User-Centered Data Visualization. Part 5 – From Wrong to Right

Here’s a cool data viz story. A large weekly magazine in Europe has recently published some statistics about Ebola cases in four African countries. It’s in German, but bear with me here.

 

Per country, the outer bubble depicts the number of Ebola cases and the inner bubble shows how many of those have died.

If you read the other parts of this blog series about user-centered data visualization, you’ll recall the reasons why using the size of bubbles to convey numbers is not a good idea. We’re just not wired to estimate and compare areas well.

When I tried to replicate this visualization I found that the size of the inner bubbles (representing the number of people that died) is incorrect. They’re too large. But even without that, using the bubbles in bubbles leads to something that I find alarmist. You look at those bubbles and you see that the inner bubbles are almost the same size than the outer bubbles. Consequently, you get the sense that most people who got infected also died. Ebola is deadly and currently no vaccination is available to the public. However, based on just these numbers, it’s not the case that almost everybody has died.

Here’s a visualization that shows the percentage of people who died vs. those who have not died. For the purists out there who hate pie charts, only look to the right! It shows that for the two countries on top (Liberia and Sierra Leone) most people have not died (sadly, I have to add “at the point of assessing the numbers”).

 

I wrote an email to the magazine who forwarded it to their graphic designers. After some praise for their magazine (which I’ve been reading for a long time) I provided them the reasons why bubble charts don’t work well, showed them the pie chart/bar chart above and humbly suggested to use pie or bar charts to show that kind of data.

I didn’t hear back from them, but two issues later, they published an update on the situation for those four countries.

 

They did use bar charts and these convey the situation much better. It’s much easier to estimate the length of these bars than the size of the bubbles. I believe that people oftentimes shy away from using the classic and established chart types because they fear they look boring and non-sexy. From a viewpoint of a magazine with a circulation of 900,000 copies a week, I could understand that. Yet, they were open to make changes and looking at this new chart, I don’t think there’s an issue with attractiveness (probably should mention that the red, black and white are their three brand colors). 

So let’s heal the world of data visualization one chart at a time! More importantly, let’s hope that an effective vaccination is made available to those who need it soon.