During the coronavirus outbreak, should we change the data visualization rules?

Alice Casey raises an interesting point about a chart about coronavirus (“What are the chances of dying? from the BBC):

What’s the problem? It’s that the bar extends all the way to the right. We perceive this as a “maximum” value. Our eyes stretch to the right, and the risk is we perceive that to mean 100%. The implication being that 100% of 80yr olds who get the disease will die. That is enough to worry anyone.

We perceive a bar that is full length to be max value. In this case, that might be 100%

Of course, the information is on the chart – down in the bottom right. The 15% label is clear:

The max length is only 15%

Is this a mistake? In most cirumstances, it’s not. When we’re communicating data, we’re normally trying to make it easy to compare one mark against another. By truncating the axis to 15%, the reader can easily compare one category against another.

However, at this time, our responsibility is to communicate all the data. Here are 2 ideas. On the left, I’ve labelled the bars: people will see the values as they see the bars. On the right is, I suspect, the chart that Alice wanted to see. I also changed the title to be a question.

Two makeovers

I think both of these are improvements, but what we need is to add the missing data: those who survived. Furthermore, where does this death rate come from? Is it consistent around the world? The numbers refer to 44,000 coronavirus cases in China, as published in a report by the Chinese Centre for Disease Control and Prevention. The death rate in China isn’t necessarily the same around the world. Digging into table 1, the death rate seems to be going down, too. So a chart like this might not apply in your country.

With this in mind, my final version is below. It shows survivors and has a more descriptive title:

A makeover showing survivors as well as those who died.

I much prefer this version. The blue bars are concerning, especially for the elderly or those with some health conditions. But the grey bars are, in all cases, far bigger. This visual clue shows a reality of coronavirus: it’s a very serious disease, worthy of a major global response, but not as dangerous as the original might have led us to believe.

EDIT: Chris Love tweeted that perhaps you could visualize the survival rate more prominently than the death rate:

That’s a great idea – this way, the data remains accurate, but greatly reduces the chance people could be alarmed. Jorge Camoes and Alberto Cairo also tweeted good arguments for the original chart.

What do you think? Is my makeover less alarmist?


Add Yours →

Could the health conditions section also be alarmist for a younger person with asthma? It suggests a person of any age has a 6% chance of dying but i am assuming that, in reality this is probably weighted someway to older people.

Definately….my husband and I are healthy ‘young’ 80 yr olds and original graph was frightening…

Your chart is definitely better; looking at the original, I was confused by how both male and female could total less than 5% but your expanded title makes it perfectly clear.

Much less alarmist, thank you! Especially with Chris Loves flavor with survival rates. It’s equally accurate – and even less alarmist for the majority of people who would view the vizn – to show the age cohorts from youngest to oldest. this emphasizes the younger cohorts that have tremendously high survival rates.

Andy- thank you for this lucid representation of a very contentious topic. Responsibly illustrating the impact of this outbreak with relevant context that informs tradeoffs in treatment & policies in mitigating risk appears to have been brushed aside in favor of (1) non-normalized counts (of cases, tested, deaths), akin to alarmist boxscores, and (2) universal calls for curve-flattening via social controls, a blunt instrument that has its own dire consequences.
C.Love’s ‘survival chart’ is rich on so many levels. It effectively visualises an inter-generational civil war of sorts, pitting the interests of the Greys & greying, i.e. the old & infirm (the old crown?), against the Blues, a hale, robust & much younger cohort which is at much less risk of dying (from the ‘new crown’ virus). Such a survival chart might support a debate on whether the prevailing ‘treatment’ via state-mandated shutdowns, encroachments & societal stasis is truly worse than the disease which itself appears quite survivable.
Are you aware of any survival charts for other viruses or behaviors to add context? And might you speculate as to why charts such as yours and C.Love’s are not getting more attention? [I’ll also try to connect via Twitter.]

I think this is a great way to show this data in a more rational light. Next, I’d love to see this chart put into perspective with data sets from other viruses. This version feels less alarmist, but I suspect comparing the mortality rate for older or health-compromised individuals with that of a common flu would add a layer of understanding of how this is different from other familiar viruses.

Leave a Reply