Measuring radicalisation: risk assessment conceptualisations and practice in England and Wales

ABSTRACT

Individual ‘radicalisation’- extremism- and terrorism-related risk assessment tools have become increasingly central instruments of counter-terrorism. The scholarship on such tools, however, is still its infancy, and remains concentrated on methodological issues and on identifying the ‘best’ indicator list for carrying out assessments. This article takes a different approach, and examines England and Wales’ Extremism Risk Guidance (ERG22+) and Vulnerability Assessment Framework (VAF), concentrating on the 22 risk indicators that both tools share, and their current uses in counter-terrorism. The article explores the conceptualisations of ‘radicalisation’ that emerges from the tools’ indicators and from their actual uses at two different ‘ends’ of England and Wales’ counter-terrorism system; to assess sentenced terrorism offenders in prison, and to assess non-criminal individuals referred over concerns over their possible ‘radicalisation’. The article hence offers both a conceptual clarification of the ideas of ‘radicalisation’ underpinning counter-terrorism policies in England and Wales, and reflections on the operational utility of the present use of the ERG and VAF indicators. While not rejecting the possible value of specialised terrorism-related individual risk assessment tools, the article finds that the conceptualisations underpinning the tools’ indicators and their use make their present counter-terrorism roles questionable.

KEYWORDS:

• Radicalisation
• counter-terrorism
• risk assessment
• UK
• Prevent

Introduction

As countries strive to identify successful ways to prevent terrorist attacks, tools for individual risk assessment have become increasingly central instruments of counter-terrorism. At present, ‘radicalisation’-, extremism- and terrorism-related risk assessment tools are proliferating both geographically and in terms of the risks they intend to capture. In different countries and forums, tools for assessing the risk associated with specific individuals are in use or being discussed for offenders in prisons, those identified as possibly becoming ‘radicalised’, persons suspected of planning ‘lone actor’ attacks, and, recently, returning Western Foreign Terrorist Fighters (see Lloyd & Dean, 2015; Meloy, 2015; Meloy & Gill, 2016; Pressman, 2009; RAN, 2017).

This article concentrates on the Extremism Risk Guidance, ERG22+ (ERG for short), as well as the Vulnerability Assessment Framework (VAF). The VAF was developed based on the ERG and the two tools assess individuals based on the same 22 risk indicators (see below). Both the ERG and the VAF were created by and are in wide use in the UK, a country often regarded as a forerunner on the field of European counter-terrorism. The centrality of the tools to the work of the UK to combat terrorism can be illustrated by their current uses at each of the two ‘extremes’ of the counter-terrorism system in England and Wales; the ERG is used to assess sentenced terrorism offenders in prison and the VAF is used to assess those considered vulnerable to possibly becoming (but not necessarily yet being) ‘radicalised’.

Although the ERG relies on the terminology of ‘extremism’ rather than ‘radicalisation’, both tools, as will be explained, have as central aims to capture individuals’ risk of radicalisation as well as expected outcomes of this process. The 22 indicators both tools are made up of and their present counter-terrorism uses are thus well-suited to illuminate the conceptualisations of radicalisation at the heart of UK counter-terrorism policies more broadly. Investigating the ideas of radicalisation underpinning the tools and their use will also aid a clear-headed assessment of their operational roles as measures of counter-terrorism. Moreover, looking at the actual uses of the tools’ indicators at two different parts of England and Wales’ counter-terrorism system will assist any further assessment of their instrumental utility as well as the potential of individual radicalisation-oriented risk assessment tools as such.

Before turning to the background and uses of the ERG and VAF and their conceptualisations of radicalisation, it must be stated that the very concept of ‘radicalisation’ is contested. Ever since it became part of conventional policy parlance in the mid-to-late 2000s, scholars have criticised it for being misleading, vague, and based on ill-founded assumptions, with some calling for discarding it altogether (see Neumann, 2013). ‘Radicalisation’ used in the context of counter-terrorism, according to many, seems to suggest that there is a clear and causal link between radical ideological, political or religious thought and violent terrorist acts; but no such link exists, these claim (Borum, 2011; Horgan, 2014; Kundnani, 2012; Richards, 2015; Schmid, 2013; Sedgwick, 2010 provide various perspectives). Moreover, it is argued that in order to understand what causes terrorism and how to fight it, it is crucial to distinguish between (cognitive) ‘radicalisation’ and (behavioural) ‘mobilisation’, ‘extremism’, or ‘engagement’ on the one hand, and ‘deradicalisation’ and ‘disengagement’ on the other (see e.g. Bjørgo & Horgan, 2009).

In the lexicon of European policy makers and practitioners, however, ‘radicalisation’ is a common reference. Authorities often define the term as a process culminating in the endorsement or use of terrorist violence (e.g. Regjeringen, 2014). The UK’s are among these, and define radicalisation as ‘the process by which a person comes to support terrorism and extremist ideologies associated with terrorist groups’ (HM Government, 2015b, p. 21). The endurance of the term seems explained in part by its perceived usefulness to counter-terrorism practitioners, to whom it appears to sum up a phenomenon that is both observable and in need of a name.

