Topic

How to choose a chart type

Topic Progress:

Click on the numbered tabs to navigate this lesson.

When creating a data-visualisation, the reader’s interpretation may vary drastically based on your choice of chart. It may seem obvious, but different types of charts are better suited for different types of things. 

For example, if you are trying to show the changing value of a commodity in dollars over time, you wouldn’t try to use a map. You would likely choose a line chart that can be read as the story of how prices move month by month.

Generally speaking, your aim is to make charts as easy to understand for the reader as possible, and help guide them to the insight you want to show. Including just one or two variables is ideal – the more complex a chart, and the more data you try to put into it, the harder it will be to read.

The chart above shows the population of Nigeria by state. Can you tell which the most populous state is at a glance, or whether there are more people in Ondo than Gombe? This is not a good example of clear communication.

The choices we make matter

Here’s exactly the same data, visualised using a different chart type. We’ve used a couple of highlight colours to pull out some details – and you won’t need to think for more than a few seconds to work out what the highlights mean, even without any annotations.

This chart is a lot easier to read and much more useful for a reader interested in the relative size of populations in Nigeria.

It’s not ideal, however, as there are so many columns on the x-axis the text is very hard to read. 

Let’s try again…

Making things simple

Better? We’ve even included a title and the source this time, so pretty much all the information you need is there.

 

How do we choose a chart type?

When we create a data visualisation, we are creating a perceptual task for the reader who wants to understand it. How humans complete these perceptual tasks was described in a famous paper by William Cleveland and Robert McGill in the early 1980s, and is often cited today. 

By observing how students were able to interpret data that had been visualised, Cleveland and McGill came up with a scale for how the human eye and brain work to decipher information that is presented graphically. 

The higher up the scale a chart type is, the more accurately readers can distinguish between small changes in numbers. Column, bar and line charts all make this task very easy.

The lower down you go, the harder it is to make accurate distinctions – but you may choose to use the techniques described because you want to portray general observations about a population quickly and intuitively.

The Cleveland McGill scale 

By Alberto Cairo

Chart choice in practice

Bar and column charts are useful for when you are showing distinct differences within a population. Line charts show a single variable as it changes (usually over time). 

Shading and saturation are often used on maps, when you the reader needs to know obvious differences but not the details. Like where have the most cases of Covid-19 been reported in Gauteng Province, South Africa?

The Data Visualisation Catalogue

Understanding which chart is appropriate for the story you want to tell can take experience. The Data Visualisation Catalogue is an excellent online resource, which can guide you to a good choice based on what you want to visualise, and includes extra information about each chart type.

Adding complexity

Most of the time, you will want to create simple line graphs or column or bar charts in order to present data as clearly as possible. It is possible to create charts with a lot of complexity that create a compelling narrative, however. 

The chord diagram above was created by Quartz to illustrate net migration flows around the world. Not only is it visually compelling, it tells powerful stories about where the world’s largest population movements occur – which ones run counter to common narratives?