Data Visualisation

Thoughts on Colour

I never even considered this until the last year or so but I stumbled across a website talking about graphs and colour blindness. It turns out that roughly 1 in 10 (~8%) males suffer from some form of red-green colour blindness (Deuteranope or Protanope). In women it’s somewhat less than 1% (wow – is there anything they don’t do better than us?).

It rarely occurs to most of us to allow for this when presenting our graphs but statistically, we probably should. People with this affliction struggle to differentiate red/green contrasts and the chances are in any non-trivial gathering there’ll be someone who can’t work out what the hell you’re talking about.

Below is a sample of a pretty standard dashboard created in Tableau:


And here is a side by side comparison showing how it would appear to a colour blind person (there’s a great simulator here for creating these):


It’s pretty obvious that understanding the bar chart on the right is going to be hard to appreciate if you struggle differentiating between reds and greens.

Fortunately tools such as Tableau have built in support for colour palettes that are designed to highlight contrast to address this, however you can implement this yourself in your tool of choice – for example in R you just specify an appropriate colour vector.

By applying Tableau’s “Color Blind 10” palette to the bar chart above we get the following:


And this is how it looks if you’re colour blind:


In summary, data visualisation is all about sending a simple clear concise message to people. But people are complex and on any given day may be tired, distracted, hung-over, bored or simply colour blind! Make your graphs easy for everyone to see and it’ll be easier to tell your story.

So as tempting as it is to have awesome looking rainbow colour spectrums, in many cases the message is lost – it’s up to professional practitioners to be aware of this.

Categories: Data Visualisation

Tagged as: , , ,

Leave a Reply