Within UK counter-terrorism, one idea sometimes associated with the label of ‘radicalisation’, of there being a continuum between radical thought and terrorist action, has been behind the repeatedly announced intention to criminalise ‘non-violent extremism’. These legislative attempts have recently been put off indefinitely (through the formation of an investigative commission) due to problems with defining ‘extremism’ in a legally enforceable way (Blackman, 2017; Townsend, 2017; UK Parliament, 2016). Nonetheless, ‘extremism’ remains the pronounced topic and target of the ERG; the official UK definition of ‘radicalisation’ is not referenced as part of that tool. Meanwhile, vulnerability to ‘radicalisation’ is explicitly what the VAF seeks to identify.

Law enforcement, intelligence, probation services and forensic psychiatry are among the professions that have long made use of various individual risk assessment tools. Techniques for conducting assessments of this kind have ranged from large-scale data collection and algorithms to one-to-one personal clinical interaction (Leese, 2014; McCulloch & Wilson, 2016; Singh, Bjørkly, & Fazel, 2016). The development of tools especially dedicated to identify terrorism-related risk is a more recent phenomenon, emerging especially from the mid-2000s onwards (Monahan, 2012, 2015).

Despite the last years’ policy trend of increasingly embracing various risk assessment tools as inherent to counter-terrorism, the scholarship on terrorism-related risk assessment is in its infancy. The topic remains predominantly covered by the fields of forensic psychology and psychiatry as well as the quantitatively oriented parts of terrorism studies. It is typically instrumental in its orientation, aiming above all to improve tools’ predictive value. Besides scrutinising the methodology behind developing such tools in order to improve it, articulating the ‘best’ model or list of indicators by which individuals can be assessed have been a core aim of this emerging body of thought (see Gill, 2016; Sarma, 2017).

This article has a different methodological anchoring and diverging ambitions. Inspired by the parts of Quentin Skinner’s work of intellectual history that concerns the uses and legitimation of ideas in political discourse, it identifies the conceptualisations of radicalisation underpinning the ERG and VAF indicators and their prominent places in UK counter-terrorism (Skinner, 2002). In light of how the tools are used in practice, the article indicates the implications of these conceptualisations for the tools’ counter-terrorism roles. Besides analysing the ERG and the VAF in particular, this endeavour will lay the ground for further investigations – be these operationally, methodologically, normatively or otherwise oriented – of terrorism-related risk assessment tools in general.

The article finds that the ERG and VAF risk indicators – perhaps unsurprisingly as these are risk assessment tools developed by psychologists – suggest a conceptualisation of radicalisation as a psychological phenomenon associated with significant risk. This conceptualisation holds the promise of being both more effective and less potentially challenging to democratic norms and human rights than an approach circling in on the religious or ideological content of individuals’ thinking. However, the present use of the tools’ indicators, especially in the VAF context, seems based on a ‘strong’ methodological individualism that contradicts recent scholarship on radicalisation (Hodgson, 2007; Udéhn, 2001; Udéhn, 2002). In particular, while the ERG and VAF indicators are not intended to capture the full political and societal context of an individual’s radicalisation, England and Wales’ actual counter-terrorism reliance on these indicators make their limited explicit incorporation of such context problematic.

The ERG and the VAF: background and uses

Like on other fields of European counter-terrorism, the UK was early in initiating the creation of a specialised tool for assessing terrorism-related risk, with a key impetus behind the ERG occurring in 2006. The UK Terrorist Act (TACT) of that year included several non-violent terrorist crimes along with violently terrorist ones. The majority of terrorist offences did now not involve the use of physical violence against other people, but were crimes like encouragement of terrorism (including indirect encouragement); disseminating terrorist publications; preparing terrorist acts; giving and receiving training for terrorism; and attendance at a place used for terrorist training (CPS, 2017). With the 2006 legislation’s expansion of terrorist crimes, some of the forensic psychologists working to assess offender risk in UK prisons found that the generic and largely violence-oriented tools they had used thus far had become unable to capture the range of risks now of interest (Lloyd & Dean, 2015, p. 40).

At the same time, scholarly research on risk assessment was questioning the overall usefulness of generic risk assessment tools to the fields of terrorism, extremism and radicalisation. One point this scholarship made was that the personal backgrounds of those involved in terrorism appeared distinct from those involved in (other) criminal violence; another suggestion was that the psychological factors that disposed some people to engage in terrorist activity seemed to differ from those disposing some people to act violently in other ways. In consequence, specialised tools needed to account for these differences if they were to effectively capture terrorism-related risk (Monahan, 2011; Pressman, 2009, p. 16–17).

