Further, a recruit might be initially satisfied, even excited, in a role and then become dissatisfied as the day-to-day reality fails to live up to one’s expectations. Tass Saada writes of his assignment as a sniper in Palestinian Liberation Organization:

More than in weaponry, however, I needed training on the mental aspects of sniping. They taught me how to sit or lie absolutely still for hours on end. In fact, some of the exercises went on for three or four days, around the clock … That was my introduction to the Simonov, which has been specifically designed for assassinations from a distance. It had a

STUDIES IN CONFLICT & TERRORISM 9

silencer to reduce the noise of the shot and to limit the muzzle flash. I became eager to master this weapon and learn what it took to be a sniper.49

Soon, however, Saada grew disillusioned and became increasingly bored, even ques- tioning his use of violence: “Looking through the telescopic sight, I watched village peo- ple come and go. The hours dragged on. Why am I doing this? Why should I erase this life? I asked.”50

Ben Turok, of the African National Congress (ANC) similarly writes:

All this casework held some interest and gave satisfaction, but the chamber itself brought nothing but torture. Although reasonably experienced as a public speaker as a result of the Congress meetings, I had little gift for the debating style required in the chamber. Furthermore, I had an intense dislike of the formality required and, indeed, I refused to use the word ‘honourable’ in referring to members of the House … My job was not to persuade the House, but rather to reach the press, which had been permanently installed above the Speaker’s chair, and so I directed my speeches to them. But it led to sleepless nights in anticipation and I am not sure that I succeeded very well. For one thing, being absorbed in the Treason Trial and remote from Cape Town, it was difficult to do the necessary research on the local issues dealt with in the council. Also, I had no time to familiarize myself with the complex parliamentary procedure.51

The qualitative data further suggest that role conflict, in addition to role strain, may generate dissatisfaction with involvement. David Hamilton, of the Ulster Volunteer Force, writes:

Shortly after I had joined we planned a bank robbery. I was annoyed because they planned the heist for Thursday morning. That was the day, I had to pick up my paycheque so I couldn’t take part in the robbery.52

A female member of an Islamist extremist group described conflicting work and familial roles, the stress it created, and how she tried to create boundaries for herself and her family:

In the beginning it was fine, but it just got several months down the line … it became very nitty gritty where … like, for example, basic necessities of working. I remember one time I was working as a chambermaid at one bed and breakfast because I didn’t come up with good grades from school. One sister thought she was doing me good by asking the sheik if it was okay for me in Islam to work and the sheik totally said, ‘No’. The more intrusive they became in my life um I thought okay I need to start putting barriers and boundaries whether that was physically at home. I’d stop them from coming to my home because it was not only overtaking my home. It was overtaking my family as well … So dropping my kids off and having to go to events or organize events – that was stressful. Trying to hold down a job, a normal day job, it’s a very, very schizo-situation because you are having to be normal yet you’re working for a cause.53

A former member of a right-wing terrorist group similarly explained the role conflict his friend experienced as a father. He elaborates:

His second child was born and he was working, like, 60 h [hours] a week to feed his family and keeps the light on, and so he would just come around less and less and I would be like, Well, I you know, where have you been? How come you haven’t been drinking in the hall with us and you know, rolling up Racial Loyalty newspapers?” And he would say, “Well, you know, I got to work. I’ve, I’ve been working all the time. We got, we got a new house, the rent, blah, blah, blah.54

10 M. B. ALTIER ET AL.

When considered through the lens of Rusbult’s investment model, these accounts highlight how role strain and role conflict may increase dissatisfaction with involve- ment. In terrorist groups where individuals have more autonomy in their initial role assignment or the ability to shift roles if they are unhappy or where the terrorist group seriously considers the match between an individual’s skill set, characteristics and com- peting demands and the role(s) to which one is assigned, we would expect a lower inci- dence of role-related dissatisfaction among members. Two examples of individuals who recounted role-related dissatisfaction are illustra-

tive. The first is Ahmed Kathrada, a former member of the ANC and the ANC’s mili- tant wing, Umkhonto we Sizwe (MK). Kathrada explained his decision to leave MK and assume a more political role in the movement:

At no time did I object in principle to the decision to move to an armed struggle, and I have never harboured the slightest regret about MK’s formation. I did, however, terminate membership of the regional command during the first half of 1962, after discussing certain reservations with some of my senior comrades. Not all men are born to be soldiers, and I realized at a fairly early stage that my aptitude lay in the political rather than the military field.55

The second is Aukai Collins, who grew increasingly disillusioned with the menial, behind-the-scenes roles he was assigned. Collins recounts how he expressed his growing disillusionment to the group’s leadership, who did not take measures to mitigate his role-related dissatisfaction:

I talked with Umar about the situation over the next few days. We’d benefitted from the camp, but I wanted to call it quits if I had to sit around and wait instead of making jihad somewhere … I wanted to make jihad on the front lines … One day everything came to a boil. Umar and I sat down with the commanders and had a very candid discussion. I told them that I didn’t care for the way they were treating me and that I hadn’t come all the way from America to sit and watch other mujahideen train and fight in Tajikistan. I tried to get a straight answer out of them about when they would send me to fight in Tajikistan … They didn’t conceal their anger either and told me to hit the road if I didn’t like the way things were going … getting past the NWFP tribal police [border checkpoint guards] with Harakat-ul Jihad’s blessings had been dangerous and difficult. If they washed their hands of me and I left now—friendless, penniless and foreign—getting out of Afghanistan alive would be next to impossible. I should have apologized to the commanders and continued to stick it out in camp, but I was young and hotheaded and told them good-bye.56

These accounts suggest disengagement may be more likely in groups that do not allow for the expression of role-related dissatisfaction and accommodate such dissatis- faction accordingly. Further role-related dissatisfaction, for some, may be sufficient to trigger disengagement. Indeed, a quantitative analysis of the frequency and importance of certain push/pull

factors in explaining individual, voluntary disengagement decisions highlights its importance. For our analysis, we considered not only whether one stated a particular push/pull factor played a role (large or small) in the decision to voluntarily disengage, but we also coded whether individuals were experiencing certain push/pull factors likely associated with role-related dissatisfaction during their involvement and then examined whether the experience of these push/pull factors are associated with an increase in the probability of individual, voluntary disengagement. As individuals are likely to experi- ence multiple push and pull factors associated with disengagement, we coded all rele- vant push/pulls experienced and self-reported to play a role in one’s disengagement.

