Computational Statistics to Tackle Modern Slavery

If we are to meet the United Nation's Sustainable Development Goals by their target of 2030, we need to develop better statistical methods to map the prevalence of vulnerable populations. In this fellowship, I will A. carry out foundational research into effective computational statistics methods for hidden populations, B. use the methods to map modern slavery at local, national and international levels, and C. work with my project partners to change policy based on our evidence-based research.

To meet the Sustainable Development Goals, we need to measure how close we are to meeting them, quantify who is most in need of support and evaluate how successful interventions are in creating sustainable development. Take, for example, victims of modern slavery. Victims are often marginalised and hidden, with abuses going unreported and unmonitored. Estimating how many victims there are, where the abuses are happening and evaluating the effectiveness of interventions to support victims remain a challenge to the field of modern slavery and sustainable development more broadly. Data about victims and abuses is often noisy, poor quality or simply not collected. Developments in computational statistics can be really powerful here. They will provide a framework to deal with poor quality and missing data, while simultaneously avoiding specific and arbitrary assumptions about how the abuses are happening. Current methods require researchers to make specific assumptions about the abuses they are modelling which are difficult to justify from the data. The methods I develop will move away from this, instead making more general, mathematical assumptions. This will allow the data to speak for itself and can provide better counterfactual evidence and more realistic conclusions. To meet this aim, I bring a strong track record of developing these methods for epidemics, where my methods have been shown to reduce the need for specific assumptions when the data is poor quality.

However, this flexibility comes at the cost of a larger computational burden, increased uncertainty in the results, and a requirement for technical expertise when using the methods. To speed up progress to meeting the Sustainable Development Goals, researchers need methods that can be used in practice. I will lead the development of effective computational statistical methods. By reducing the computational burden, providing mechanisms to deal with the uncertainty in the results, and making methods easy to implement, they will become much more attractive to non-statisticians. I have already shown how my developments can considerably reduce the data collection burden when mapping poverty, making these methods more attractive to research and organisations working in poverty reduction. A key part of this fellowship is collaboration with a research software engineer who can develop data systems and software that other researchers and organisations can use to implement my methods.

I will use my methods to solve pressing problems in modern slavery and advance the field to meet the UN's goal to end slavery by 2030. I will work with my project partners to map modern slavery at local, national and international levels. This fellowship has the potential to save lives and show how computational statistics can advance progress towards the Sustainable Development Goals. By leveraging support from my project partners, I will influence politicians and policy makers to use my results to safeguard victims and prevent potential victims from suffering from modern slavery abuses.