ESRC Research Centre on Micro-Social Change (MiSoC)
Paul is convenor of MiSoC‘s cross-cutting strand of methodological work on the use of machine learning in the social sciences together with with Spyros Samothrakis, Damian Machlanski and British Academy Postdoctoral Fellow Annalivia Polselli. We have developed machine learning for causal inference of conditional average causal effects, explored the importance of hyperparameter selection for both causal effect and structure/causal discovery methods, and are developing double machine learning estimation for causal panel data analysis. See here for more details about the project.

Understanding Society
Paul is a co-investigator and Statistical Methods Topic Champion on Understanding Society. He has collaborated with Yanchun Bao on methods for estimating different kinds of mode effect from Understanding Society and other surveys, and is currently working with Jamie Moore developing novel weighting schemes for the Covid-19 Survey. Additionally, he develops advanced longitudinal methods for the analysis of Understanding Society and other complex longitudinal household surveys.