LSE Statistics research in response to COVID-19

The data generated by the Covid-19 pandemic should be seen by statisticians and data scientists as a natural entry point to helping address this global health crisis. Sarah Callaghan, Editor-in-Chief of “Patterns”, refers to Covid-19 as a “data science issue” (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7144860/). Indeed, statisticians and data scientists the world over are hard at work trying to understand this unique event from the perspective of their disciplines. This page gathers our own research on this theme, and includes papers, summaries, opinions, trend estimates and blog posts. We will continue to add information to this page, so keep coming back to see what else we have been working on as a department during this pandemic.   

Blog post by Professor Jouni Kuha and Professor Patrick Sturgis


How many people in Britain have already been infected with the coronavirus? The truth is, we don’t really know. Yet, this is a very important number because, without it, it is difficult to properly assess the severity of the virus, its likely trajectory, and the consequent burden on health and social care systems. Read more in this insightful blog post by Professor Jouni Kuha and Professor Patrick Sturgis. 

Trend estimates by Professor Piotr Fryzlewicz


Professor Piotr Fryzlewicz has worked on some trend estimates and forecasts in relation to COVID-19:




Modelling time evolution of spatial linkages for UK and Europe under COVID-19 by Dr Clifford Lam


Dr Clifford Lam has written a short summary below on his work regarding COVID-19.


The research is on modelling the time evolution of the spillover effects – or, the spatial linkages on the rate/number of infections/deaths, across different countries in Europe or different regions in UK as restrictions of movements and closure of businesses are imposed over time.

Why is such a modelling important? Traditionally such spillover effects are specified by a researcher, and other aspects of a spatial econometric model are estimated based on the specifications. Such a traditional method assumes that we have correctly specified all spillover effects between each pair of regions, which is in fact an overwhelmingly difficult task in itself. Moreover, it is also assumed constant, which is of very limited use for situations such as COVID-19, where the dynamics of such spillover effects can change based on policy changes.

With our modelling of changes of spillover effects over time using COVID-19 data, we can trace back the effectiveness of the policies implemented in different countries and regions from our data analysis. Over a longer time horizon, we will also be able to trace how the reduction in movement of people affected regional integration and economic outcomes.

The above can in fact be used for broader regional science studies. The effects of economic integration across Europe from the effects of Brexit, for example, can be studied using our method. 

What will be done?

Since the modelling of time evolution of spatial linkages is completely new, we first need to develop the corresponding spatial model for this. Lam and Souza (2019) has successfully analyse a model where spatial linkages can be estimated from data with inference tools available as well. The model lays down a solid foundation for an extension of modelling time evolution of spatial linkages, and the necessary theoretical analysis for the development of inferencing tools. R package will be developed for the purpose, adding with visualisation of how the spatial linkages, as a network, changes over time.