In 2008, the UK National Offender Management Service (NOMS), gave two experienced forensic psychologists the lead role in developing a tool for assessing risk among imprisoned terrorism offenders. The forensic psychologists behind what came to be the ERG were motivated by a need to prioritise between cases in their work in UK prisons (HMPPS 2017; Lloyd & Dean, 2015, p. 40–41). Tasked by the prison and probation authorities, they hoped to find a way to better identify persons of particular concern and select the right ones for targeted attention and intervention. The ERG was formulated based on their clinical forensic work in UK prisons and their reading of theoretical literature. While developed based on experiences with chiefly Al Qaeda-inspired offenders, the ERG is meant to be useful on individuals with any ideological reference, of any gender, and on lone actors and group actors alike.

ERG22+ with indicators (Lloyd & Dean, 2015, p. 46).

Since 2011, the ERG has been used throughout prisons in England and Wales. Specifically, England and Wales use the tool to assess all terrorism offenders sentenced under TACT, as well as a shorter version of the ERG to assess prisoners sentenced for non-terrorist crimes if someone has a serious concern over their ‘possible involvement or interest in extremist groups, causes, or ideas’ (Lloyd & Dean, 2015, p. 40). ERG assessments are used to contribute to determine, among other things, prisoners’ security categorisation, bail, release, and targeted intervention programmes (see also Rose, 2012). An ERG is initially carried out upon the entry of a prisoner into the correctional service and at that point to retrospectively identify the factors that contributed to the original offence. Subsequently, the tool is used with regular intervals to measure the development of a person and to pinpoint areas of concern and further intervention.

The ERG was formulated for use by either a trained forensic psychologist or an experienced probation officer, and anyone using the tool must have gone through a two-day training dedicated to using the ERG (Lloyd & Dean, 2015, p. 48). Usually only one assessor carries out each assessment, but sometimes more than one are involved (Lloyd & Dean, 2015, p. 49). Although cooperation with the assessed is sought out, this is not always granted and is no requirement for an assessment. The now renamed NOMS – the HM Prison and Probation Service (HMPPS) is currently finalising a review and update of the ERG for continued use.

In 2012, the ERG became the basis of another tool, named the Vulnerability Assessment Framework (VAF). The VAF was developed without the HMPPS, but with the involvement of one of the two creators of the ERG, and came to consist of the same 22 indicators. The VAF was in 2012 applied at the ‘opposite end’ of England and Wales’ counter-terrorism system, for use by local authorities with regard to the possible risk of ‘radicalisation’ of an individual who had not committed a crime, but about whom someone had reported a concern under ‘Prevent’. The Prevent programme is one component of the UK’s counter-terrorism strategy CONTEST, and one part of Prevent is called Channel. Prevent places a statutory duty upon certain authorities, including those of education and health, to report worries over someone’s possible radicalisation. It also allows anyone to refer anyone else over radicalisation-related concerns (HM Government, 2011, 2015a; Home Office, 2012; see also Heath-Kelly, 2017a).

Some of the referrals to Prevent will be transferred to Channel, where an overall assessment by local authorities will include carrying out a VAF as well as collecting information from the referred person’s family and wider network. Although the framework is one of vulnerability rather than risk, the stated aims of the VAF are to ‘identify individuals at risk of being drawn into terrorism; assess the nature and extent of that risk; and develop the most appropriate support plan for the individuals concerned.’ (Home Office, 2012, p. 2, my emphases, also Krahmann, 2011). Following a Channel assessment, a panel will meet to decide on appropriate interventions. While Channel is meant for ‘vulnerable children and adults of any faith, ethnicity or background’ (HM Government, 2015a, p. 5), more than half of those referred to Prevent are under 20 years old, and close to one third are below 15 (Home Office, 2017).

The ERG is not used in Scotland, which instead uses the Canadian-originated Violent Extremism Risk Assessment, version 2 (VERA2), which was formulated based on theory rather than clinical practice. Elsewhere in the UK, the Royal College of Psychiatrists has criticised the ERG over lack of transparency regarding its source basis and for a possibly flawed methodology involved in its development (RCP, 2016). Scholarly research on specialised terrorism-related risk assessment has also highlighted methodological issues such as low base rates, validation and reliability problems as well as the diversity of the risks tools intend to capture and the differences in the profiles of people who become involved in terrorism (see e.g. Brown & Cox, 2011; Garrick, 2002; Gill, 2016; Herzog-Evans, 2018; Sarma, 2017; Scarcella, Page, & Furtado, 2016). Activists, scholars, and parts of the media have expressed alarm over alleged problems with the VAF in particular as well as with the UK’s approach to counter-terrorism more broadly (Qureshi, 2016; The Guardian 2016). The Prevent part of CONTEST has been particularly strongly criticised, with the most vocal claiming that programme has created a climate of suspicion and fear and resulted in stigmatisation of Muslims (Halliday & Dodd, 2015; Sian, 2017).

