Analyzing person‐exposure patterns in lone‐actor terrorism

• Date: 01 May 2020
• Author: Caitlin Clemmow,Noémie Bouhana,Paul Gill
• Published date: 01 May 2020
• DOI: http://doi.org/10.1111/1745-9133.12466

DOI: 10.1111/1745-9133.12466

ORIGINAL ARTICLE

COUNTERTERRORISM

Analyzing person-exposure patterns in lone-actor

terrorism

Implications for threat assessment and intelligence gathering

Caitlin Clemmow Noémie Bouhana Paul Gill

University College London

Correspondence

CaitlinClemmow, Department of Security &

Crime Science, UniversityCollege London,

35Tavistock Square, Kings Cross, London

WC1H9EZ, U.K.

Email:c.clemmow.17@ucl.ac.uk

Thisstudy received funding from The Euro-

peanResearch Council (ERC) under the

EuropeanUnion’s Horizon 2020 research

andinnovation programme (Grant 75883).

Seehttps://www.grievance-erc.com. This

material is partly based on worksuppor ted

bya European Commission 7th Framework

ProgrammeGrant [608354 (PRIME) FP7-

SEC-2013-1].More information is available

fromhttp://www.fp7-prime.eu/.

Research Summary: The lone-actor terrorist population

can be extremely heterogeneous and difficult to detect.

Intelligence is vital to countering this threat. We devise

a typology of person–exposure patterns (PEPs) that

could serve as a framework for intelligence gathering

and threat assessment. We use cluster analysis and a risk

analysis framework (RAF) to identify relations among

three components: propensity, situation, and network.

The results of the analysis reveal four PEPs: solitary,

susceptible, situational, and selection. The solitary PEP

lacks common indicators of a propensity to pursue terrorist

action. The susceptible PEP reveals cognitive suscepti-

bility, manifesting as mental illness, to be a key factor in

the emergence of a terrorist propensity. The situational

PEP demonstrates how situational stressors may act as

warnings of acceleration toward violent action. Lastly, the

selection PEP demonstrates higher frequencies of leakage

and antecedent violent behaviors.

Policy Implications: Our findings have two key policy

implications. First, given the multifinality of terrorism

risk indicators, we suggest a move toward a structured-

professional judgement approach to the risk analysis of

lone-actor terrorists. Second, we present the PEP typology

as a framework for intelligence gathering. Existing

frameworks are predominantly focused on mobilization

Criminology & Public Policy. 2020;19:451–482. wileyonlinelibrary.com/journal/capp © 2019 American Society of Criminology 451

452 CLEMMOW ET AL.

indicators. We suggestexpanding data collection to include

propensity and situational indicators, as operationalized

here, and using the PEP typology to inform decisions about

the emergence of the motivation to commit an attack. To

do so, it is necessary to pursue a multiagency approach to

intelligence gathering.

KEYWORDS

intelligence-led policing, lone actor, risk analysis framework, terrorism,

threat assessment

Post 9/11, counterterrorism policing has evolved toward an intelligence-led model of policing

(ILP). Threat assessment, defined as “the application of the collection and analysis of information

related to crime,” is a key principle of ILP (Capellan & Lewandowski, 2019, p.17). The lone-actor

terrorist population can be extremely heterogeneous and a challenge for law enforcement to detect,

and so a framework for guiding the threat assessment of these offenders may serve as a beneficial

tool for ILP. Typologies can be a useful way to conceptualize complex, heterogeneous offending

populations and crime events. The results of various studies of a range of crimes provide evidence

for the usefulness of typologies in terms of increasing arrest rates (Fox & Farrington, 2015),

decreasing recidivism (Carbajosa, Catalá-Miñana, Lila, & Gracia, 2017), and predicting violence

risk (Mohandie, Meloy, McGowan, & Williams, 2006). We argue that disaggregating an empirical

typology of lone-actor terrorism can have a similar practical utility, specifically, as a framework

for guiding the threat assessment of these offenders within the context of an ILP approach to

counterterrorism.

Detecting and disrupting lone-actor terrorist attacks is a central focus of policing, globally.

Intelligence gathering, defined as “a process that involves the collection and transformation of

data into knowledge and finally actionable and useable recommendations for courses of actions”

(Kebbell & Porter, 2012, p. 213, see also Ratcliffe, 2008), is vital. Threat assessment involves

making intelligence- and evidence-based decisions about the allocation of limited resources and

the responses of appropriate agencies (Tusikov & Fahlman, 2009). Hence, a great emphasis is

placed in ILP on acquiring, evidencing, and actioning appropriate intelligence. Kebell and Porter

(2012) described the need for a framework to guide intelligence gathering. Specifically, they iden-

tified the need for clarity regarding what intelligence to collect, as well how to operationalize

intelligence in decision-making. The authors conceptualized this as a “risk-based, intelligence-led

approach to counter-terrorism” (Kebell & Porter, 2012, p. 224). A typology in which the relations

between common risk indicators are articulated and tied to causal mechanisms may provide the

necessary guidance for a risk-based, intelligence-led approach to countering the lone-actor terrorist

threat.