STUDIES IN CONFLICT & TERRORISM 11

The results (Table 3) suggest individuals who voluntarily disengaged were more than eight times as likely to experience disillusionment with their day-to-day tasks (i.e. role strain) than those in our comparison group. A v2 test reveals this difference is statistic- ally significant. Further, 43% reported disillusionment with one’s day-to-day tasks played a role in their decision to leave the group (22.5% a large role, and 20.9% a small role). To examine the potential influence of role conflict in explaining disengagement decisions, we examined the frequency and importance of the following pull factors, which highlight a likely incompatibility between one’s terrorist role and a more conven- tional social role: desire to seek employment/education, dedicate more time to family, too hard to balance with family life, dedicate more time to existing job/educational pro- gram, desire to marry and desire to have children.57 Although all with the exception of “too hard to balance with family life” were more likely to be experienced among those who voluntarily disengaged, the only pull factors that were statistically significant were the desire to dedicate more time to one’s family and the desire to have children. Although less frequently cited as playing a role in disengagement decisions that disillu- sionment with one’s day-to-day tasks, it is crucial to note that the desire to marry and have a family, the demands of family, the desire to seek conventional employment or education, or work/school demands were all reported as pivotal in explaining the disen- gagement decisions of some individuals who chose to walk away from terrorism. Despite these findings, it is important to reiterate that individuals may experience

role strain or conflict and remain because of the satisfaction they obtain from other aspects of involvement or because they have invested heavily in the group and/or pos- sess few alternatives. Cecilia Bobrovskaya, a former member of the Revolutionary Cells in Russia, for example, discussed how she was experiencing a conflict between her roles as a terrorist and a midwife. Due to her increasing involvement in the Revolutionary Cells, she had little time for outside employment and, as such, could often not afford food. Bobrovskaya explains:

During those times of intense hunger I would be in utter despair. I would rather die than give up Party work and daily intercourse with the comrades; yet if I looked for employment it would mean that I would have to give up my Party work and become occupied with something that I neither knew nor liked. I hated midwifery.58

Table 3. Relationship between disengagement & role strain and role conflict. Comparison

group (n¼ 54) Disengagement (n¼ 49)

Push/pull factor Experienced

(%) Experienced

(%) Any

role (%) Large role (%)

Small role (%)

Role strain Disillusionment with day-to-day tasks��� 5.6 42.9 42.9 22.5 20.4 Role conflict Desire to seek employment/education 16.7 26.5 14.3 8.2 6.1 Dedicate more time to family� 13.0 26.5 12.2 6.1 6.1 Too hard to balance with family life 24.1 20.4 14.3 4.1 10.2 Dedicate more time to existing job/

educational program 5.6 14.3 10.2 4.1 6.1

Desire to have children�� 0.0 10.2 8.2 4.1 4.1 Desire to marry 5.6 8.2 4.1 2.0 2.0 �p< .10.��p< .05.���p value for v2 test for experiencing push/pull factor is less than .01.

12 M. B. ALTIER ET AL.

Just as satisfaction in other areas can mitigate the effects of role conflict and role strain on one’s decision to exit, so too can satisfaction in one’s role offset the effects of growing dissatisfaction in other arenas. Eamon Collins described how the pride he obtained from his new role helped counteract the doubts that he was experiencing about certain operations and the PIRA’s ideology:

My promotion to the security unit had given me an initial glow of satisfaction. I tried to allow the swelling of my ego to overshadow the horror of what I was involved in. My “promotion” became a crutch to support myself in the collapse of my belief in the armed struggle. But so advanced was the process of mental deterioration that I could no longer quite believe my own desperate attempt at self-deception. I knew I was staring at my own defeat. Yet my vanity gave me the strength to continue acting out for a little while longer a role that was no longer mine.59

Roles, Sunk Costs and the Quality of Available Alternatives

It is possible one’s role influences not only the satisfaction component of Rusbult’s investment model, but also the investments and alternatives (H3). Although this rela- tionship was not commonly expressed explicitly in the autobiographical data, Leyden, a former neo-Nazi skinhead, noted that he considered his role and what part it would play in his ability to leave:

Since Skins don’t take kindly to people who just get in their way, I couldn’t imagine how they would feel against anyone in power who broke the oath and left the movement. I remember thinking, “Don’t bother trying to get out or you’ll get seriously messed up” [emphasis added].60

For Leyden, his high-ranking role meant that any alternative to involvement would entail the threat of continual persecution by the group, thereby limiting the appeal of leaving. Leyden’s account is corroborated in interviews with three other former members of

right-wing terrorist groups. These individuals mentioned that it was much easier for those “less involved” in the group or on the periphery to leave without facing retribu- tion.61 One elaborates:

You’ve been in it longer, you know you’ve got more at stake … you’ve lived the ideology, you’ve lived the lifestyle for so long, you have people that look up to you, you know, you have people that listen to you, that follow what you say, and you know, and when they lose what they consider a leader, that’s not a good thing.62

Another interviewee, who was a member of a nationalist group, also recounted:

Not common at all, but it wasn’t impossible to leave… I think it all depends how valuable they were to the organization. Then, it would be easier to leave or just drift away or just not be involved. So it wasn’t that difficult… So I suppose it all depends on your seniority in the organization and what you know about that organization. If you’re in a crucial position. Say you’re somebody who’s in charge of weapons. That’s not somebody who’s going to walk out of the organization, you know.63

In addition to leaders, it seems individuals in violent operational roles may incur greater sunk costs, possess fewer alternatives and have a harder time disengaging. One inter- viewee, who operated as a bomb-maker and bomb-planter in a nationalist terrorist group, noted his role shaped his ability to leave in two somewhat different ways. First, because he served in a direct violent operational role, the possible alternatives were limited:

STUDIES IN CONFLICT & TERRORISM 13

It was possible for very, very prominent people to work their way out, you know, but you see it was easier if they were more involved in the command structure and hadn’t been in the physical jobs front… like they hadn’t been in the actual operations. They might have ordered the operations where they were more morally and legally responsible than the young person they sent out to do the job, but for somebody like me who was the front- line [terrorist], who made, planted or carried the [explosive] … For me the options to get out without necessarily going over to the enemy, you were kind of constrained by that.64

Second, the deep investments he made in the group made it more difficult to leave because they signaled to the leadership his commitment to the cause and his capacity to continue:

You see I wouldn’t have been in the category of somebody who needed to leave. Here they’d be looking at somebody who was quite active, fully gritty and determined, and they definitely would have been… it would’ve been quite difficult to pop your head above the part and say I have a fundamental disagreement with the aims and philosophy and ideology of the organization…There were no rules or rubrics for how people were going to be facilitated in that area [i.e. leaving the terrorist group because they no longer agreed with its underlying ideology].65

To further test these claims we ran a series of logistic regressions, which compare the effect of one’s role – specifically leadership and violent, operational roles – at the time of disengagement on the probability an individual chose to leave terrorism. Due to the sample size, there is not enough variation to use all 15 role categories. We therefore col- lapse our role categories into leadership (executive leader & low-level leader), violent operators (violent operator, bomb-maker, operational planner) and support/logistics (trainer, intelligence officer, financier, recruiter, security officer, courier, publicity offi- cer, political officer, materials acquisition, general member). We control for whether the individual was experiencing certain push/pull factors likely to influence the satisfaction component of Rusbult’s investment model in order to help isolate the influence of role- related sunk costs and alternatives. One could argue leaders and violent operators are generally more ideologically committed and thus, less likely to disengage. We therefore control for the level of ideological commitment at the time of disengagement. Finally, following Abadie et al.,66 we cluster the standard errors by terrorist group, as Altier et al. note individuals from certain groups are more likely write an autobiography and thus be “sampled.”67 We also rerun the analysis clustering standard errors by individual to address dependencies across engagement events in our sample by individual. These and other robustness checks68 are in the Online Appendix. The results (Table 4) suggest that holding a leadership role or a violent role decreases

the probability of voluntary disengagement from terrorism (Models 1 and 2), while holding a support role increases the probability. These results in conjunction with our interview data suggest that leaders and those more directly involved in violent operations are less likely to walk away from terrorism and that this may be driven, in part, by the greater investments they make in the group and/or the fewer alternatives available. To calculate the substantive effects of one’s role, we simulated the expected probabil-

ity of disengagement by setting all control variables to their means and allowing one’s role to vary.69 We find holding a leadership role reduces the probability of voluntary disengagement by 19 percentage points (Model 1), holding a violent role reduces the probability by 24 percentage points (Model 2), and holding a support role increases the probability by 26 percentage points (Model 3).