Radicalisation snapshots

The ERG and VAF indicators were articulated by psychologists as part of their forensic clinical work with a particular prison population in order to guide psychologically-informed interventions; it should thus come as no surprise that the conceptualisation of radicalisation emerging from these indicators is a psychological one. The 22 risk indicators are oriented towards mapping an individual’s ways of thinking, reasoning and perceiving. Less emphasis is placed on the content of this thinking or the context in which it occurs.

Before unpacking these aspects of the ERG and VAF indicators it is worth comparing the tools’ shared structure and conceptual orientation towards risk with that of the tool most widely used and acknowledged within the general field of psychological risk assessment and management, the HCR20. Such a comparison is interesting not because the HCR20 would necessarily be a better tool with which to assess extremism-related risk, but because the HCR20 is the most established tool with which to assess risk connected with selected individuals; its conceptual structure can serve to put that of the ERG and VAF into relief. What is more, given the prominence of the HCR20 on the field of risk assessment, the ERG creators must not only have been familiar with the HCR20 when developing the ERG. They must also have found the HCR20 wanting when seeking to capture risk associated with post-2006 terrorism offenders in England and Wales. And they must have formulated the ERG indicators and structure in part as a response to the HCR20’s perceived deficiencies when addressing such risk.

The HCR20 was developed as a structured clinical assessment tool for use by mental health professionals to assess individuals’ risk of committing violent acts, especially violence deemed to be likely, immediate and serious (Logan, 2014; Webster, Douglas, & Hart, 1997). It is divided into three parts with one Historical, one Clinical, and one Risk management scale with in total 20 risk indicators. Also the ERG and the VAF have three parts; engagement, intent and capability (see above). Broadly speaking, the three parts of the HCR20 are directed towards mapping an individual’s past, present and future. The historical scale consists of in total ten risk factors, including young age at first violent incident, employment, and relationship problems. The HCR20 clinical scale has five indicators associated with the individual’s present characteristics such as impulsivity, negative attitudes, lack of insight and major mental illness. The risk management scale also has five factors, each assessing the individual’s likely future, such as the feasibility of future plans and likely exposure to destabilisers.

By contrast, none of the three parts making up the ERG and the VAF alike is as a whole primarily looking towards a person’s past. Although assessors carrying out ERG assessments in prisons are encouraged to be aware of aspects of a person’s history, only three of the tools’ shared 22 indicators raises this directly. These regard criminal history, mental health, and transitional periods. As mentioned, the suggestion that terrorist offenders’ histories can differ from that of those committing other forms of crime was one important part of the background to developing separate risk assessment tools for extremism/terrorism/radicalisation in the first place. But accepting that backgrounds may differ might not necessarily mean that historical factors should be downplayed to the extent of making up only three out of 22 indicators (instead of half, as with the HCR20). It could instead lead one to focus on different ones. For instance, the VERA2 tool used in Scotland includes several ‘historical’ and ‘demographic’ items, although not the same ones as those of the HCR20.

Neither the future is a strong focus of the ERG and VAF indicators. Whereas the whole ‘risk management’ part of the HCR20 is fully oriented towards evaluating a person’s probable future, VERA2 has a part dedicated to ‘protective’ items, many of which serve a similar function. Of the ERG and VAF indicators, even the ‘intent’ ones are not really aimed at clarifying individuals’ intentions, future plans, or situations they are likely to find themselves in. The authors of the ERG have explained that ‘intent’ refers to ‘the mindset associated with a readiness to perform or contribute to an extremist offense’ (Lloyd & Dean, 2015, p. 42); these indicators target attitudes or ways of being, such as over-identification with a group or cause and dehumanisation of the enemy. The absence of separate indicators explicitly focusing on a person’s plans or prospects seems odd given that a core aim of ERG and VAF assessments is to direct individuals onto interventions guiding them to a non-radicalised, non-terrorist future.

A key difference between the ERG and VAF indicators on the one hand, and the HCR20 and VERA2 indicators on the other is thus that the ERG and VAF ones are more concentrated on the present, whereas the other two tools overall have a wider temporal range. By comparison, single ERG and VAF assessments seems to provide a ‘radicalisation snapshot’ of a person’s psychology at the moment an assessment is being carried out; possibly sharp and accurate, but nonetheless narrow and frozen in time. In this way, the uses of the ERG and the VAF seem to represent a transfer of what risk scholar Louise Amoore has labelled the ‘derivative’ approach to risk from big data analysis and algorithmic predictions at population level to a small selection of information drawn from an individual’s intimate and personal spheres. Couched in Amoore’s vocabulary, the ERG and VAF indicators enable the translation of an uncertain future into actionable decisions through novel forms of correlation (Amoore, 2013, pp. 57, 76, 60, 58). Although the basis for assessments would remain rather narrowly delineated, the ‘derivativity’ involved in the uses of the ERG and VAF is alleviated somewhat by the practice of conducting several assessments over time and comparing these against one another.

