2 April 2020
Data visualisation is a powerful tool. It has the ability to inform, persuade and steer people to take specific actions. And with great power, comes great responsibility.
The aim of dataviz should be to guide end-users to take the right action, rather than being purely evocative.
Communicating information that directly affects the well-being of the world’s populace is cause to produce dataviz responsibly. More so in times of globally pervasive uncertainty and fear.
Under the current conditions, it’s imperative that dataviz conveys facts and expresses those facts accurately and in context.
In a crisis, people are grappling for answers and clear solutions. And in our digital age, it’s particularly easy to publish the ‘answers’. But perhaps too easy.
COVID-19 is a phenomenon that can potentially affect every continent, country, town, and individual. As a result, developers, now more than ever, must produce responsible dataviz.
Dashboard developers should aim to support this health crisis by providing accurate data. If not, the data message could spread false information and contribute to a state of panic rather than constructive action.
To weigh in on this, let’s introduce Megan Danielson.
Megan takes this premise further by adding, “It isn’t just about using accurate data, but about using accurate data appropriately. Selecting the wrong visual can distort and misrepresent otherwise accurate data.”
Image credit: Mapbox. The choropleth map showing raw counts per country overemphasizes areas, like China, with large populations and large total area. (Click on the image to view the full-size map in their blog)
And here is how the above data set should have been mapped:
Image credit: Mapbox. The choropleth map showing raw counts divided by population highlights the country with the most widespread impact on its population — in this case, it’s Iceland. (Click on the image to view the full-size map in their blog)
If you take a look at the vast number of dashboards popping up on official and social platforms, they do provide information on COVID-19. Yet, many fall short of being both responsible and informative.
Let’s review a few best practices for creating effective COVID-19 data visualisation.
We’ll begin with practical pointers from Megan.
1. Outline your objectives and your target audience before you start developing. Answering questions such as: ‘What information do I want to convey to my audience with this data set?” at the onset will help narrow your design focus. Remember, keep it simple when you can!
2. Don’t throw the kitchen sink at the problem. Too often, people add extraneous information or interactions into their visualisations that don’t need to be there. Overwhelming your visualisation by adding too many layers, too many interactions, or too many colours can detract from your main objective.
3. Reduce the alarmist nature of your visualisations by choosing the right colours, for instance.
Image credit: Mapbox. Red is alarming and fear-inducing for readers. (Click on the image to view the full-size map in their blog)
4. Mapbox also recommends choosing the right dataviz medium for the data you have. Your data might be better suited to a chart rather than a map.
You can find more practical steps from the Mapbox team in their article: best practices for mapping a pandemic.
Those are great suggestions. And here are two additional tips to help you tune up your COVID-19 dashboards.
5. Show data truthfully. To accomplish this, you’ll need to have a sound understanding of the numbers and not just copy and paste data sets.
6. Keep the message in context. Without context, data can be ambiguous and confusing. Not something you want in a time of crisis. Instead, consider:
The core elements of a data story
And, to elaborate on the importance of data context, we have input from our next collaborator, Joe Warbington.
Joe added that the general population is far more aware of visual elements like line charts and KPIs with the constant updates flooding the news and social media. “It’s easy to show spikes and draw attention to BIG RED LETTERS to draw in your audience and show trends,” he continued.
He also highlighted a key point: The focus should be on the question: “So what?” What can you do with the information? How can you take action on it? It’s always about context, he reiterated. And what you want the person consuming the data visualisation to do with it. Yes, you have to attract their attention first with the visuals, but then you have to provide the context to get the brain processes flowing, he advised.
And, context applies to all varieties of dataviz. Now we’ll take a look at tips pertaining to mapping data.
“Mapping has been the focal point for tracking and helping resolve epidemics since John Snow helped resolve the Cholera outbreak in London back in 1854. They are a sexy visualization for sure. One that piques a reader’s curiosity and engages them if the area in which ‘they live’ stands out from another area. Because at that point it’s ‘important’ and not ‘just data.’”
Dalton Ruer, Qlik.
Here are five expert mapping tips from our final collaborator, Dalton Ruer, to guide your COVID-19 map dev process:
1. Don’t just visualise a single number, show its relationship to other data so that it has context. For instance, in the case of COVID-19 visualisations, why show the Number of Cases when you can show the Number of Cases per million people? Forming a contextual analysis will draw more readers in and provides a working number everyone can relate to.
2. Don’t be afraid to tell a story, to drive action. We pay big money to see a movie and have a bucket of popcorn and drink a soda because humans are suckers for a good story. So, don’t just plop a map on the screen and expect people to jump out of their seats.
3. Avoid jargon. Your labels and explanations should be in a language that can be consumed by your intended audience. You’ve created a masterpiece, let it be enjoyed, not feared.
4. If one map is good, two maps are better. Or rather, a bi-variate map is better because it can show the contextual relationship visually between different data sets. One layer of the map could be the number of cases per million in a county, and the other colour layer could be the number of hospital beds available.
Dalton’s bivariate map depicting County Health Rankings in one colour scale at the county level and the CDC Social Vulnerability Index at the census tract level in another colour scale
5. Add a personal touch. Put yourself into what you are visualising through the maps you create. I love to integrate social determinants of health data. Perhaps you work in shipping and have access to data on shipping schedules, and could visualise where toilet paper will be available day by day! Ok, that’s silly, but you get the point.
Dalton makes more constructive points in his video on mapping COVID-19, if you’re keen to get more information.
There are new COVID-19 dashboards published every day, many focusing on the growing number of recorded cases. And more data is useful, right? That may be true, but not everything that’s published is communicating the data in a suitable way.
Joe worked with the Qlik team to create a COVID-19 group on the Qlik Community, a platform aimed at sharing best practices, inspiring members with new possibilities for visualising data in Qlik Sense, and providing support for creating responsible data messaging.
Many dashboards were built upon the well-known Johns Hopkins data sets to give additional functionality or blended with other data sources, like weather, stock data, flight data, and even social determinants of health data, Joe explained.
COVID-19 Vizlib Heatmap showing new confirmed cases – Robert Mika
One significant outcome, he added, was that healthcare organisations started collaborating on system-level and local-level applications.
“These apps harness internal clinical, staffing, and supply chain data to give better and faster insights into what they are facing on the front lines of the pandemic fight. Often these apps have the high-level KPIs you’d expect, but the more functional ones bring in visualizations that help with operational needs, such as capacity planning or telehealth outreach,” Joe voiced.
This platform, and other beneficial COVID-19 data analytics initiatives, where the right stakeholders are working closely together, is a true reflection of the power of data collaboration!
It’s never easy going it alone. And where would we be if we couldn’t learn from others in the BI field? Here, Joe urges us in the right direction. He asserts that for visualisation apps to bring out the story in the data and empower people to take action, the most technical savvy developers in the world cannot do it alone.
He urges developers and designers to work collaboratively with subject-matter experts (the end-users) to build impactful analytics. “Though this is necessary across all sectors and functions, we’re seeing it highlighted for healthcare providers and staff that must be able to consume information and act on it quickly,” he prompted.
If you’re looking for practical resources to help you simplify your data messaging and get the right, responsible data message across to end-users, here’s a great place to begin:
We hope this helps you to create COVID-19 or other crisis-related dataviz that prompts the right action, rather than adverse reactions. Now, go use your powers for good!
Thank you, Megan, Joe and Dalton for your valuable insights and contributions!
Need to supercharge your Qlik Sense apps? Try Vizlib today!
Check your email
We've sent you an email to with all the information to get you started.