14 M. B. ALTIER ET AL.

Roles and the Likely Presence of Push/Pull Factors

Qualitative evidence from the autobiographies also supports hypothesis H4 that individ- uals in certain roles are more likely to experience certain push/pull factors associated with leaving. Tass Saada, a former member of the Palestinian Liberation Organization, for example, described the toll his role as a sniper took:

Back in camp, I realized with surprise how mentally exhausted I was. Killing someone is neither easy nor insignificant. You think right away about what you’ve done. Especially if you planned the act ahead of time. It’s different being in hot battle: there, you kill on instinct to avoid getting killed yourself in the next two seconds. But when you’ve laid an ambush and deliberately set a trap to destroy another human being, it takes its toll (2008:36).

By contrast, Eamon Collins had little difficulty coping with the effects of carrying out attacks. He mentioned:

Yet I was satisfied: satisfied that I had acted as an IRA man, as a volunteer, and I was prepared to move on to the next operation, the next hi [sic], the next kill – for that was what it was all about (1997: 28).

Instead, Collins seemed to experience burnout as a result of the careful planning that attacks required:

It was always like this: I would try to get on with the normal things of life but I would spend my time continually thinking about violent operations. Going for a drink, playing

Table 4. Effect of roles on probability of disengagement. Model 1 Model 2 Model 3

Leadership role �1.18� (0.61)

Violent role �1.49� (0.87)

Support/logistical role 1.70��� (0.53)

Disillusionment with strategy �0.50 (0.73)

�0.42 (0.63)

�0.42 (0.66)

Disillusionment with leadership 1.06� (0.63)

1.25�� (0.61)

1.06� (0.60)

Disillusionment with members 1.18 (0.74)

0.98 (0.63)

1.19� (0.68)

Disillusionment with day to day tasks 2.58�� (1.00)

2.75��� (1.05)

2.74�� (1.08)

Burnout 0.22 (0.74)

0.85 (0.72)

0.66 (0.78)

Difficulty coping with attacks �1.01 (1.09)

�1.39 (0.93)

�0.76 (1.19)

Ideological commitment (none)

Some �0.76 (1.14)

�0.17 (0.98)

�0.16 (1.05)

A lot �0.91 (1.00)

�0.83 (0.77)

�0.35 (0.92)

Constant 0.04 (1.06)

�0.23 (0.88)

�1.89� (1.13)

n 98 98 98 Pseudo R2 0.2854 0.2905 0.3298

Logistic regression results. Robust standard errors clustered at group level.�p< .10, ��p< .05, ���p< .01.

STUDIES IN CONFLICT & TERRORISM 15

with the kids, visiting the relatives, preparing Sunday lunch, going for a drive in the country, reading the newspapers – I liked to do all these normal things which ordinary people did, but the enjoyment of such normality was impossible. I could pretend, but I could never be relaxed or absorbed by normal activities, because thoughts of the IRA were always at the front of my mind, nagging me, putting me on edge, torturing me with images of what might go wrong… (1997:157).

Collins further suggested his role not only caused burnout, but socially isolated him:

I had come to ditch almost everything and everybody not connected in some way to the IRA. It had become my whole life and I was beginning to ask myself what sort of life I had. I went through the motions of enjoying myself, but how could I live happily when I spent most of my time in the company of people whose business was death? And I was one of them. Always looking for people to kill, finding people to do the killing, constantly exposing myself to danger, more and more danger. There was no respite. (1997:157–158)

To systematically test these claims, we ran a series of logistic regressions to examine whether one’s role70 increased the probability of experiencing a particular push/pull fac- tor for disengagement, regardless of how the engagement event ended. We controlled for ideological commitment and, when appropriate, marital status at the time of disen- gagement. The latter may influence the likelihood of certain push/pull factors such as family demands or the desire to marry and have children. Following Abadie et al.,71 we cluster the standard errors by terrorist group, as individuals from certain groups are more likely write an autobiography and thus be “sampled.”72 Results clustered by indi- vidual are in the Online Appendix.73 Table 5 lists the results for leaders, Table 6 for vio- lent operators and Table 7 for support/logistical roles. Our findings suggest, on average, those in leadership roles are more likely to report

experiencing disillusionment with members, difficulty living a clandestine lifestyle and regret for their role in attacks (Table 5). Those in leadership roles are also less likely to report maintaining ties to nonextremists. Violent operators are more likely to experi- ence physiological distress as a result of involvement and burnout (Table 6). Those in support or logistical roles are less likely to experience difficulty living a clandestine life- style, difficulty coping with attacks and regret for their role in attacks (Table 7). One’s role does not have a statistically significant effect on the experience of other push fac- tors including disillusionment with day-to-day tasks or the strategy or actions of the group, and pull factors including the desire to marry and have children, to dedicate more time to one’s family, to seek employment or education or the offer of amnesty or material incentives. Our results further show that even when controlling for one’s role, a deep ideological commitment reduces the probability an individual will report the experience of most push/pull factors except the desire to dedicate more time to one’s job or education, the desire to marry and offers of amnesty. While these findings pro- vide insight into which push/pull factors individuals in certain roles more commonly experience, it is important to reiterate that whether they precipitate disengagement depends on a myriad of other factors including what motivates the individual’s involve- ment, the available alternatives and any investments incurred. In order to calculate substantive effects, we simulated the expected probability of

experiencing a particular push/pull factor by setting all control variables to their means and allowing one’s role to vary.74 We find the largest effect of being a leader is on the likelihood of experiencing regret for one’s role in attacks, which increases by 14

16 M. B. ALTIER ET AL.

Ta bl e 5.

Th e ef fe ct

of ho

ld in g a le ad er sh ip

ro le

on th e pr ob

ab ili ty

of ex pe rie nc in g pu

sh /p ul lf ac to rs

fo r di se ng

ag em

en t.