In making such ‘radicalisation snapshots’ the basis of significant counter-terrorism decisions – such as whether to release sentenced terrorism offenders and how to intervene (or not) against non-criminal children and adults – England and Wales seem reliant on a ‘strong’ form of methodological individualism where the ultimate explanation and arena for intervention and change are located in the individual, personal space. Unlike the traditional and ‘weaker’ form of methodological individualism, the ‘strong’ version tends towards psychological reductionism, to some extent atomism (Udéhn, 2001, p. 216, 204; Udéhn, 2002, p. 484). Claiming that social, political and economic phenomena can and should be explained in terms of individuals and their psychology, ‘strong’ methodological individualism would only secondarily consider relations between people, and would regard political and social structures as even less important (Heath, 2015; Hodgson, 2007; Udéhn, 2002, p. 488). England and Wales’ uses of the ERG and the VAF indicators come across as one materialisation of assumptions of this kind translated onto the field of contemporary security policy, placing selected persons’ cognition and psychology as the basis and scene of counter-terrorist action.

To return to the differences between the ERG and VAF versus the HCR20 (and VERA2): apart from these tools’ diverging temporal scopes, one distinction between them is that the latter two tools are aimed at identifying the risk of violent acts – VERA2 extremist violence, and HCR20 general violence, especially among mental health patients. Meanwhile the primary focus of the ERG and VAF as these are used at present is to identify risk among people who have not committed direct acts of violence against others, and who are not necessarily suspected of doing so in the future: although the ERG is used on violent offenders too, non-violent offenders make up the majority of those sentenced under England and Wales’ terrorist legislation, and violence is generally not the core issue regarding those assessed within Channel.

The VAF is explicit on the point that its aim is to identify individuals’ risk of radicalisation rather than violence (HM Government, 2012, p. 2). The ERG authors’ definition of ‘intent’ above combined with their understanding of ‘engagement’ (Lloyd & Dean, 2015, p. 42) come close to how authorities often define ‘radicalisation’. This centrality of the concept of radicalisation to the practice of identifying extremism-related risk with these tools suggests a conceptualisation of radicalisation as both non-violent in itself but also inherently as risk. Indeed, this seems to be the unarticulated legitimating idea behind the uses of the ERG and the VAF in England and Wales: that radicalisation represents a terrorism-related (but not necessarily violent) risk warranting targeting, capture, and intervention.

One may ask whether and how these two conceptual differences between the HCR20 and VERA2 on the one hand, and the ERG and VAF on the other, could be related; in other words, whether the comparative concentration of the ERG and VAF indicators on the present may pertain to radicalisation, rather than violence, being their primary target. The 22 indicators’ relative emphasis on the present in aiming to capture the risk of radicalisation does seem to suggest an understanding of radicalisation as a condition of the here and now, a state rather than a process, but at the same time a situation of becoming and potentiality. The condition of being radicalised comes to resemble being ‘pregnant’ with a form of risk that is potent but as of yet unrealised: while possible outcomes for the radicalised are not unlimited, the condition’s precise materialisation will depend on the effectiveness of interventionist actions.

The slipperiness and disputes associated with the concept of ‘radicalisation’ would seem to make it ill-suited for recording and measurement through the conventional scoring format of a risk assessment tool. Like the HCR20, the ERG or the VAF are not formally actuarial instruments – that is, based on summative scales aimed at producing a number representing an individual’s overall risk. Now largely abandoned, the actuarial approach to risk assessment has generally been found too rigid and overall wanting in its ability to capture individual risk (see e.g. Geraghty & Woodhams, 2015; Mendoza, Rose, Geiger, & Cash, 2016). In practice, however, and like with most other risk assessment tools including the HCR20 and VERA2, carrying out ERG and VAF assessments does take the form of recording individual ‘scores’ on each of the indicators in a manner that will result in a sum, apparently quantifying individual risk. With regard to the ERG, the danger of relying too heavily on the resulting number is mitigated by the score sheet being accompanied by a narrative report considered to be the main product of an assessment. Nonetheless, the indicator summarising involved in every ERG and VAF assessment does suggest a conceptualisation of radicalisation not only as risk – but as measurable and quantifiable risk. And whereas incorporating uncertainty is becoming increasingly important for scholars and practitioners of other forms of risk assessment (see Guikema, 2012), the recording of the 22 ERG and VAF indicators does not encourage recording genuine uncertainty.

Recalling the ‘risk society’ literature produced in the wake of Ulrich Beck’s much-lauded and much-criticised landmark book, the ERG and VAF indicator list would appear like a characteristic ‘risk society’ attempt to pin down uncertainties that may in fact be incalculable (Beck, 1992, 1999, 2006; Rasmussen, 2006, pp. 38–39). Put in line with Amoore’s partial reframing of risk onto the specific field of post-9/11 counter-terrorism: as applied today, the ERG and VAF indicators seem aimed at making unknown futures actionable and amenable to intervention and management by extrapolating from limited information (Amoore & de Goede, 2008; Amoore & Raley, 2017). Indeed, a key part of the legitimating basis of the prominent place of the two tools’ indicators in England and Wales’ counter-terrorism seems to be their perceived ability to quantify radicalisation-related risk in a way that can provide the grounds for appropriate interventions.

