Science of violent extremism risk assessment and management: An agenda for moving beyond the ‘state of the art’

1. Synthesise past research

There is no doubt that studies on violent extremism are going through an empirical renaissance (Schuurman, 2020). Recently, scholars have considered the growth in violent extremism research through reviews of recent advancements (Pressman & Davis, 2022; Gill et al., 2020), summaries of instruments (Clesle et al., 2024), and meta-analyses for specific populations, such as juveniles (Emmelkamp et al., 2020). However, there is still a pressing need to take stock, further synthesise this knowledge base through broader meta-analytical designs, and to organise this knowledge conceptually.

2. Move from prevalence to relevance

Much of the existing research on violent extremism risk factors is focused upon prevalence rates of certain risk factors across different cohorts. Focusing upon the ‘presence’ of risk factors is just one half of the risk assessment and management puzzle. The other, trickier, part is understanding the ‘relevance’ of a risk factor to the genesis of violent extremism. Recent research has begun to try address this gap, for example by generating base rates of risk factors using a general population (Clemmow et al., 2020). However, limitations exist with this type of work, and unpacking this requires additional studies and more methodologically sophisticated research designs.

There is a growing understanding that many risk factors for general violence are also risk factors for violent extremism.

3. Compare across violence type

There is a growing understanding that many risk factors for general violence are also risk factors for violent extremism (Gill, P., Marchment, Z., Seaward, A., Doherty, P., & Goodman, K, 2024). This necessitates organising the general violence risk factors literature also via meta-analytical designs and weighing this evidence against our conceptual frameworks for violent extremism. More fundamentally though, it also requires an examination of whether the ‘relevance’ of a risk factor is the same across violence types. It is likely not. For example, alcohol abuse is a risk factor for domestic violence because it often immediately precedes the onset of violence (e.g., Foran & O’Leary, 2008).

Previous studies also show high prevalence rates of alcohol abuse amongst violent extremists (Gill et al., 2021). However, its functional role for violent extremism is likely very different because the individual, in many circumstances, may replace their substance abuse problems with extremism. Consistently highlighting these dissimilarities – and conducting future research to compare risk factors across types of violence – is important for risk assessment guidance.

4. Compare across contexts

While some studies have made great strides forward in identifying the base rate prevalence of risk factors within the general population (Clemmow et al., 2020), there is now pressing need for more fine-grained comparative designs to test these across different national contexts and sub-sets of the population. This will move us toward an understanding of what factors can be considered universal risk factors, and which are contextually derived.

5. Look beyond a single factor

There has been a tendency in the literature to search for silver-bullet type explanations. Single factors are regularly proffered as being crucial, especially in response to a major attack. Extremist ideologies, mental illness, prison experience, online experiences; the list of univariate types of explanations of radicalisation are endless (and usually lacking empiricism). In reality, the situation is much more complex and typically involves a crystallisation of personal, political and social drivers. The field should seek to move toward an empirically-informed, practitioner-relevant multivariate understanding of the risk of extremist violence.

Focusing upon the ‘presence’ of risk factors is just one half of the risk assessment and management puzzle.

6. Include protective factors

The field generally has a lack of understanding around protective factors. While some research has been done to evaluate possible protective factors (Lösel et al., 2018; Marsden & Lee, 2022), studies generally have had a tendency to only look for ‘risk factors.’ This was also true in the general violence research literature until only recently, but research has shown the value of including protective factors alongside the traditional risk factors (Robbe et al., 2013). In practice, there is a major demand for a greater focus on protective factors.

Protective factors directly reduce dysfunction and thus can negate risk and dampen the impact of experienced risk factors. Protective factors are therefore often the focus of intervention plans. Yet, their scientific underpinnings are thin due to a lack of concerted research focus. This requires a synthesis of existing knowledge from the general violence literature, but also a test of competing explanatory models for how protective factors work in practice in terms of their mean, moderating and mediating effects.

7. Replicate

A number of well-established disciplines are undergoing so-called ‘replication crises’ including psychology, medical sciences, economics, and the social sciences (Schumann et al., 2019). Failed replications occur for many reasons including lack of sufficient detail on design or procedure in the original study, or issues concerning sample size. A field like terrorism studies is likely not immune, especially with concerns such as small sample sizes, low base rates of offending, and limited access to populations. We require dedicated effort on replication work to set out to test the veracity of previously reported results that have been consequential for the field.