It is almost beyond doubt that violence, and the threat thereof, should disrupt the theories of change that underpins humanitarian and development programming in much the same way that it disrupts just about everything else.
Skills programmes might churn out newly qualified graduates into a labour market with collapsing demand. Infrastructure programmes could easily become a direct target of attack. Health centres and schools could become isolated or unstaffed.
At the same time, much understanding about how programmes work and are supposed to work has been developed in situations that are well suited for them to perform well; and that are ideal for data on effectiveness to be collected.
Programmes that work well in ‘ideal scenarios’ might not work as expected, if they work well at all, in more difficult settings. Yet standard programmatic typologies are seldom significantly modified when they are placed in a (potentially) violent situation.
This should not be taken as a cynical blindside at an entire field. Running good programmes and doing good research in these ‘ideal scenarios’ is hard enough, not least because no situation that requires humanitarian or development interventions is ‘ideal’ in any conventional use of the word.
Rather, ‘ideal scenarios’ are places where a confluence of circumstances – such as near-passable transportation networks, the absence of open violence, or relatively stable populations – facilitates the achievement of the programme’s aims.
It is entirely understandable why the rollout of programmes that perform well in these scenarios is applied in environments that are more complex. In a field that is notoriously ‘knowledge light’, stretching what could be accumulated is sensible.
The existence of contextual challenges to evidence-based theories is, far preferable to stabs at the dark.
This, of course, means that while many ‘standard’ programmes are implemented in risky scenarios without much modification, not all “standard” programmes might make the cut. Anecdotally at least, it is often assumed programmes with the simplest theories of change have the best chance of working as the complexity of a setting increases.
The logic here is simple and intuitive: the shorter the causal chain, the fewer links there are to be disrupted by external complexities and thus, the more durable the programme is likely to be.
However, a simple thought experiment hints that this is not always the case. The theory of change associated with the direct provision of food supplements to malnourished people, for example, requires no further elucidation.
Insofar as these things go, giving more calories to people who are experiencing a shortfall could not be simpler. Once an individual receives his or her allowance, the opportunities for violence to intervene are small, if extant at all.
Yet, the process of actually getting the supplements to those who need them is layered with logistical complexity on both the demand and supply side.
Supplements are seldom, if ever, produced in the locations where they are needed. This means that processes must be in place to distribute them to the regions, towns and villages where they are most needed.
The threat of violence poses an obvious threat to the successful functioning of these supply networks. Local distribution – often in the form of health centres – is also required.
Such places could easily be the targets of violence and, even when not, could be left unmanned as workers flee violence. Even if this does not happen, demand-side challenges also arise. People must travel – sometimes long distances – to these distribution centres in order to receive their entitlements.
They are, obviously, less likely to do so if the way is dangerous. That is without considering that conflict might drive people away from their hometowns and regions (and thus from the centres at which they are registered).
By contrast, programmes with more complicated theories appear less promising to begin with. Take a programme that aims to boost nutrition by helping households to build assets.
This relies on the household successfully building up an asset, effectively using or selling the asset, and then deciding to spend at least some of these gains to increase (the quality and quantity of) food consumption.
Violence could theoretically interrupt this process at multiple points – for example, by reducing the likelihood that the asset is built, through impacts on household behaviour or that the asset is lost as the household escapes violence. At the same time, such programmes can be administered in a much more ‘hands-off’ manner.
Providing cash as an ‘asset’ involves a one-time transfer of cash, which contrasts significantly with the ‘on-going’ nature of direct supplement provision and boosts nutrition. Even when the ‘asset’ is more complicated– such as the planting and harvesting of lumber for land rehabilitation – the inputs that set the theories of change in motion are, comparatively, simple and thus, potentially more tractable in violent settings.
This means that there is a trade-off between the susceptibility of the provision modality to violence and the susceptibility of the theories of change. The interventions with the simplest theories of change might not be those with the simplest distribution logistics, and vice versa.
Interventions with both simple theories of change and easy provision modalities, however, might not exist for all forms of provision. The question that logically stems from this is whether simple theories of change, or simple distribution logistics, are more robust to threats of violence.
Providing an answer to this goes far beyond what it is appropriate to say. At the same time, it is exactly this kind of trade-off that we analysed in Niger – a place characterised by chronic insecurity, chronic fragility and frequent manifestations of lethal violence within its borders – where we had the chance to evaluate two WFP interventions.
While the direct provision of calories is not shown to have an impact immediately after the programme period, the same is not true for nutrition-sensitive asset-building provision. Recipients of this assistance show significantly higher nutrition indicators than their compatriots in the control group and those who received direct nutrition support.
Given the support for the effectiveness of food aid in more “ideal scenarios” throughout the literature, this almost certainly confirms that the theories of change associated with nutrition programming are confounded in highly complex scenarios.
It follows, at least in this one specific setting, that programme typologies with simpler roll out logistics out-perform those with simpler theories of change, even when the latter have regularly been shown to perform well elsewhere.
This should provide food for thought for those who, understandably, seek to apply learning from elsewhere to the complicated scenarios in which they work.