A Pie Chart Makeover
I recently saw two pie charts and a table that used colour to encode each item. I replicated what I saw with some dummy data below.
I basically compare the type of clothes and accessories I had in 2007 and 2017. The pie charts show the percentages of what I owned in the given year. Now I really hope you are feeling uncomfortable looking at this. I really hope you feel a bit anxious and a bit upset even. Why? Because it is only natural. If you aren’t, there might be some mutation in your genes.
I don’t know about you, but I really, REALLY struggled when I was trying to figure out what was what in the pie chart. Let’s look at 2007. There’s four shades of blue here and trying to figure out what they were by referencing the table on the right is a bit of a hassle. Not to add that this is in no way accessible for people with colour-vision deficiencies. Quite frankly, this viz is just a paradox.
Graphs, bar charts, plots, maps, pies, these were all meant to make it easier to communicate information. But communication is not as easy as it looks. Let’s explore this with a model thoroughly used in A Primer on Communication by Ray and Charles Eames. I highly recommend this video, if you are short on time, at least watch the first couple of minutes.
What is Communication?
This duo is no stranger in the world of design, and certainly not in the art of communication. In the video still above, it explains how information is communicated. Progressing from left to right, here is what happens:
- There is an Information Source
- A Message is selected from this source
- A Transmitter changes this message into a Signal
- This Signal travels through a Channel
- The Signal is received by the Receiver
- The Receiver decodes this Signal into a message
- And this message arrives at the Destination
That’s quite a mouthful, but it is a model that explains how we communicate. In this video, they gave an example of saying “I Love You” to someone. Let’s try this model with an example between the couple Josh and Harry:
- Josh’s brain (Information Source)
- The thought Josh loves Harry (Message)
- John encodes this thought into the words “I love you”. Your mouth and vocal chords form and sound these words which is transmitted via sound
- Sound travel as vibrations through the Channel of air
- The sound of the words arrive at Harry’s ears
- Harry’s brain decodes this message
- Harry receives the feeling Josh wanted to convey (Destination)
Looking at all these steps, a lot goes on when we talk to each other, and it all happens at a rate where we don’t notice it.
Let’s take this one step further and apply it to the viz I saw recently.
Here, the information is the original set of data not shown here. From the data set, we extract the relevant information such as all the clothes I owned in 2007 and 2017. The message we want to deliver is the change between 2007 and 2017. This message is encoded into two things here, the first being the table on the right and the second are the pie charts. The communication channel is the screen which is transmitted through air as light. This signal arrives at your eyes and is decoded into your brain. And that, is how you are seeing this viz. As awesome as the process of seeing is, people will still struggle to understand the pie charts. Why is this?
There is one last thing I did not mention, which was crucial in the video, and that is Noise. Noise is essentially what distorts the message. When you say you love someone, many things can distort that message for the receiver. The receiver might be insecure and somehow receive the message that you hate them and not love them. In the case of this pie chart, the noise is mainly colour. If your screen is distorted or has low resolution, that is also considered as noise, which can disrupt the message. Colour in this case is noise because it has created an extra step in encoding information. When you look at the the dark orange, you have to take an extra step and look at the legend to understand what it represents. Generally, using a pie chart like this with colours may do OK if it had less than 7 items. But this one has over 10 and that is too many to comprehend.
As a result, I decided to figure out the best way to visualize this. To be fair, the viz I saw looked like it was made using Microsoft PowerPoint. Let’s keep this all at the same playing field and do my remakes in PowerPoint as well. Yes, I am not a big fan too, but let’s just play along here.
Let’s Not Use Colour
Colour is oftentimes dismissed as an aesthetic and not much beyond that by the general public. This is true, but not for information design. Generally if there is more than one colour, then it becomes a variable that has meaning. If colour was the issue with these pies, then it might be a good start to remove them. So I tried to use pattern fills. This achieves two things: 1) It is easier to decipher the legend through pattern than colour since the patterns differ in shape, size, and orientation; 2) This is also accessible for people with colour-vision deficiencies.
Still, I think we can make do better.
Get Rid of Crazy Legends
The legend has 18 items, which is practically…legendary…Sorry, that joke just had to happen. Anyways, back on track. The best way to work with legends is to group items into broader buckets. So if I was to stick to using one, I could just sort things by the Type, which are “Bottom”, “Top”, and “Accessory”. This would be easier to decipher and easier to read. But let’s imagine that this viz just had to visualize each item individually, which was 18 in total. In this case, the legend should just go. Bye legend!
So this looks a bit more relaxed. It’s not as crazy and it looks more sane. Data labels were added so you knew what each pie piece was representing. But this is still not easy to read.
2D Areas Aren’t Preferable
Steven Few wrote in his book “Show Me the Numbers” about how humans have a pretty tough time perceiving pie charts. Pie charts use 2D area to represent quantity, and he frankly puts it that he avoids using pie charts altogether. I can see the argument for this because you would have a easier time comparing the lengths of two lines than the areas of two circles. You can see that one line is 2x the size of the other line, but that is not the case for circles. So I thought to ditch the pie chart altogether and go with a stacked bar chart.
Looking at this, there are still barriers to accessing the information. With the limits of PowerPoint, I can’t easily move the text labels on the bar chart in a way that is easy to read.
How About Column Charts?
With this final iteration, there is no need to worry about legends, colour, or text placement. It is not ideal, but this one is certainly a lot easier to read than the pie chart that looked like it had Skittles for lunch.
I don’t mind pie charts if they are used properly. My rule of thumb is if it has about 4–5 pieces, then it is easy to read and it will do the job. But if it has over 10, it’s time to re-think the format.
======
I am aspiring to be an information designer. With no formal background in this field, I am equipped with a combination of skills from design strategy and life sciences. I hope you enjoyed this article and found value in it. Feedback is always welcome.