D is ill u-

si on

m en t

w /d ay -t o-

da y ta sk s

D is ill u-

si on

m en t

w /s tr at eg y

D is ill u-

si on

m en t

w /le

ad er s

D is ill u-

si on

m en t

w /m

em be rs

D iff ic ul ty

w /c la nd

es tin

e lif es ty le

D iff ic ul ty

co pi ng

w / at ta ck s

Re gr et

ro le

in at ta ck s

Ph ys io lo gi ca l

di st re ss

Ps yc ho

lo gi ca l

di st re ss

Bu rn ou

t

Ti es

to no

ne xt re m is t

in di vi du

al s

D iff ic ul ty

ba la nc in g

fa m ily

lif e

D ed ic at e

m or e

tim e to

fa m ily

Se ek

jo b or

ed uc at io n

D ed ic at e

m or e tim

e to

jo b/

ed uc at io n

D es ire

M ar ry

D es ire

ch ild re n

Am ne st y

Fi na nc ia l/

m at er ia l

in ce nt iv es

Le ad er

�0 .3 3

(0 .5 5)

�0 .1 8

(0 .4 6)

�0 .2 4

(0 .4 4)

0. 60 �

(0 .3 6)

0. 83 ��

(0 .4 0)

0. 71

(0 .5 2)

1. 08 ��

(0 .5 2)

�0 .2 6

(0 .6 4)

�0 .2 6

(0 .5 3)

�0 .2 3

(0 .4 1)

�0 .6 6�

(0 .3 8)

0. 55

(0 .3 8)

0. 72

(0 .6 0)

0. 05

(0 .3 6)

0. 09

(0 .5 6)

�0 .0 5

(0 .9 0)

�0 .4 6

(1 .1 0)

0. 11

(0 .8 1)

�0 .5 7

(0 .9 2)

Id eo lo g.

Co m m it.

(N on e)

So m e

�0 .4 3

(0 .6 3)

�0 .8 7

(0 .9 7)

�1 .0 4

(0 .7 2)

�2 .1 8� ��

(0 .7 6)

�1 .7 3� �

(0 .8 3)

�1 .9 2� ��

(0 .6 1)

�1 .4 6� ��

(0 .5 4)

�1 .3 0� �

(0 .6 2)

�1 .1 2�

(0 .6 3)

�0 .7 3

(0 .6 9)

�0 .6 2

(0 .8 0)

�1 .2 0�

(0 .6 2)

�2 .1 1� ��

(0 .7 2)

�1 .2 7�

(0 .6 7)

�0 .3 3

(1 .0 3)

�1 .5 5

(1 .3 2)

�1 .9 8

(1 .3 3)

1. 45

(1 .0 3)

�0 .7 9

(0 .5 7)

A Lo t

�1 .8 8� ��

(0 .4 9)

�2 .8 4� ��

(0 .9 4)

�2 .2 2� �

(0 .6 3)

�3 .3 6� ��

(0 .6 4)

�1 .3 8� �

(0 .6 9)

�2 .8 3� ��

(0 .6 9)

�2 .2 7� ��

(0 .5 8)

�1 .9 1� �

(0 .7 7)

�1 .1 8� �

(0 .5 7)

�1 .2 5� �

(0 .5 8)

�1 .1 6�

(0 .7 2)

�1 .4 9� �

(0 .6 1)

�2 .0 1� ��

(0 .5 1)

�1 .4 6� ��

(0 .5 1)

�0 .0 4

(0 .9 0)

�1 .1 0

(0 .7 6)

�2 .0 4� ��

(0 .7 2)

�0 .0 3

(1 .1 5)

�2 .7 9� ��

(0 .9 0)

M ar rie d

1. 05 ��

(0 .5 0)

1. 51 ��

� (0 .4 8)

�0 .6 2

(0 .5 7)

�0 .2 8

(0 .6 1)

Co ns ta nt

�0 .1 8

(0 .4 8)

1. 85 ��

(0 .9 0)

1. 00

(0 .6 3)

1. 62 ��

� (0 .6 2)

�1 .9 9

(0 .7 9)

�0 .3 3

(0 .5 5)

�0 .2 5

(0 .5 6)

�0 .8 3

(0 .7 7)

�0 .2 0

(0 .6 0)

�0 .2 1

(0 .6 0)

2. 05 ��

� (0 .7 2)

�0 .6 3

(0 .5 1)

�0 .8 4�

(0 .5 1)

�0 .3 1

(0 .4 1)

�2 .2 8

(0 .7 2)

�1 .6 0� �

(0 .7 5)

�1 .2 2�

(0 .7 3)

�3 .0 3� ��

(1 .1 1)

�0 .9 9

(0 .5 8)

n 15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 7

15 7

Ps eu do

R2 0. 10 99

0. 18 83

0. 11 68

0. 19 57

0. 07 59

0. 19 68

0. 13 72

0. 08 64

0. 03 02

0. 03 86

0. 04 42

0. 08 06

0. 14 87

0. 04 98

0. 00 21

0. 00 21

0. 00 35

0. 05 79

0. 17 51

Lo gi st ic re gr es si on

s. Ro bu

st st an da rd

er ro rs

cl us te re d at

gr ou

p le ve l.

� p < .1 0,

�� p < .0 5,

�� � p

< .0 1.

STUDIES IN CONFLICT & TERRORISM 17

Ta bl e 6.

Th e ef fe ct

of ho

ld in g a vi ol en t op

er at io na lr ol e on

th e pr ob

ab ili ty

of ex pe rie nc in g pu

sh /p ul lf ac to rs

fo r di se ng

ag em

en t.

D is ill u-

si on

m en t

w /d ay -t o-

da y ta sk s

D is ill u-

si on

m en t

w /s tr at eg y

D is ill u-

si on

m en t

w /le

ad er s

D is ill u-

si on

m en t

w /m

em be rs

D iff ic ul ty

w /c la nd

es tin

e lif es ty le

D iff ic ul ty

co pi ng

w /a tt ac ks

Re gr et

ro le

in at ta ck s

Ph ys io lo gi ca l

D is tr es s

Ps yc h.

D is tr es s

Bu rn ou

t

Ti es

to no

ne xt re m is t

in di vi du

al s

D iff ic ul ty

ba la nc in g

fa m ily

lif e

D ed ic at e

m or e

tim e to

fa m ily

Se ek

jo b or

ed uc at io n

D ed ic at e

m or e tim

e to

jo b/

ed uc at io n

D es ire

M ar ry

D es ire

ch ild re n

Am ne st y

Fi na nc ia l/

m at er ia l

in ce nt iv es

Vi ol en t

op er at or

�0 .0 7

(0 .5 3)

0. 30

(0 .6 4)

0. 48

(0 .6 0)

�0 .5 3

(0 .4 6)

0. 22

(0 .3 3)

1. 09

(0 .9 0)

0. 16

(0 .6 2)

1. 26 ��

(0 .6 0)

0. 78

(0 .5 3)

0. 91 ��

(0 .4 0)

0. 56

(0 .5 9)

�0 .6 2

(0 .6 6)

�1 .0 7

(0 .7 3)

0. 33

(0 .5 1)

�0 .2 6

(0 .9 6)

0. 38

(0 .7 6)

0. 77

(0 .8 3)

�0 .9 8

(1 .2 1)

O m itt ed

Id eo lo g.

Co m m it.

