Pie Charts are Off the Menu!
Pie charts are commonly used in business and data journalism, but are they ever the right choice for your data visualization?
Why you shouldn’t use pie charts and how to replace them
Nowadays, pie charts are commonly used in business and data journalism. The message conveyed by pie charts is nearly universally understood, making it tempting to use them in your own data visualisation. While the message that pie charts convey (parts of a whole) is easily understood, the insights they provide are not.
My advice; try not to use charts named after foods. So, if you want to create superior reports, dashboards, and presentations, one easy step is to stop using pie charts. Here’s why.
Exactly what is the purpose of using visual representations like graphs and charts?
Including visual aids like charts and graphs in reports and presentations helps make the material more digestible.
Data visualisations like charts and graphs are meant to simplify the information they present without reducing its quality in any way.
What's wrong with pie charts?
Typically, pie charts are used to illustrate the relative importance of various components. However, humans are notoriously bad at comparing areas (or angles), which is the main requirement of pie charts, so they are often misleading. It is human nature to exaggerate obtuse angles and minimise acute ones.
If you need an alternative to pie charts, bar graphs are usually your best bet.
To better compare data relationships, bar charts are recommended. That's because we can easily make direct comparisons between items of varying shapes and sizes using just one dimension. The bar comparison is much more straightforward than the area comparison of pie slices because humans are much better at comparing objects along a single dimension than along two. It’s just simpler.
It's true that pie charts are more aesthetically pleasing than other types of graphs. The problem is that they aren't very good at what they do.
Data presented in a table is much more readable than when it was scattered inconveniently around the edge of the pie. If that's the case, then what's the point of a graph? Why present a picture of the data if it cannot be understood and does not enhance the presentation of the data in any way? The answer is: You shouldn’t. When a visual representation of the data reveals interesting connections (patterns, trends, and outliers) that would be difficult to infer from the tabular data alone, a graph has served its purpose.
Many other academics share the opinion that pie charts are bad. William Cleveland wasn't the first to criticise pie charts, but his 1984 scientific paper found that people were 1.96 times more likely to draw the correct conclusion when presented with data in the form of a bar chart rather than a pie chart. That's because it's simpler to compare lengths than angles.
When to use pie charts
If you're trying to visualise more than two options, like yes/no, most data scientists will tell you to avoid using pie charts. Still, you'd need to add numerical labels; otherwise, the data won't be easily interpreted.
Even so, from a purely personal standpoint, I might still consider employing bar charts in this case. For me, it’s just a matter of best practices.
Not ready to give up on pie?
Understandably, your consumers may not care about the complexity of pie charts, they may want one anyway. Since they are paying you, you have to deliver it.
Do this instead:
If you insist on using them anyway, order the slices by size to make it easier to read, and limit their use to percentage breakdowns in which each slice represents a certain percentage out of 100%. Avoid using a pie chart with more than five sections, and never make it three-dimensional.
But please promise me that you will first try a bar graph and compare it to a pie chart to see which one is simpler to read and understand, and only then will you ever go ahead and add a pie chart to any report or presentation.