Bayesian Computation for Modern Slavery Statistics

Victims of modern slavery are exploited for personal and commercial gain. In 2015, the Home Office estimated that there are 11,000-13,000 victims of modern slavery in the UK (Silverman, 2020). Data about modern slavery is often, noisy, sparse or censored. Tin this project, we will develop Bayesian methods for statistical models to estimate the prevalence of and risks associated with modern slavery. There are several models available (e.g. Multiple Systems Estimation, Network Scale Up, and Comparative Judgement models), but these models rely on assumptions that are not always valid for modern slavery data. This project will be about reducing or removing these assumptions and developing ways of fitting these models quickly and efficiently.