Gun violence in South Africa – what’s wrong with this chart?

Eyewitness News,

I saw this infographic on the multimedia section of the Eyewitness News website during the early stage of the Oscar trial.

One of the charts, MOST GUNSHOT VICTIMS – MEN vs. WOMEN, inflates the number of women as gunshot victims in the way the data has been presented. Perhaps it was because of the focus on Intimate Partner Violence (IPV) and the abuse of women in our society highlighted by the trial. Seeing less women affected by gun violence would probably not have made the same impact. Or maybe it was a mistake.

The real purpose of quantitative displays is to provide the reader with important, meaningful, and useful insight, according to information visualisation expert Stephen Few. Here we have a combination of two chart types, pie chart and unit chart; both have data visualisation design flaws and the two sets of data don’t seem to match.

Pie charts

Pie charts are too easy to create and fail at trying to present useful data, which its creators don’t realise.

Start quote

But it is almost always difficult to compare the slices of a pie, because visual perception supports only rough comparisons of areas and angles, which are the two primary ways that pie charts encode quantities.
Stephen Few, Our Irresistible Fascination with All Things Circular (PDF)

End quote

With only two data sets, a pie chart is appropriate here because the values are quite disproportionate and easy to grasp.

Unit charts

A unit chart uses geometric shapes or icons to represent the quantity of things being illustrated – often seen in magazines and newspapers. Here the familiar person icon shows us that people are the subject of the data – often referred to as a pictorial unit chart. The problem with unit charts is that you can’t figure out the values without counting the icons – a slower cognitive process than simply having to read the percentages – and for some reason the graphic designers decided to include both chart types.

Gunshot victims pie chart, 11 women at 11 percente versus 15 men at 89 percent
Gunshot victims chart – 11 women vs. 15 men
Gunshot victims pie chart, 1.85 women versus 15 men, which is more realistic
Gunshot victims chart – 1.85 women vs. 15 men

The data in this chart is incorrectly represented

In this chart, 26 people represent the population of men and women affected by gun violence – I counted them. But when you try to calculate it, the numbers don’t add up. That would give us a 42/58 percent split – not what the chart shows, which we must assume is correct.

If I take 15 men representing the 89% of the population, then one man represents 5.933%. By that same math logic, 11% of women would be represented by 1.85 woman icons. It would be silly to have 85% of a person illustrated, but that would still be more accurate than the infographic that the infographic represents.

My recommendation would have been to leave out the people icons or recalculate the numbers so that they match whole icons – like 2/16.