How can we tell if our efforts to prevent violent extremism are working?

The problem

Trying to work out if the money spent on development projects has made a real difference is hard; when this money is directed at sensitive and intangible goals like countering violent extremism (CVE), this gets even harder. CVE programmes, by nature, are designed with ambitious goals; they often seek to reduce or eliminate the violent extremist threat in a specific area, which is seemingly impossible to prove. So how, then, do we try to identify any sort of impact or changes in CVE programmes and attribute these changes to specific interventions?

A number of reports have been published in recent years (such as this one, from USIP) discussing the challenges and proposing partial solutions to this problem, with no single approach being perfect and each one having its drawbacks.

First though, we have to figure out where to start; that is, what particular change are we trying to measure? For example, we could try to assess whether there has been a reduction in recruitment, in violent extremist attacks, or in support for a given ideology. We also may be interested in changes in individuals’ or communities’ perceptions toward the government and security sector or toward an extremist organisation. Each of these higher-level impact goals comes with its own set of challenges. How do we prove, for example, that there has been a reduction in violent extremist activity? While over a given period, we may see a decrease in the scale and frequency of attacks, it may not mean that the threat is decreasing. Instead, the organisation could be restrategising, planning or biding its time until the heat dies down, or it could be moving attacks to a different location while recruitment remains high in a given area. If we are to take the case of Kenya as an example, the volume of extremist attacks in country is comparatively low. Instead, Kenyans are often recruited to join Al Shabaab in Somalia or ISIS in Syria or Libya. Looking only at a decrease in attacks as an indicator of a decrease in violent extremist incidents prevents us from seeing the full picture.

At the same time, looking at patterns and trends of recruitment into violent extremist organisations is equally challenging. Without datasets or even accurate estimates on the number of individuals joining these organisations, it is almost impossible to discern any sort of pattern. Gathering data on recruitment trends is likely an insurmountable challenge in and of itself; how do we know if someone who disappeared has left to join Al Shabaab or ISIS, gone abroad for a job and not told anyone, been arrested by the police, or simply decided to run away? How do we even know if someone has disappeared in the first place? Lacking any trusted and effective reporting mechanisms, many family members and peers of those who have disappeared do not file a report and the disappearances are not recorded.

So what do we do?

Increasingly, programs are starting to measure changes in behaviours and perceptions as proxies for measuring a reduction in violent extremism. The argument, according to the USIP article cited above, is that ultimately, CVE programmes intend to adjust the mindsets and therefore the actions of individuals or communities in order to reduce their support for or propensity to engage in violent extremism. A range of innovative tools are being developed and implemented to try to better understand changes at this level, but limitations remain.

Firstly, some of these approaches assume causal links that may not necessarily exist. Even if an individual supports more radical beliefs and ideologies, or harbours more negative perceptions of the state, they may not necessarily be motivated to action. What triggers one individual is likely very different than what triggers another. Secondly, and particularly within the Kenyan context, there are numerous examples of individuals recruited into violent extremist organisations for reasons that have nothing to do with ideology, beliefs or perceptions. In this case, simply using measures of these indicators hinders the development of a more holistic understanding of drivers, motivating factors, and therefore an understanding of changes in extremism trends.

Using perceptions, attitudes and beliefs to understand propensity to support an extremist organisation is a good starting point, but we must also seek to evaluate other factors at play within a given context. The importance of social or familial networks, for example, likely have an impact on an individual’s motivation to action, as well as other vulnerabilities that may marginalise some segments of a population over others, such as access to national identification cards, job or educational opportunities, or feelings of unequal representation in government or politics. The trick is that the significance of these factors (and likely many others) differs quite significantly by location. Even in Kenya, relevant drivers or stressors in Isiolo, for example, are not applicable in Garissa, and vice versa.

With all of these compounding considerations, I am still left wondering, what’s the solution? What’s our magic bullet? Short of the ability to see five years into the future to understand if the relevance of these extremist organisations has increased or decreased, we remain unable to completely and accurately understand our impact. But this limitation presents an exciting opportunity: the chance to innovate and experiment with new techniques and ideas and learn from each other about what is working and what is not working.