Radicalisation and context

Reflecting the ‘strong’ methodological individualism underpinning the uses of the ERG and the VAF in England and Wales, the tools do not include indicators aiming to capture wider political or societal context; the main focus of the indicators is on the psychological domain of the assessed individual (see also Udéhn, 2002, p. 494). Although the ERG authors suggest that assessors incorporate also wider political and societal context in their assessments (Lloyd & Dean, 2015, p. 47), this is no requirement. The only forms of context the indicators explicitly open for accounting for is the immediate, private and situational ones of the assessed person’s family and personal network (see also Herzog-Evans, 2018, p. 14). In 2015, the authors of the ERG themselves acknowledged the danger of ‘political, cultural, and social context, together with protective factors (…) potentially slip(ping) from the analysis’ when using that tool (Lloyd & Dean, 2015, p. 50). With regard to the VAF, even such post-hoc awareness appears to be lacking – the only mention of context in the UK Government’s description of the VAF refers back to the tool’s ‘engagement’ category of indicators (HM Government, p. 2, see also above).

The absence of ERG or VAF indicators pronouncedly addressing political or societal context does not reflect the emphasis put on various forms of context by most of the terrorism scholarship, which generally conceptualises terrorism and radicalisation as thoroughly political phenomena. Most definitions of terrorism frame it as violence carried out for political, ideological or religious purposes, and definitions of terrorism-related radicalisation often describe it as a process where individuals increasingly become ready to support or carry out such acts. Simply put, an act of violence would not be terrorism were it not motivated by political, ideological or religious aims pointing outside of and beyond the perpetrator and that person’s most immediate surroundings. It thus seems odd for any collection of specialised terrorism-related risk indicators to not take this into account. With the benefit of hindsight, the authors of the ERG have also noted that it might have been an omission not to include political context in the tool (Lloyd & Dean, 2015, p. 42).

Political and societal context would also have bearing on the counter-terrorism utility of the ERG and the VAF since such context would impact on the evaluation of how much risk should be associated with each of the tools’ shared indicators. Or, put differently, how an assessment’s descriptive findings should be analysed in terms of risk and what interventionist consequences findings should have. To some extent, many of the 22 ERG and VAF indicators in effect entail judgements on the (in)appropriateness of various psychological responses, judgements which would make little sense in a vacuum.

For instance, many of the ‘engagement’ indicators, such as a need or desire ‘for identity, meaning and belonging’ might be unproblematically present in certain settings among seeking and questioning individuals; a need or desire ‘to defend against threat’ might be on point if context revealed that someone actually was being threatened. None of the 22 indicators seem to allow for systematically capturing relevant context of this kind. In a situation of military conflict, individuals might be expected adopt some level of ‘us and them thinking’, even ‘dehumanisation of the enemy’, to cite two additional indicators. When assessing individuals with recent experiences of war, for instance, recording this as part of an individual’s context could put findings into perspective and be relevant for determining their consequences. The same could be said for systematically incorporating more immediate kinds of possibly relevant context, such as – in the already mentioned words of the HCR20, ‘exposure to destabilisers’. Notably, factoring in context of either an immediate or broader kind would not mean registering different descriptive findings or ‘scores’ for the assessed individuals, but it could have significant implications for the analytical and risk assessment meaning of scores.

As already mentioned, although awareness of context is encouraged when carrying out an ERG, none of the ERG or VAF indicators obliges assessors to take wider political or societal context into account in a systematic and critical manner. This, of course, does not mean that assessments are context-free. Rather, the tools’ anchorage in ‘strong’ methodological individualism here appears to have led to a practice of implicit (over-)reliance on individual assessors: by not making mandatory the systematic and transparent inclusion of relevant political and societal context, the ERG and VAF indicators allow for each assessor’s own understanding of context to become assessments’ normative foundation. This increases not only the power imbalance between assessors and the assessed, but also the importance of assessors being appropriately experienced and trained.

Disconcertingly, however, the VAF tool is presently being used by professional groups that do not possess any specialised psychological risk assessment expertise, such as local authority bureaucrats and police officers (CT and Security Act 2015, p. 2, 37 1)). While one should assume that these would be both serious and well-intentioned, it is far from clear how they would be able to access even the basic descriptive evidence the VAF seeks to pin down, let alone be able to appropriately analyse these findings. Trying to map a person according to the 22 indicators through direct questions (such as ‘are you susceptible to indoctrination?’ Or: do you have a ‘need for dominance’ or ‘harmful end objectives?’) seems unlikely to yield useful results.

