If there is one thing I’ve learned recently, it’s that violent extremism is a complex problem. Here in the Horn of Africa, a vast number of potential causal factors exist to incite someone to violence and the connections between them are often hard to see.

And yet, despite the complexities, we (the international community) often fall into the trap of trying to design responses to such threats using tools and approaches built for simpler, more conventional problems. This is fraught with risk.

On our quest to learn more about how complexity theories can be applied to real-world problems such as violent extremism, we’ve been grappling with the question of how to construct more adaptive tools (such as theories of change and results frameworks) for CVE.

In doing so, we’ve drawn heavily from thinkers such as Ben Ramalingman, Duncan Green and Marcus Jenal who have broken new ground in the field. Below we offer a compilation of (mostly their) thoughts to stimulate more discussion. Let us know what you think – we would love to hear your thoughts!

A number of problems with conventional approaches to creating theories of change and results frameworks

  • They assume that causal pathways are known in advance of implementation. The assumption that there is sufficient knowledge about the chain of results can be highly dangerous when applied to the uncertainties of violent extremism (VE).
  • They over-simplify messy realities that then become entrenched in implementation. Defining the causal pathway at the outset ignores the dynamic interactions among various parts of the VE problem.
  • They assume the problem can be treated in reductionist ways. Conventional approaches risk assuming that the problem can be separated into its component parts and that solutions can be applied in a replicable and reproducible fashion.
  • They typically engage with contextual factors in delivery only, rather than in design. They can often assume certainty about the design of an intervention from the outset, which is nigh on impossible in relation to the nature of VE.
  • They reduce the willingness to adapt the design over time. They create strong incentives to stick to and report on what was agreed from the outset.
  • They impose limitations on capturing unexpected outcomes. Further to the point above, these can impede the ability of CVE programmes to explicitly look for wider outcomes and results.

Some thoughts on creating an ‘adaptive Theory of Change’[i]

  • A Theory of Change (ToC) is only useful if it actually informs decision-making. Sounds self-evident, however many Theories of Change gather dust once they have been created to sell the concept of a project.
  • A ToC can never be perfect or fixed, at least for CVE programmes. A ToC should be an idea that is alive and dynamic. A ToC has, in the first place, to be useful for the people who work with it. It is a tool to discuss, debate, experiment, learn, change.
  • We may never figure out how change will happen in complex systems. Not only can we not know the pathways, we are also unable to predict the exact shape of how the change will look. In some cases we might even not know what a good outcome would look like before we see it. Complex systems can only be understood if we interact with them.
  • We may need to start with, and test, multiple Theories of Change. For complex problems such as VE, there may be multiple competing hypotheses of how the intended change could be achieved, or what it could look like. The available evidence may support different and even competing perspectives.
  • A well-designed ToC cannot change complexity. The complexity of VE cannot be simplified or reduced with a nice, neat, logical model or theory. Recognise the dangers that any ToC is a potentially dangerous simplificiation of the real world.
  • In a complex system, different people will have different perspectives about how things work, which may not be amenable to analysis with a single, simple model. On the contrary, it is important to understand where there is agreement on causalities among the stakeholders and where there is not – this gives us important insights on the complexity of specific links in the logical chain.
  • The ToC needs to be presented as an overarching framework that explains how the programme intends to work, but without detailing the specific mechanisms of change (i.e. interventions). This will help ensure the theory of change remains valid even if individual interventions are adapted, closed down, or scaled up.
  • Mini Theories of Change may be required to allow adaptation and testing. In conditions of significant uncertainty, an adaptive, learn-as-you-go approach is essential. It makes sense for programmes to include a range of exploratory interventions that can be scaled up, or brought to an end. These projects may run independently of each other, and each should be thought through with its own mini ToC.
  • Design it to evolve over time based on new evidence. Recognise that any Theory of Change is just that – a theory. Given the (a) lack of data and (b) debate over what data there is in the world of CVE, we should expect current theories to evolve, and quickly. We must apply this thinking to the ToC from the outset.
  • Its possible that the most important goal may be about how responsive the programme is to the local system. Given the outcomes of intervening in a complex system may be unknowable, it is worth considering that the actual goal of the intervention itself may be the extent to which it is responsive to the problem (rather than defining the impact it will have).

Adapted from Essays on complexity and Theories of Change. Marcus Jenal. 2016 https://marcusjenal.wordpress.com/

Also see Navigating Wicked Problems in Development