Disaggregated analysis became a research agenda for terrorism scholars, in part, after the pub-

lication of a unique insight into far-right homicide in the United States. Gruenewald, Chermak,

and Freilich (2013b) analyzed data from the Extremist Crime Database, looking at differences in

characteristics such as race, mental illness, weapon-use, and victim–offender relationships. They

found differences between right-wing–inspired homicide offenders and group-based offenders.

These results had a significant impact on the field by tangibly demonstrating the heterogeneity of

CLEMMOW ET AL.453

extremist offending populations. The need to disaggregate terrorist populations has been called for

and demonstrated empirically (Gill, 2015; Gill & Corner, 2013; Horgan, Gill, Bouhana, Silver, &

Corner, 2016; Horgan, Shortland, & Abbasciano, 2018; Horgan, Shortland, Abbasciano, & Walsh,

2016; Perliger, Koehler-Derrick, & Pedahzur, 2016).

Typologies of a range of equally complex offending populations demonstrate the utility of a more

rigorous, quantitative approach to disaggregating these populations. For example, Gruenewald and

Kelley (2014) disaggregated a population of anti-LGBT homicides by characteristics of victim selec-

tion and conceptualized two types; predatory and responsive homicide (see also Gruenewald, 2012).

They compared differences across several characteristics of the offense process, including offender,

victim, situational, attack, and aftermath characteristics. The focus on situational characteristics of bias

homicide stems predominantly from Tomsen’s (2002) work on anti-homosexual homicide in Australia,

which has particular relevance here. Often, explanations of lone-actor terrorist behavior are focused

predominantly on individual-level characteristics and not on the role of the individual’scontext or situ-

ation. We argue it is necessary fora typology of lone-actor ter rorism to articulate the relations between

a person and his or her environment to inform judgments of risk, which are conceptualized here as

person-exposure patterns (PEPs).

In current typologies of lone-actor terrorists, offenders are predominantly differentiated along a sin-

gle behavioral dimension, suchas social connectedness (Pantucci, 2011), r isk aversion(Phillips & Pohl,

2012), or ideology (Simon, 2013). This approach is evident across work with other types of terrorists

and includes factors such as recruitment (Brzica, 2017), the variables that limit an organization’sactiv-

ities (Ganor, 2008), suicidal motivation (Lankford, 2014), personality disorders (Miller, 2006), target

group (Merari, 1978), and individual differences such as motivation (Hacker & Hacker, 1976) and per-

sonality differences (Strentz, 1988). Terrorism can be extremely heterogeneous, and so the function

of these typologies is often to “bring order to this chaos” (Merari, 1978, p. 331). For the purposes

of guiding intelligence gathering, however, disaggregating populations along a single dimension may

inhibit the detection of patterns of behavior that exist naturally within the data.

Moreover, these dimensions are mainly constructs identified by the researcher, which are then

retroactively applied to a set of cases. Deductive typologies, such as these, are often based on detailed

analyses of case studies, drawn from established theory, or from inferences based on the existing lit-

erature. These typologies serve an important function in terrorism research; however, their external

validity may be limited given the use of predominately qualitative methods. For the purposes of threat

assessment, we argue that more rigorous, quantitative methodologies, and an inductive approach to

typology development, may be more appropriate.

To address many of these issues, Horgan et al. (2018) used multidimensional scaling methods to

disaggregate terrorist involvement and proposed a behavioral typology of violent extremist offenders.

Drawing on methods from investigative psychology, this approach offers an alternative to deductive

typologies in that types are not organized around predefined dimensions. We aim to disaggregate

lone-actor terrorists in a similar way. We seek to detect multidimensional subtypes that are embedded

within the data, while conserving the benefits of theory-informed approaches through interpretative

guidance from an explicit analytical framework. Such an ambition has been called for elsewhere

(Borum, Fein, & Vossekuil, 2012), but remains untested empirically.

In forensic psychology and criminology, techniques such as cluster analysis have been used to

derive empirical typologies of a range of criminal behaviors. These typologies are multidimensional,

polythetic, and span the offence trajectory. Researchers have drawn from theories of crime to propose

causal mechanisms across multiple offending components at multiple levels of analysis of the

crime-commission process. In one notable design, the results of a series of cluster analyses informed

by developmental theories of crime, were used to deduce a range of typologies of sexual offenders