(N on e)

So m e

�0 .3 4

(0 .5 8)

�0 .8 7

(0 .9 4)

�1 .0 6

(0 .6 8)

�2 .2 2� ��

(0 .7 0)

�1 .9 5� �

(0 .7 9)

�2 .3 2� ��

(0 .6 8)

�1 .6 9� ��

(0 .5 5)

�1 .5 3� �

(0 .6 2)

�1 .2 0� �

(0 .5 8)

�0 .8 4

(0 .7 0)

�0 .5 2

(0 .7 4)

�1 .2 7� �

(0 .6 3)

�2 .2 2� ��

(0 .7 5)

�1 .3 4�

(0 .6 9)

�0 .3 2

(1 .0 1)

�1 .6 1

(1 .2 8)

�2 .0 4

(1 .2 6)

1. 55

(1 .0 0)

�0 .4 7

(0 .5 9)

A lo t

�1 .8 8� ��

(0 .4 8)

�2 .8 7� ��

(0 .9 4)

�2 .2 8� ��

(0 .6 0)

�3 .2 5� ��

(0 .6 8)

�1 .3 2�

(0 .6 8)

�2 .9 2� ��

(0 .6 5)

�2 .1 0� ��

(0 .5 4)

�2 .1 2� ��

(0 .7 4)

�1 .2 8� �

(0 .5 4)

�1 .3 7� �

(0 .6 3)

�1 .5 23 �

(0 .6 7)

�1 .3 7� �

(0 .5 9)

�1 .8 4� ��

(0 .5 3)

�1 .4 9� ��

(0 .5 1)

�. 01

(0 .8 8)

�1 .1 4

(0 .7 6)

�2 .1 7� ��

(0 .7 4)

0. 03

(1 .1 6)

�2 .7 3� ��

(0 .9 1)

M ar rie d

0. 98 ��

(0 .4 8)

1. 40 ��

(0 .4 9)

�0 .5 8

(0 .5 9)

�0 .1 3

(0 .7 2)

Co ns ta nt

�0 .2 8

(0 .4 6)

1. 77 ��

(0 .8 1)

0. 88

(0 .5 9)

1. 85 ��

� (0 .6 0)

�0 .7 2

(0 .5 5)

�0 .1 9

(0 .6 6)

0. 08

(0 .5 7)

�1 .0 6

(0 .8 5)

�0 .3 6

(0 .6 5)

�0 .3 8

(0 .6 1)

1. 75 ��

� (0 .6 5)

�0 .3 7

(0 .5 5)

�0 .4 7

(0 .5 8)

�0 .3 2

(0 .4 2)

�2 .2 3� ��

(0 .7 2)

�1 .6 7� �

(0 .7 0)

�1 .5 0� �

(0 .6 8)

�2 .9 3� ��

(1 .0 1)

�1 .0 3� �

(0 .4 9)

n 15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 7

12 9

Ps eu do

R2 0. 10 72

0. 18 92

0. 12 03

0. 19 13

0. 05 45

0. 20 66

0. 10 34

0. 12 03

0. 04 72

0. 05 85

0. 03 52

0. 07 72

0. 14 71

0. 05 22

0. 00 31

0. 05 16

0. 11 84

0. 07 04

0. 17 74

Lo gi st ic re gr es si on

s. Ro bu

st st an da rd

er ro rs

cl us te re d at

gr ou

p le ve l.

� p < .1 0,

�� p<

.0 5,

�� � p < .0 1.

18 M. B. ALTIER ET AL.

Ta bl e 7.

Th e ef fe ct

of ho

ld in g a su pp

or t/ lo gi st ic al ro le

on th e pr ob

ab ili ty

of ex pe rie nc in g pu

sh /p ul lf ac to rs fo r di se ng

ag em

en t.

D is ill u-

si on

m en t

w / da y- to -

da y ta sk s

D is ill u-

si on

m en t

w / st ra te gy

D is ill u-

si on

m en t

w / le ad er s

D is ill u-

si on

m en t

w / m em

be rs

D iff ic ul ty

liv in g

cl an de st in e

lif es ty le

D iff ic ul ty

co pi ng

w /a tt ac ks

Re gr et

ro le

in at ta ck s

Ph ys io lo gi ca l

D is tr es s

Ps yc h.

D is tr es s

Bu rn ou

t

Ti es

to no

ne xt re m is t

in di vi du

al s

D iff ic ul ty

ba la nc in g

fa m ily

lif e

D ed ic at e

M or e

tim e to

fa m ily

Se ek

jo b or

ed uc at io n

D ed ic at e

m or e tim

e to

jo b/

ed uc at io n

D es ire

M ar ry

D es ire

ch ild re n

Am ne st y

Fi na nc ia l/

m at er ia l

in ce nt iv e

Su pp

or t/

Lo gi st ic

Ro le

0. 29

(0 .5 5)

�0 .0 3

(0 .3 6)

�0 .1 1

(0 .3 5)

�0 .1 9

(0 .3 2)

�0 .9 6� �

(0 .4 1)

�1 .4 2� �

(0 .7 0)

�1 .1 0� �

(0 .5 3)

�0 .6 7

(0 .5 5)

�0 .2 8

(0 .4 8)

�0 .4 2

(0 .3 0)

0. 28

(0 .4 2)

�0 .1 9

(0 .3 3)

�0 .2 3

(0 .5 2)

�0 .2 5

(0 .3 7)

0. 06

(0 .4 9)

�0 .1 9

(0 .6 3)

�0 .1 0

(0 .7 8)

0. 42

(0 .7 2)

1. 40

(0 .8 5)

Id eo lo gi ca l

Co m m itm

en t

(N on e)

So m e

�0 .3 8

(0 .6 2)

�0 .8 2

(0 .9 4)

�0 .9 8

(0 .6 8)

�2 .2 7� ��

(0 .6 8)

�1 .8 9� �

(0 .7 8)

�2 .1 3� ��

(0 .6 5)

�1 .6 7� ��

(0 .5 5)

�1 .2 1� �

(0 .5 5)

�1 .0 4

(0 .6 4)

�0 .6 5

(0 .6 7)

�0 .4 7

(0 .7 7)

�1 .3 1� �

(0 .6 4)

�2 .2 6� ��

(0 .7 2)

�1 .2 6�

(0 .6 6)

�0 .3 6

(1 .0 1)

�1 .5 3

(1 .2 2)

�1 .8 8

(1 .1 9)

1. 39

(0 .9 7)

�0 .8 1

(0 .5 7)

A Lo t

�1 .8 6� ��

(0 .5 0)

�2 .8 5� ��

(0 .9 2)

�2 .2 4� ��

(0 .6 0)

�3 .3 1� ��

(0 .6 6)

�1 .4 7� �

(0 .7 1)

�3 .1 2� ��

(0 .7 3)

�2 .3 5� ��

(0 .6 1)

�2 .0 3� ��

(0 .7 6)

�1 .2 4� �

(0 .5 7)

�1 .3 2� �

(0 .6 1)

�1 .1 5�

(0 .6 8)

�1 .4 7� �

(0 .6 1)

�1 .9 7� ��

(0 .5 2)

�1 .5 0� ��

(0 .5 1)

�. 02

(0 .8 9)

�1 .1 2

(0 .7 8)

�2 .0 7� ��

(0 .7 7)

0. 03

(1 .1 3)

�2 .7 1� ��

(0 .9 1)

M ar rie d

1. 05 ��

(0 .4 7)

1. 50 ��

� (0 .4 7)

�0 .6 0

(0 .5 2)

�0 .2 2

(0 .6 7)

Co ns ta nt

�0 .4 6

(0 .5 2)

