The British public that headed into the pandemic was not particularly data confident. A pre-Covid study found that only 10% of UK workers felt that mainstream education had prepared them to work with data. Given this, data practitioners have had to tailor their presentation of statistics to ensure that key messages land with little room for misinterpretation.
Tailored messaging has created a better appreciation of the need for policies that have placed unprecedented restrictions on our everyday lives. Simplified, surface-level statistical concepts such as “flattening the curve” and the “R number” have entered the lexicon. An increasingly data-savvy population has utilised open and accessible data sources to keep informed of, and scrutinise, decision making.
Public engagement with data has allowed for models and processes to undergo iterative improvements – the Covid-19 dashboard itself has gone through 11 redesigns, evolving from a few line charts to an interactive hub to satiate growing user demand. The latest iteration of the dashboard publishes around 200 metrics – or around 40 million individual figures – daily.
“Simplified, surface-level statistical concepts such as ’flattening the curve’ and the ’R number’ have entered the lexicon.”
Practitioners are hopeful that, unlike the virtual pub quiz, the demand for real-time information and accessible data will persist long after the pandemic has passed into the rear-view mirror. This would support the government’s National Data Strategy, published in September 2020, which aims to “harness the power of data to drive growth and innovation, fuel new jobs and businesses, support scientific research, revolutionise the public sector and create a fairer society for all.”
In turn, a more data literate public would better understand how, and why, their data is being used – enabling for the protection and strengthening of people’s rights as citizens and consumers.
Creating a narrative
Releasing data into the public realm is a delicate act. Increased visibility doesn’t necessarily translate to increased literacy. At the onset of the pandemic, methods of data presentation had to be carefully considered in order not to confuse an already anxious population.
“You have to tailor to your audience,” says Rony Arafin, head of analytics at NHS England and NHS Improvement. “It’s the same problem when you go to the GP and they use all sorts of medical jargon – it’s a total waste of time.”
A study conducted by LSE in May 2020 found that respondents looking at a logarithmic graph held different attitudes and policy preferences towards the pandemic than those shown the same data on a linear graph. Logarithmic graphs display exponential growth using a scale that increases by a factor of ten on the vertical axis – a complex methodology that led sections of the public to underestimate the rate of infection.
“There are great analysts that want to do really intricate charts, but those charts have to tell you the required information straight away,” says Ming Tang, chief data and analytics officer at NHS England and NHS Improvement. “Otherwise, they could be misinterpreted.”
Indeed, poorly presented, or easily misinterpreted, data has fuelled societal divisions over the severity of the virus and the safety of vaccines.
The most intricate, data-filled graphs are trumped by basic pie charts when the latter is more readily understandable than the former. As the public’s interaction with data grows, it becomes increasingly important that data practitioners can weave a clear narrative around their analysis.
“The pandemic has taught us that it’s not just about numbers, it’s about narrative and storytelling,” says Tang. A strong narrative helps to ensure that there is little room for misinterpretation when data is accessed via services such as the Covid-19 daily dashboard. Narratives also help to support the government ministers, broadcasters and journalists relaying vital information on the news and in their reports.
The best example of this in action was during the 5pm pandemic briefings, which became something of a British cultural staple in 2020. The stars of the show were chief medical officer Chris Whitty and his deputy Jonathan Van Tam. “Having our brilliant CMOs join ministers during the briefings allowed for the accurate interpretation of data, both during their speeches and when responding to the media’s questions,” says Johanna Hutchinson, director of data and data science at the Joint Biosecurity Centre with the Department for Health and Social Care (DHSC). “Their work helped to increase confidence in the analytical work being done to shape decision making and allowed us to weave an accurate and useful narrative around the statistics.”
“The pandemic has taught us that it’s not just about numbers, it’s about narrative and storytelling.” – Ming Tang, CDAO at NHS England and NHS Improvement
Communications teams within the NHS, Public Health England and the DHSC worked closely with practitioners and the media to ensure that complex research was boiled down into insightful but widely understandable content. “Taking the time to explain some of the more complex models to the media was incredibly important,” says Hutchinson.
Take the R number, for example. The basic reproduction number is a statistical concept used in epidemiology to show the expected number of cases generated by one case in a sample population. Modelling the R number is a complex and intricate process which most of the general public would struggle to grasp if explained in its entirety. “The public doesn’t need to understand the ins and outs of the R number,” says Tang. “But they did need to understand that if it goes from .8 to 1.2 its bad.”
Instead of the logarithmic graphs typically used to display R rates within epidemiological circles, broadcasters including the BBC relied on simple line graphs. Some intricacies were lost, but the message landed. By pushing understanding of the R number out into public discourse, practitioners were able to clearly demonstrate the rate of the infection’s spread, in turn shaping an understanding of the factors influencing decision making.
This has been an iterative process for the teams involved. “The graphs produced by my team and across the wider PHE/NHS landscape have been through various iterations over the course of time,” says Hutchinson. “Developing the public’s understanding of the R number is the best demonstration of the power of graphical displays of data.”
A new appetite for data
We aren’t always going to be in a situation where conversations around “flattening the curve” are commonplace within WhatsApp group chats. The pandemic has been an exceptional period during which there has been a unique demand for real-time information. Practitioners hope that the peaked engagement and interest in data will be taken forward into the new normal.
“This is where we’ve always wanted to get to,” says Hutchinson. “Models are created and based on certain assumptions and parameters, and we’re now enabling a greater debate about the analysis that’s being done.” Indeed, social media was ablaze during the pandemic with discussion around statistical concepts, data points and interpretation of analysis. “Covid has enabled the public to interact in a wider sense not just with our analysis, but on wider issues led by government data,” says Hutchinson.
A persistent appetite for timely information would encourage the creation of platforms through which the public can engage with data. “It’s a bit like when you got your first smartphone, you were never going to go back to your Nokia,” says Tang.
“Covid has enabled the public to interact with our analysis and on wider issues led by government data.” – Johanna Hutchinson, director of data and data science at DHSC
Organisations such as NHSX are spearheading this movement by encouraging best practice for data sharing and transparency. OpenDataSavesLives is an organisation that supports the NHS in making more evidence-based decisions by sharing data and code. One of its four foundational pledges is to “encourage people to put their data in a safe place to learn from others health statistics and inform research.”
The public has to be encouraged to take ownership of its data if such initiatives are going to be successful. Models such as the Covid-19 public dashboard can only succeed when they are informed by large quantities of high-quality data, which in turn depends upon cooperation from the public. “The dashboard itself was designed to be a two-way discussion; relaying information provided by users back to those same users,” says Hutchinson. Iterative improvements can only be made when users are engaged.
The pandemic has had a significant impact here, as the public were able to see, in real time, the impact that their data was having. For example, when the Test and Trace app first launched, many members of the public were reluctant to download it due to privacy concerns. As the programme has matured, more data has been released into the public realm. Transparency has improved trust, and the application now serves as a high-quality data source that continues to guide the government’s approach to the crisis. Hutchinson hailed the publication of weekly Test and Trace statistics as one of her team’s great successes.
This is the relationship that practitioners are trying to set in stone as conditions stabilise: a reciprocal relationship, in which the public takes ownership of its data while reaping the rewards of better modelling and analysis. This, it is hoped, would dramatically improve literacy rates and deliver the “economic and public benefits” outlined within the National Data Strategy.
NHS England, NHS Improvement and Public Health England won in the Data for society category at the recent DataIQ Awards ceremony. Ming Tang was named Data and analytics leader of the year.