Six Chart Design Lessons from Visualizations of COVID-19
Important reminders for social scientists when creating visualizations
Adapted from a talk for the American Evaluation Association’s Data Visualization and Reporting Technical Interest Group
I remember the first COVID-19 chart that came across my news feed that prompted questions I couldn’t answer. At the time, the charts were double-digit case counts from Singapore with active, recovered, and fatal cases displayed on individual graphs — a simpler time than what we see on the tracker charts of cases and deaths from COVID-19, crowded with lines from dozens of countries and still trending up in December.
A month later, I still couldn’t get a clear definition for what qualified as a “recovered” case, even asking colleagues working at the CDC. The Morbidity and Mortality Weekly Reports (MMWRs) had started to contain early learning about the novel coronavirus and the growing case counts in the US. I felt alarmed when I saw that Tableau had launched a ‘ready to use’ workbook for anyone to jump start their analysis of the case data and wrote 10 Considerations Before You Create Another Chart about COVID-19, which still has relevant lessons months later.
Now, I’ve delivered nearly 50 talks, written articles, and given podcast interviews about the messiness and challenges of visualizing COVID-19 data, thanks to my intersecting public health and dataviz expertise. I collaborated with Tableau as a public health data advisor on the COVID-19 Resource Hub to share useful context and information for analysts from a public health perspective.
The US is still hungry for information, though I know many of us are just fatigued at this point. Maybe some are hoping for a glimmer of a number that will give some hope of a return closer to ‘normal’? But instead, we’re peaking in a third wave as I write this article.
What have we learned as the charts evolved?
Have we ever watched visualizations on a topic evolve in real time on a topic with such a personal impact to our lives? The only parallel I can think of is visualizing polls and elections, but seldom are those charts updated for six or more months with significant changes…