1. 81 ��

(0 .8 1)

0. 98 �

(0 .5 5)

1. 91 ��

� (0 .5 9)

�0 .1 8

(0 .5 9)

0. 71

(0 .7 4)

0. 74

(0 .7 2)

�0 .5 6

(0 .7 9)

�0 .1 3

(0 .6 6)

�0 .0 5

(0 .6 8)

1. 63 ��

� (0 .6 1)

�0 .3 4

(0 .5 9)

�0 .4 6

(0 .6 8)

�0 .1 5

(0 .4 8)

�2 .2 9� ��

(0 .7 5)

�1 .5 2

(0 .8 3)

�1 .3 2� �

(0 .6 1)

�3 .2 5� ��

(1 .0 1)

�2 .0 9� �

(0 .7 7)

n 15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 8

15 7

15 7

Ps eu do

R2 0. 11 00

0. 18 75

0. 11 55

0. 18 75

0. 08 26

0. 23 47

0. 14 00

0. 09 81

0. 03 50

0. 16 91

0. 03 16

0. 07 18

0. 13 45

0. 05 20

0. 00 20

0. 05 00

0. 10 84

0. 06 26

0. 21 26

Lo gi st ic re gr es si on

s. Ro bu

st st an da rd

er ro rs

cl us te re d at

gr ou

p le ve l.

� p < .1 0,

�� p < .0 5,

�� � p

< .0 1.

STUDIES IN CONFLICT & TERRORISM 19

percentage points. The probability of reporting disillusionment with other members increases by 10 percentage points and difficulty living a clandestine lifestyle by 9 per- centage points. Leaders also experience a 13 percentage point reduction in the probabil- ity they possess ties to nonextremists. For those in violent roles, the probability of reporting burnout increases by 15 percentage points and the probability of experiencing physiological distress increases by 10 percentage points. For those in support roles, the probability of reporting regret for one’s role in attacks decreases by 14 percentage points and the probability of reporting difficulty coping with attacks and difficulty living a clandestine lifestyle decrease by 11 percentage points each.

Conclusion

Our analysis provides the first systematic evidence of a potential link between the role(s) individuals hold within a terrorist group and their likelihood of voluntary disen- gagement. We theorize the ways in which one’s role(s) influence the key components of Rusbult and colleagues’ investment model and thus, the probability of exit. We find evi- dence to support the claim that individuals experiencing role conflict and role strain are more likely to disengage. Evidence also suggests those in leadership and violent, oper- ational roles are less likely to leave even when controlling for factors likely to influence one’s satisfaction suggesting that these individuals may incur greater sunk costs and possess fewer alternatives. Those in support roles, conversely, are more likely to exit. Finally, we show even when controlling for ideological commitment certain roles are more commonly associated with the experience of particular push/pull factors hypothe- sized to drive disengagement decisions. These findings indicate that greater attention to a terrorist’s role(s) may lend import-

ant insights for encouraging or facilitating their disengagement. For example, those experiencing role conflict or role strain may be more susceptible to leaving. As such, terrorist groups which allow some autonomy in role selection or those that factor in an individual’s knowledge, skills and abilities when assigning roles may be less prone to shedding members. Individuals in leadership or violent, operational positions may be less likely to disengage unless there are policies in place that provide them with adequate alternatives outside the group (e.g. amnesty, protection from reprisal). Further, knowing one’s role in the group may provide additional insight into the push/pull fac- tors for disengagement he or she may be experiencing. Our findings indicate that lead- ers are more likely to experience regret over their role in terrorist attacks and a harder time living underground. Violent operators are more likely to experience burnout and physiological distress. Such information can help in the development of more tailored interventions especially when knowledge of the individual’s motivations and circumstan- ces are lacking. Finally, a deeper understanding of the link between terrorist roles and terrorist disengagement can also inform policies for countering violent extremism. Counter-narratives that emphasize the mismatch between the roles one expects to hold in a terrorist group and the reality of the roles that one will hold or the dissatisfaction associated with certain roles may help deter involvement. Although each individual’s characteristics and experiences in a terrorist group, like any social group, are unique and must be considered, individuals assigned to certain tasks and functions may

20 M. B. ALTIER ET AL.

encounter many of the same challenges and these challenges may influence the likeli- hood of disengagement and the reasons for disengagement in a relatively systematic way. Future research should continue to explore the relationships outlined here.

Notes

1. Eamon Collins and Mick McGovern, Killing Rage (UK: Granta, 1998). 2. See, Tore Bjørgo and John Horgan, eds., Leaving Terrorism Behind: Individual and

Collective Disengagement. (London: Routledge, 2008); John Horgan, Walking Away from Terrorism (Abington, UK and New York, NY: Routledge, 2009); Fernando Reinares, “Exit from Terrorism: A Qualitative Empirical Study on Disengagement and Deradicalization among Members of ETA,” Terrorism and Political Violence 23, no. 5 (2011); Rogelio Alonso, “Why do Terrorists Stop? Analyzing Why ETA Members Abandon or Continue with Terrorism,” Studies in Conflict and Terrorism 34, no. 9 (2011); Mary Beth Altier, Christian Thoroughgood, and John Horgan, “Turning away from Terrorism; Lessons from Psychology, Sociology, and Criminology,” Journal of Peace Research 51, no. 5 (2014); Kate Barrelle, “Pro-Integration: Disengagement from and Life After Extremism,” Behavioral Sciences or Terrorism and Political Aggression 7, no. 2 (2015); Neil Ferguson, Mark Burgess, and Ian Hollywood, “Leaving Violence Behind: Disengaging from Politically Motivated Violence in Northern Ireland,” Political Psychology 36, no. 2 (2015); Mary Beth Altier, Emma Leonard Boyle, Neil Shortland, and John Horgan, “Why They Leave: An Analysis of Terrorist Disengagement Events from 87 Autobiographical Accounts,” Security Studies 26, no. 2 (2017).

3. See Arie Perlinger, Gabriel Koehler-Derrick, and Ami Pedhazur, “The Gap between Participation and Violence: Why We Need to Disaggregate Terrorist ‘Profiles’,” International Studies Quarterly 60, no. 2 (2016); John Horgan, Neil Shortland, Suzette Abbasciano, and Shaun Walsh, “Actions Speak Louder than Words: A Behavioral Analysis of 183 Individuals Convicted of Terrorist Offenses in the United States from 1995 to 2012,” Journal of Forensic Sciences 61, no. 5 (2016); Robin Simcox and Emily Dyer, Al Qaeda in the United States: A Complete Analysis of Terrorism Offenses (London, UK: The Henry Jackson Society, 2013); Paul Gill and John Horgan, “Who were the Volunteers? The Shifting Sociological and Operational Profile of 1240 Provisional Irish Republican Members,” Terrorism and Political Violence 25, no. 3 (2013).