Trained forensic psychologists of the kind the ERG was created by and often used by in prisons may well be able to articulate the prompts that would bring to light the information the 22 indicators are after, translate this into the record sheet, and analyse their implications for interventions. For someone without their training – and even for the experienced probation officers among the ERG’s intended and continued users – accessing the psychological ‘data’ the indicators seek to pin down would be much less straightforward. The importance of assessor skill and training is further increased by the fact that in contrast to the HCR20, ‘the ERG is now completed by a single assessor’ (Lloyd & Dean, 2015, p. 49), with no requirement of either consent from or an in-person meeting with the assessed person.

Radicalisation and ‘content’

Not dissimilar to the way the ERG and VAF indicators are silent on the issue of political and societal context, they do not explicitly address the content of the radicalisation they seek to capture. The 22 indicators are not targeted at clarifying, for instance, religious affiliation, left/right political orientation, nor the ideology, values, or life philosophy of the assessed. Instead, they are oriented towards how individuals think, are, and hold their convictions. Radicalisation is associated with thought processes and ways of perceiving, rather than specific political, ideological or religious leanings. For this psychological conceptualisation of radicalisation, attitudes and manners of thought that hold the greatest definitional power. One could say that a legitimating conceptualisation behind the indicators and their counter-terrorism use is of radicalisation as a ‘transport infrastructure’ of human cognition, where routes and structures are more important than what exact ‘cargo’ is being transported.

Informed by experiences with more established risk assessment tools and especially the HCR20, distilling and acting on the kinds of psychological features the 22 indicators concentrate on intuitively come across as more useful to counter-terrorism than an approach preoccupied with particular political or religious affiliations. Decades of experience with the HCR20 used on mental health patients and its updates, validations and peer-reviews suggest that well-structured psychological analyses carried out by trained practitioners might help to identify the risk of violence among defined populations. From legal and democratic perspectives too it seems potentially less problematic to focus on cognitive structures and modes of thought in a ‘content-neutral’ way than starting from a point where certain ideological or religious beliefs are made suspect per se. All the same, the indicators appear simultaneously to capture a too limited selection of information, and to have a too indiscriminate – and thus far not validated – risk ‘filter’ (Rasmussen, 2006, p. 109) to be meaningfully used as they are at present.

This is not to exclude the possible usefulness of the ERG or other psychologically-informed risk assessment tools when considering, for instance, the bail or release of those sentenced for serious terrorist crimes, as one of several factors contributing to a decision. Notably, one of the drivers of the ERG creators was their experience of seeing all TACT offenders automatically being categorised as ‘high risk’ and placed under a maximum security regime, and their desire to accurately prioritise between this group of prisoners (Lloyd & Dean, 2015, p. 40; see also Liebling & Arnold, 2012; McCann, 2016). Another concern was that generic violent risk assessments could classify some terrorism offenders as ‘low risk’, ‘when in reality they presented with considerably more problems’ (HMPPS 2017, p. 2). Both concerns seem like understandable parts of the background to formulating a risk assessment tool dedicated to identify specified forms of risk among sentenced terrorism offenders.

The use of the same 22 indicators under the VAF umbrella, however, is a different matter. At that end of England and Wales’ counter-terrorism system, the VAF seems like a prime materialisation of the ‘precautionary principle’ that it has been claimed characterises post-9/11 counter-terrorism approaches to risk (Gelev, 2011; Rasmussen, 2006, p. 9; Salter, 2008; Stern & Wiener, 2008). From effectiveness and logical standpoints too, the transfer of the 22 ERG indicators to the Channel context seems based on tenuous grounds: The indicators were based on and record a retrospective terrorism offender profile, which the VAF then uses to predict risk among rather different populations.

While questionable, England and Wales’ transfer of the indicators from the ERG to VAF tools also seems easily explicable: Faced with the complexities of radicalisation, extremism and terrorism, authorities must have welcomed the indicators as a chance to possibly quantify risk in a way that could become the basis of counter-terrorism action, transmitting them from the more controlled and regulated prison setting to the very different context of Channel, and possibly beyond (see the reference to a ‘terrorist mindset’ at BBC, 2017). This transfer would also seem predictable to those risk researchers who have long warned that once risk assessment techniques are introduced as forms of governance, their proliferation would be difficult to constrain (Amoore & de Goede, 2008; Krahmann, 2011; Salter, 2008, p. 246).

The very fact that England and Wales today use the same 22 indicators to assess the potentially radicalised, the already radicalised, non-violent terrorist offenders, violent terrorist offenders as well as violent and non-violent non-terrorist offenders blurs the lines between these categories of people and indirectly posits a relationship or continuum between such actions and states of being; between being (possibly) radicalised and being a terrorist, being violent, and/or being criminal in other ways (see also Lloyd & Dean, 2015, p. 44). UK authorities’ now infamous description of the Channel programme as dealing with people in a ‘pre-criminal’ (rather than simply non-criminal) space suggests that those assessed for their ‘vulnerability to radicalisation’ with the VAF are expected to enact their (possible) radicalisation as terrorist crime unless intervened upon (NHS, 2015; also Goldberg, Jadhav, & Younis, 2016; Heath-Kelly, 2017b). The terminology of pre-crime now seems to have been abandoned, while the conceptualisation of radicalisation underpinning the continued counter-terrorism use of the 22 indicators appears to remain the same (NHS, 2017). Although concerns were raised with England and Wales’ prison authorities already in 2012 regarding whether an early version of the ERG really was valid for all the groups it was meant to cover (HMPPS 2017, p. 3, 20, 26, 39), this does not seem to have resulted in any decrease in the ambitions for the the tool’s uses.

