Blog

When it comes to visualizing data, the charts you choose to use matter . The right chart can be the difference between your audience understanding your data or being left more confused than before. One commonly-misused chart is the donut chart, which has gained popularity in recent years due to its aesthetic appeal. Here’s an example of a donut chart gone awry:

Awful! But where did it all go wrong? Well, this donut chart fails in multiple ways.

Part to whole

A donut chart should always add up to 100% and each segment should form part of the whole 100%. In this case, the segments add up to a whopping 230%, indicating a donut chart may not be the best fit for this data.

Bad use of color

Even if the story had been a part-to-whole, the use of color is atrocious. It can be tempting to use a lot of color to make a chart look more visually-appealing, but, in this case, the reader may not know where to focus their attention. In this case, if we wanted to highlight Williamson’s incredible contribution to his team, there is nothing separating him from the rest of the batters.

Accuracy with Area and Angles

Our donut chart relies on angles and area to convey information. This can be challenging to read accurately. To understand why this is difficult for humans to understand, we need to talk a little bit about the Cleveland-McGill scale.

Cleveland and McGill

In 1984, statisticians William Cleveland and Robert McGill conducted a comprehensive study to understand how humans perceive data encoded by charts. In this study, Cleveland and McGill measured how well participants interpreted types of charts by asking them to make “a quick visual judgment…” Based on the accuracy of those responses, Cleveland and McGill ranked the tasks from least accurately perceived to most accurately perceived, creating something called the Cleveland-McGill Scale. Let’s look at the Cleveland-McGill scale and how it can help us make better data visualization choices:

As you can see, Angles and Areas, or the task needed to interpret a donut chart, fall far down the accuracy scale. So, not only does our donut chart tell the wrong story, but it is also difficult to read!

Making over our donut chart

We need to use a chart that is comparable on a common scale, presents percentages without needing to add up to 100%, and leverages can color in an effective way.

Here’s our proposed makeover:

This is an instant improvement. With just one glance you instantly know everything the chart is trying to tell you.

When to use a pie or donut chart

Within the data visualization industry, there are mixed feelings as to the use of pie and donut charts due to where they sit on the Cleveland-McGill scale. At StoryIQ, we believe pie charts and donut charts are ok to use as long as they adhere to a few guiding principles:

• There should be no more than five segments in the chart. If you have more than five segments, think about grouping the lesser segments into an “other” category.
• Segments should be ordered in ascending or descending order (either clockwise or counter clockwise are fine, as long as its consistent within your presentation).
• Since segments of similar value are difficult to differentiate, the segments should be meaningfully different.