4. John Monahan, “The Individual Risk Assessment of Terrorism,” Psychology, Public Policy, & Law 18, no. 2 (2012).

5. Altier et al., “Why They Leave”. 6. We define terrorist activity according to the U.S. Department of Defense as “the calculated

use of unlawful violence or threat of unlawful violence to inculcate fear, intended to coerce or to intimidate governments or societies in the pursuit of goals that are generally political, religious, or ideological.” This definition is consistent with those usually employed in the academic literature by Bruce Hoffman, Inside Terrorism (New York: Columbia University Press, 1998) and Alex Schmid and Albert Jongman, Political Terrorism: A New Guide to Actors, Authors, Concepts, Databases, Theories, and Literature (New Brunswick, NJ: Transaction Books, 1988).

7. James Krantz and Marc Maltz, “A Framework for Consulting to Organizational Role,” Consulting Psychology Journal: Practice and Research 49, no. 2

8. Ibid, 137. 9. Some have argued terrorist share certain characteristics including insanity or psychopathy

(e.g. Friedrich Hacker, Crusaders, Criminals, Crazies: Terror and Terrorism in our Time (New York, NY: Norton, 1976); K. I. Pearce, “Police Negotiations,” Canadian Psychiatric Association Journal 22, no. 4 (1977) or psychopathic tendencies (e.g. H. H. A. Cooper, “What is a Terrorist: A Psychological Perspective,” Legal Medical Quarterly 1, no. 1 (1977), and H. H. A. Cooper, “Psychopath as Terrorist,” Legal Medical Quarterly 2, no. 4 (1978).

STUDIES IN CONFLICT & TERRORISM 21

Jeff Victoroff, “The Mind of a Terrorist: A Review and Critique of Psychological Approaches,” Journal of Conflict Resolution 49, no. 1 (2005) notes that psychological research on terrorism is “largely flawed” and based on unscientific methods.

10. Jeffrey Ross, “A model of the Psychological Causes of Oppositional Political Terrorism,” Peace and Conflict 2, no. 2 (1996).

11. Perlinger et al., “The gap between participation and violence”. 12. Gill and Horgan, “Who were the Volunteers”. 13. Simcox and Dyer, “Al Qaeda in the United States”. 14. Altier et al., “Turning Away”. 15. Altier et al., “Turning Away”. 16. Ibid. 17. D. Farrell and C. E. Rusbult, “Exchange Variables as Predictors of Job Satisfaction, Job

Commitment, and Turnover: the Impact of Rewards, Costs, Alternatives, and Investments,” Organizational Behavior and Human Performance 28 (1981); C. E. Rusbult, “Commitment and Satisfaction in Romantic Associations: A Test of the Investment Model,” Journal of Experimental Social Psychology 16 (1980); C. E. Rusbult, “A Longitudinal Study Test of the Investment Model: The Development and Deterioration of Satisfaction and Commitment in Heterosexual Involvements,” Journal of Personality and Social Psychology 45 (1983); C. E. Rusbult and D. Farrell, “A Longitudinal Test of the Investment Model: The Impact on Job Satisfaction, Job Commitment, and Turnover of Variations in Rewards, Costs, Alternatives, and Investments.” Journal of Applied Psychology 68 (1983).

18. Benjmain Le and Christopher Agnew, “Commitment and Its Theorized Determinants: A Meta-Analysis of the Investment Model,” Personal Relationships 10, no. 1 (2003); Karen van Dam, “Employee Attitudes Toward Job Changes: An application and extension of Rusbult and Farrell’s Investment Model,” Journal of Occupational and Organizational Psychology 78, no. 2 (2005); Caryl Rusbult, Dan Farrell, Glenn Rogers, and Arch Mainous III, “Impact of Exchange Variables on Exit, Voice, Loyalty, and Neglect: An Integrative Model of Responses to Declining Job Satisfaction,” Academy of Management Journal 31, no. 3 (1988); Meni Koslowsky and Avraham Kluger, “Commitment to Participation in Musical Activities: An Extension and Application of the Investment Model,” Journal of Applied Psychology 16, no. 9 (1986).

19. Yuan-Huei Lin and Caryl Rusbult, “Commitment to Dating Relationships and Cross-Sex Friendships in America and China,” Journal of Social and Personal Relationships 12, no. 1 (1995); Caryl Rusbult and John Martz, “Remaining in an Abusive Relationship: An Investment Model Analysis of Nonvoluntary Commitment,” Personality and Social Psychology Bulletin 21, no. 6 (1995); Caryl Rusbult, Dennis Johnson, and Gregory Morrow, “Predicting Satisfaction and Commitment in Adult Romantic Involvements: An Assessment of the Generalizability of the Investment Model,” Social Psychology Quarterly 49, no. 1 (1986); Thomas Raedeke, “Is Athlete Burnout more than Just Stress? A Sport Commitment Perspective,” Journal of Sport and Exercise Psychology 19, no. 4 (1997); Paul Carpenter and Richard Coleman, “A Longitudinal Test of Elite Youth Cricketers’ Commitment,” International Journal of Sport Psychology 29, no. 3 (1998)

20. Le and Agnew, “Commitment and”. 21. Altier et al., “Turning away”. 22. W.J. Goode, “A Theory of Role Strain,” American Sociological Review 25, no. 4 (1960); S. R.

Marks, “Multiple Roles and Role Strain: Some Notes on Human Energy, Time and Commitment,” American Sociological Review 42, no. 6 (1977); J. D. Snoek, “Role Strain in Diversified Role Sets,” American Journal of Sociology 71, no. 4 (1966).

23. J. W. Getzels and E. G. Guba, “Role, Role Conflict, and Effectiveness: An Empirical Study,” American Sociological Review 19 (1954); J. T. Gullahorn, “Measuring Role Conflict,” American Journal of Sociology 61, no. 4 (1956).

24. Orsini, for example, documents role conflict and role strain producing dissatisfaction in interviews with far left and far right terrorists. Alessandro Orsini, Anatomy of the Red Brigades (Cornell University Press, 2011), 72, 266-67.

22 M. B. ALTIER ET AL.

25. R. L. Kahn, D. M. Wolfe, R. P. Quinn, and J. D. Snoek, Organizational Stress. (New York, NY: Wiley, 1964); J. R. Rizzo, R. J. House, and S. I. Lirtzman, “Role Conflict and Ambiguity in Complex Organizations,” Administrative Science Quarterly 15, no. 2 (1970); R. T. Keller, “Role Conflict and Ambiguity: Correlates with Job Satisfaction and Values.” Personnel Psychology 28 (1975).

26. Jacob N. Shapiro, The Terrorist’s Dilemma: Managing Violent Covert Organizations (Princeton, NJ: Princeton University Press, 2013); Jacob N. Shapiro and David A. Siegel, “Moral Hazard, Discipline and the Management of Terrorism Organizations,” World Politics 64, no. 1 (2012); Mary Beth Altier, John Horgan, and Christian Thoroughgood, “In their Own Words? Methodological Considerations in the Analysis of Terrorist Autobiographies,” Journal of Strategic Security 5 (2012); Bonnie Cordes, “When Terrorists do the Talking: Reflections on the Terrorist Literature,” Journal of Strategic Studies 10, no. 4 (1987).

27. Michael Kenney, Stephen Coulthart, and Dominick Wright, “Structure and Performance in a Violent Extremist Network: The Small World Solution,” Journal of Conflict Resolution 61, no. 10 (2016); Barrelle, “Pro-Integration”; Anne Speckhard and Ahmet S. Yayla, “Eyewitness Accounts from Recent Defectors from Islamic State: Why They Joined, What They Saw, Why They Quit,” Perspectives on Terrorism 9, no. 6 (2015); Horgan, Walking Away.

