To project overview

Can we predict infection before it happens?

Annual Review Article 2020

The corona crisis is the biggest global crisis in the 21. century. Photo: Martin Sanchez, Unsplash

Virus infects from person to person and, accordingly, people’s behaviour are a key to understanding how an epidemic unfolds. If people practise social distancing and maintain good hand hygiene, the epidemic is kept at bay. We have heard that time and again during the corona epidemic. So if that is right, can we use indicators of people’s behaviour to predict where and on whom the infection pressure will fall and increase during an epidemic? This is the focus of Michael Bang Petersen’s Semper Ardens project “HOPE – How Democracies Cope with Covid19: A Data-Driven Approach”.

By Michael Bang Petersen, professor, PhD, Department of Political Science, Aarhus University

An epidemic is a special form of crisis. The politicians cannot just ask people to “keep calm and carry on”, as was the British slogan during the Second World War. The politicians cannot resolve an epidemic by themselves. 

They are reliant on significant changes in people’s behaviour. People need to practise social distancing, avoid large gatherings and maintain a high level of hand hygiene if the infection curve during an epidemic is to be flattened. In order to understand the course of an epidemic, we must therefore understand people’s behaviour. 

This is the focus of the researchers behind the HOPE project. HOPE stands for “How Democracies Cope with COVID-19”, and the project is being led by a multi-disciplinary, crossuniversity team comprising Rebecca Adler-Nissen (University of Copenhagen), Andreas Roepstorff (Aarhus University), Sune Lehmann (Technical University of Denmark) and myself. 

The hope is that the project can give us a deeper understanding of how, in a crisis, democracies can navigate dilemmas related to health, the economy and personal freedoms. The starting point is the corona pandemic, which hit the world in 2020. 

But the project is not just important here and now. It is also important for us to have effective preparedness for the next time a pandemic or other significant crisis hits the globalised world.

How do the Danes act during the time of Corona-crisis?

Big data revolution

The HOPE project adresses how democracies react during the corona crisis. Photo: Markus Spiske, Unsplash

A unique feature of the corona epidemic is that it arose in the middle of the big data revolution: that there are unprecedented possibilities for collecting and analysing huge quantities of data around people’s behaviour while the epidemic is ongoing. 

In the HOPE project, we are studying how the authorities’ communication and decisions are cascaded down through the whole of society across a number of western democracies, with a particular focus on Denmark. 

With the help of computer power, we are continually harvesting data on the content of press conferences, on the content of the media’s articles, and on the mood and the balance between information and misinformation on social media. 

With the help of surveys, we are continually collecting data on people’s attitudes and concerns. With the help of data from mobile phones, we are continually harvesting data on people’s movements around society. 

And with the help of anthropological field studies, we are learning not just what Danes are doing, but also why they are experiencing, thinking and feeling what they are.

The politicians cannot resolve an epidemic by themselves. They are reliant on significant changes in people’s behaviour

Early warning signs?

An epidemic is not just a matter for health research. People’s behaviour is crucial, which means an epidemic also requires insights from the social science and humanities. 

A central research question in the HOPE project is whether classic and new data collection methods within social science and the humanities can increase our possibilities of predicting the course of epidemics. One challenge is that people’s infection-preventing behaviour is largely a private matter. 

How can we investigate, for example, whether people refrain from meeting up with friends? Or whether they are sanitising their hands at home? Behaviour tracking using mobile phones’ GPS functions can provide some information. 

But the data mostly tells us how people move around in society. And even if people are moving around a lot, they are not infecting anyone if they are mainly doing so in a car with washed hands. 

An alternative data source is the population’s own responses collected by means of questionnaire surveys. Through simple random selection, we can survey a representative sample of the population. 

And through well-designed questions, we can measure private behaviour such as frequency of hand-washing, number of contacts and degree of physical distancing. 

Finally, questionnaire surveys have another benefit. The use of data from Facebook, Google and telephones can make people concerned about whether data is stored securely and whether privacy is adequately protected. Questionnaire surveys, by contrast, are easy to complete in a fully anonymous fashion. 

An epidemic is not just a matter for health research. People’s behaviour is crucial, which means an epidemic also requires insights from the social science and humanities

But questionnaire surveys also present some challenges. Can you trust what people say? And can they actually remember what they did yesterday? These are important questions, but there are at least two things that show that we can extract important signals from people’s noisy responses. 

Firstly, much of the noise disappears as a result of aggregation. We know this from the TV programme Who Wants to be a Millionaire? Here, contestants can ask the audience if they are unsure about an answer. 

Miraculously, the audience almost always gets the answer right! Random noise in individual responses is cancelled out when you take an average. 

Secondly, we are interested in changes. Have people become less aware of the health advice and are they keeping less of a distance? Provided the noise is constant over time, the changes in the data will give us a clear signal about changes in the population. 

In the HOPE project, we are relying on GPS data from mobile phones and daily survey responses from 500 randomly selected Danes. 

In this way, we can compare the strength of the new and classic ways of collecting information on people’s behaviour. And if we are able to identify a strong association between people’s infection-preventing behaviour and the infection’s actual transmission, in the long term this will give us new tools for handling epidemics. 

Questionnaire surveys of people’s behaviour can thus serve as a cost-effective early warning system that can tell us whether a new wave of epidemic is on the way. And, as a result, we can change course before the infection spreads and people end up in hospitals.