Empirical evidence for terrorist targeting frameworks

With infinite time and resources available, all potential targets for significant terrorist activity could be protected from harm as far as is practically possible. Given real world time and resource limitations, it is necessary to prioritise some locations over others.

Existing Frameworks

To better understand the locations that might be at greatest risk, we have risk assessment frameworks for assessing the desirability of targets to terrorists. These are based on principles of situational crime prevention (SCP). For proponents of SCP, the offender always has a choice, and action can be impeded by factors such as increasing the effort (e.g., target hardening), increasing the risk (e.g., strengthening surveillance), or reducing the rewards available.

The terrorist targeting frameworks are typically lists of criteria thought to make a location more desirable to an offender. Two well-known frameworks (with very handy acronyms) are EVIL DONE and TRACK.

Exposed
Vital
Iconic
Legitimate
Destructible
Occupied
Near
Easy
(Clarke & Newman, 2006)

Tolerable
Relevant
Accessible
Close
Known
(Marchment & Gill, 2022).

The least self-explanatory of the criteria may be legitimate, tolerable, and relevant. Legitimate refers to the targeting of victims who are ‘most deserving’ (e.g., military personnel or government, rather than the general public). Tolerable refers to pre-attack detection (including whether the individual can reach the point of attack without being overcome by fear). Relevant refers to fit with the ideology of the offender.

...there has been little evaluation or empirical investigation about the capacity for these frameworks to actually inform target selection.

Empirical evidence

At first glance, these frameworks are made up of sensible factors that we might expect to inform decision-making (an actor might target a location with a large number of potential victims that also has little security etc.). However, nearly two decades since the release of EVIL DONE, we find it surprising that there has been little evaluation or empirical investigation about the capacity for these frameworks to actually inform target selection.

For the most part, the frameworks have been used in descriptive exercises. For example, in 2006, Clarke and Newman illustrated only that the components apply to well-known sites in Washington DC that one might imagine are potential targets. We lack empirical insight in the UK context, and in the limited cases where data has been used in an attempt to evaluate utility using cases overseas, results tend to diverge from expectations of the frameworks (for a comprehensive review, see Monaghan et al. 2023).

Our findings

In light of this, we responded to a call from the National Counter Terrorism Security Office to examine the relevance of the ‘EVIL DONE’ framework (plus some additional components from ‘TRACK’) to terrorism cases in the UK between 2015 and 2021 (Monaghan et al., 2023). By looking back at 184 available cases, we investigated which of the components were consistently present or absent across cases (i.e., attractive to, or avoided by, offenders).

...we investigated which of the components were consistently present or absent across cases (i.e., attractive to, or avoided by, offenders).

Of course, most cases of terrorism in the UK are based in Northern Ireland. Many of these incidents are localised and do not fit with the kind of international terrorism the frameworks were designed for. Therefore, many of the components didn’t map well onto UK cases, with some being absent on almost all occasions.

For the 22 cases from Great Britain, we found that the components fit somewhat better. For GB, it seems that offenders tend to avoid locations that are relatively indestructible, unoccupied, far away, highly secure, likely to expose them pre-attack, or irrelevant to their ideological motivation. A greater spread of scores for the other components (exposed, vital, iconic, and legitimate) indicate that they may be less helpful in determining attractiveness.

However, there were also obvious limitations of utilising retrospective rating techniques to make determinations, like trying to deduce motivation and intent of an offender after the fact, and without all the relevant information. To overcome some of these limitations, we have been working with NaCTSO on a red-team approach to better understand the decision-making process.

How much can we understand?

There is an overarching question about the extent to which target selection can actually be predicted. A predictive model of targeting behaviour (and a response rooted in situational crime prevention) relies on an assumption of ‘rational’ decision-making processes and struggles to account for elements of chance and unforeseen opportunity. We have seen that decisions can change at the last minute or even during the course of an attack. These frameworks also do not account for value-based decision-making. We know that, particularly people who act in the name of sacred values, utilise rule-bound logic rather than weighing up costs and benefits in a way that might be conducive to the principles of situational crime prevention (Berns et al., 2012).

So, can these frameworks really reflect hostile decision-making with the necessary complexity to be informative? We hope that this series of studies will contribute to a better understanding of these questions, and an evidence-based approach to understanding and mitigating the threat.