The Freedom Fund has laid out an explicit commitment to learn about what works in the anti-slavery field and to share that knowledge. The evidence we generate will add to the growing body of wider research – but only if we ensure that the questions we ask are the right ones and are examined in the right way.

The question we (and others) sometimes face is ‘are you going to use a Randomised Controlled Trial?’ (RCT – see here for more information). But the question should be ‘what is the most appropriate research method for this context?’ The answer also depends on who is asking the research question – funder, program implementer, academic – and what they want to know including information on cost, contribution towards policy change, replication – there are a host of possible questions.

RCTs can sometimes be superior when the purpose is to understand an intervention’s true effect and to be able to assign cause to that intervention. DFID has laid out the requirements for the stringent design conditions that RCTs require to be effective. So might RCTs be appropriate for our field?

Interventions within social development are often complex and multi-faceted. But as we understand more about the extent of the modern slavery problem, and the vulnerabilities and the circumstances under which it exists, our projects and interventions are still being developed and evolving. Here, the answer we need is more often how and why an intervention might work and for whom in what context, rather than a single answer on attribution or evidence of the true effect of a specific intervention or component, so that we can shape and improve the program for its participants.

We know from many years of development experience that for interventions to be successful, they need to be tailored to the local context. This is also true for anti-slavery interventions, which often aim to foster ‘community ownership’. This means that the community is able to select their immediate priorities for work and may decide that the best protection against enslavement within their individual context is to make the local school functional, ensure that health visitors are coming to the local area, or develop a ‘collective resistance’ to debt bondage by a local landowner. Without this creation of local capacity for organising together and determining priorities towards a shared goal of ending slavery, efforts against slavery will be dependent on outsiders, and any results will be short-lived. In short, one group in one area may select to prioritise one action while another prioritises something rather different – and they build pressure on key actors through different strategies. Within this context, articulating a common protocol for intervention that could be tested through an experimental design is extremely difficult.

RCTs are also generally expensive to run, so it is critical that interventions are sufficiently advanced and well-defined to derive maximum value from the research. Very few anti-slavery interventions have reached this point. Investment in developmental research prior to large scale evaluations can help to ensure research is conducted at the appropriate time and can provide a better return on investment, especially when the nature of the intervention will also limit the ability to ensure fidelity of the RCT design. For example, changes in local government services or study participants moving location means following up participants can be difficult, which is needed to ensure appropriate statistical power.

Interventions also often seek to achieve wider systems change by encouraging learning and collaboration between different communities. In RCT terms, this might constitute ‘contamination’ between groups, thus limiting the ability to assign cause to the impact of the intervention – as the control or counterfactual is exposed and we cannot say what would have happened without the intervention.

Ethical concerns are paramount, especially with vulnerable populations that may often still be within a situation of exploitation. To generate evidence in this field, we need to consider the most appropriate research method to answer the question but also need to pay close attention to the current needs of the target population, political and ethical concerns, as well timelines for using the results to help refine and develop projects.

When the conditions are right, evidence generated by the RCT approach could provide robust evidence for our field in well-defined programs with specific outcomes such as targeted information campaigns or training schemes and may be particularly useful to inform decisions about investing significant sums of money to scale up an intervention. However, it is critical that we draw from the range of evaluations methods and select those that are most appropriate to the research question and context.

Photo: Ginny Baumann © The Freedom Fund