28. Altier et al., “In their Own Words?”. 29. Cordes, “When Terrorists do the Talking”; Rachel Gamson, Jessica Gottesman, Nicholas

Milan, and Sitara Weerasuriya, Cues to Catching Deception in Interviews (College Park, MD: START, 2012).

30. Cordes, “When Terrorists do the Talking”; Altier et al., “In their Own Words?” 31. Two individuals in the sample, Michael “Bommi” Baumann and Martin McGartland, each

wrote two distinct autobiographical accounts. 32. We define terrorist activity as “the calculated use of unlawful violence or threat of unlawful

violence to inculcate fear, intended to coerce or to intimidate governments or societies in the pursuit of goals that are generally political, religious, or ideological” (U.S. Department of Defense).

33. See the Online Appendix for a description of how the dataset was created. 34. Omar Cabezas, Fire From the Mountain: The Making of a Sandinista (Crown Publishing

Group, 1985), 13. 35. Sean O’Callaghan, The Informer (London, UK: Corgi, 1998), 198. 36. Ibid, 28. 37. Maria McGuire, To take arms: A year in the Provisional IRA (London, UK: Macmillan,

1973), 135. 38. Table 1A in the Supplementary Appendix lists the inclusion criteria for each role. 39. For our analysis of roles, we analyzed text on three questions: (1) What positions or roles

did the individual hold within the terrorist organization or its affiliates? (2) Was the individual satisfied (happy) with his/her position(s) or role(s) within the organization and the day-to-day tasks that each entailed? and (3) Did the individual change positions or roles within the organization? If so, please include information as to when and why these changes occurred.

40. Leila Khalid, My People Shall live: The Autobiography of a Revolutionary (Hodder and Stoughton, 1973), 173–75.

41. N. Ascherson, Memoirs of an Italian Terrorist (Basic Books, 2003), 94. 42. Abu Iyad with E. Rouleau, My Home, My Land: A Narrative of the Palestinian Struggle

(New York, NY: Times Books, 1981), 54–56. 43. Ibid, 95. 44. Abdul Salam Zaeef, A. S. Van Linschoten, and F. Kuehn. My life with the Taliban (New

York, NY: Columbia University Press, 2010), 22. 45. Shane Paul O’Doherty, The Volunteer: A Former IRA Man’s True Story. (Strategic Book

Publishing, 2011), 53–54. 46. Ibid, 53–54. 47. Raymond Gilmour, Dead Ground: Infiltrating the IRA (Warner Books, 1999), 85.

STUDIES IN CONFLICT & TERRORISM 23

48. Omar Nasiri, Inside the Global Jihad (Jacana Media, 2006), 36. 49. Tass Saada and D. Merrill, Once an Arafat Man: The True Story of how a PLO Sniper found

a New Life (Tyndale House Publishing, 2008), 34. 50. Ibid, 35. 51. Ben Turok, Nothing but the Truth: Behind the ANC’s Struggle Politics (Jonathan Ball

Publishers, 2003), 70–73. 52. David Hamilton, A Cause Worth Living For: The Journey of David “Packie” Hamilton

(Highland Books, 1997), 51. 53. Interview 8/20/13. 54. Interview 10/1/12. 55. Ahmed Kathrada, Letters from Robben Island: A selection of Ahmed Kathrada’s prison

correspondence, 1964–1989 (Zebra, 2000), 141–42. 56. Collins, My Jihad, 36–37. 57. Summary statistics for these and variables included in the remainder of our analyses are

located in Table 2A in the Online Appendix. 58. Cecilia Bobrovskaya, Twenty Years in Underground Russia: Memoirs of a Rank-and-File

Bolshevik (International Publishers, 1934), 30–31. 59. Collins and McGovern, Killing Rage, 233. 60. T. J. Leyden and M. B. Cook, Skinhead Confessions: From Hate and Hope (Sweetwater

Books, 2008), 104. 61. Interviews 10/2/12, 11/16/12 and 8/4/13. 62. Interview 8/4/13. 63. Interview 8/20/13. 64. Interview 9/11/12. 65. Ibid. 66. Alberto Abadie, Susan Athey, Guido W. Imbens, and Jeffrey Wooldridge, “When Should

You Adjust Standard Errors for Clustering?” NBER Working Paper #24003 (2017). 67. Altier et al., “Why They Leave”. 68. Following Altier et al. (2017), Table 3A in the online appendix excludes Aukai Collins, who

is a potential outlier given the number of times he disengaged (n¼ 11) and re-engaged (n¼ 10). Table 4A excludes individuals involved as spies. Table 5A excludes Collins and spies. Table 6A includes age at disengagement, which significantly reduces the sample size and why it is not included here. Table 7A presents the results with robust standard errors clustered by individual. When clustering by individual or excluding spies, the effect of holding a violent role is close to, but no longer statistically significant at the .10 level. When excluding Collins or Collins and spies or including age at disengagement, holding a leadership or violent role are no longer statistically significant, but holding a support role is. Given that the coefficients remain relatively the same and in the same direction, we believe this is a reflection of the reduction in our sample size. In every specification, holding a support role increases the likelihood of disengagement.

69. We used Gary King, Michael Tomz, and Jason Wittenberg’s CLARIFY macro. Gary King, Michael Tomz, and Jason Wittenberg, “Making the Most of Statistical Analyses: Improving Interpretation and Presentation,” American Journal of Political Science 44, no. 2 (April 2000).

70. Because we lack temporal data on when the individual experienced the push/pull during engagement and what role s/he was in at the time, we choose to use the most salient role at disengagement. While this is an imperfect measure, it is the best available.

71. Alberto Abadie, Susan Athey, Guido W. Imbens, and Jeffrey Wooldridge, “When Should You Adjust Standard Errors for Clustering?” NBER Working Paper #24003 (2017).

72. Altier et al., “Why They Leave”. 73. When clustering by individual, difficulty living a clandestine lifestyle and ties to non-radical

individuals are no longer statistically significant for leaders though difficulty living a clandestine lifestyle is very close (p< .101); the findings for violent operators and supporters remain relatively the same.

24 M. B. ALTIER ET AL.

74. We used Gary King, Michael Tomz, and Jason Wittenberg’s CLARIFY macro. Gary King, Michael Tomz, and Jason Wittenberg, “Making the Most of Statistical Analyses: Improving Interpretation and Presentation,” American Journal of Political Science 44, no. 2 (April 2000).

75. Adapted from Altier et al., “Why They Leave”.

Acknowledgments

The authors thank the editor, anonymous reviewers, Kurt Braddock, Neil Shortland and Joe Young for helpful comments. We also thank Christian Thoroughgood for research assistance.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

The research from which this article is drawn was sponsored by the U.S. Department of Homeland Security’s Science and Technology Directorate and coordinated through the U.K. Home Office. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Department of Homeland Security or the Home Office.

ORCID

Mary Beth Altier http://orcid.org/0000-0001-9939-6022

STUDIES IN CONFLICT & TERRORISM 25