This broad actual use of the 22 indicators at different parts of England and Wales’ counter-terrorism system suggests an idea of radicalisation as risk both in its own right, and as a likely precursor to violent, terrorist and other criminal outcomes. This idea, along with the ‘strong’ methodological individualist assumptions behind the use of the indicators, stand in contrast to recent years’ scholarship on radicalisation and terrorism. Much of the last decade of scholarly research has not only been eager to distinguish between radical thought and violent terrorist action. It has also increasingly highlighted various political and societal contextual factors as crucial to understanding radicalisation and terrorism. Although many questions remain regarding why some people commit terrorist acts while others do not, there is wide consensus that broader context is relevant (e.g. Basra, Neumann, & Brunner, 2016; Crone, 2016; Hegghammer, 2016; Nesser, 2016). Even, for instance, in the much-publicised debate between French scholars Gilles Kepel and Olivier Roy none of the parties dismisses the role of political and societal context, although their emphases differ (e.g. Kepel, 2017; Roy, 2015).

Conclusions

Given the ERG’s disciplinary rooting and original ambitions, that tool or its authors should not be faulted for concentrating on a person’s psychology in addressing individual radicalisation. Some of the arguments for developing specialised practices for assessing terrorism-related risk seem well-founded, and it would appear relevant to explore psychologically grounded versions. The actual use of the ERG in England and Wales’ prisons has also incorporated some relevant safeguards such as encouraging incorporation of context as well as vetting and training assessors, and it is promising that the tool is presently under review.

Any benefit associated with the 22 indicators, however, would seem conditioned on them being reserved for the uses, target groups and assessors they were originally created for, and on them being used as only one component of a broad, contextual assessment of individuals about whom there is an evidence-based terrorism-related concern. As has been shown, this would disqualify the VAF from its present counter-terrorism use, while also raising questions as to parts of the present uses of the ERG, especially when applied to non-terrorist offenders, and when used by only one assessor which is not a trained psychologist or psychiatrist. One might also ask whether aiming to measure ‘radicalisation’ at all might be a non-starter, given the concept’s controversial nature, unclear meaning, and uncertain precise role in leading to terrorist offences.

Besides the methodologically weak grounds for using the ERG-originated indicators in the Channel setting, the conceptualisation of radicalisation underpinning the indicators also puts the utility of their present uses into question. The indicators map out radicalisation as a risky psychological condition that will probably lead to terrorism if left unchallenged, that can be tracked down within individual patterns of cognition. The evidence basis for such a conceptualisation of radicalisation, however, has not been established. Finally, the use of the same indicators at two opposite points of an assumed radicalisation-to-terrorism scale not only relies on the much-criticised idea that such a scale exists. It also seems underpinned by ‘strong’ methodological individualism assumptions which posits that the primary cause and interventionary locus concerning terrorism is each person’s psychology and cognition. This places intra-individual psychologically conceived radicalisation at the core of the terrorism problem in a way that seems disproportionate if not in conflict with what research has found thus far.

It is imaginable that specialised risk assessment models and indicators could play a useful role in evaluating the terrorism-related risks associated with individuals about whom there is a reasonable security concern. However, as new variants for assessing and analysing such risks continue to multiply, policy-makers, practitioners and scholars would be well advised to take one note of caution away from the story of the ERG and the VAF. In particular, the transfer of the 22 ERG indicators to the VAF context illustrate the importance of creating safeguards against risk assessment ‘mission creep’ – both in terms of unwarranted expansions of tools’ target groups and in terms of the bloating of ambitions. In light of the uses of the ERG and VAF thus far, it seems necessary to protect tools, assessors, and assesses from applying indicators to phenomena that might in fact not be measureable, and from targeting populations beyond which they were originally intended.

Acknowledgements

Two anonymous readers are thanked for their critical comments to this article, which helped to strengthen it. The piece also greatly benefited from discussions with colleagues at NUPI and C-REX and at an EISA 2017 Young Researchers Workshop. Part of a pre-phase of the research for this article was enabled by funding from the Norwegian Ministry of Justice and Public Safety.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Dr Rita Augestad Knudsen is the Senior Researcher on counter-terrorism at the Norwegian Institute of International Affairs (NUPI), Security and Defence Group. She is also affiliated with the Centre for Research on Extremism at the University of Oslo (C-REX) and is the Managing Director of the Consortium for Research and Terrorism and International Crime.