It's fun to make fun of things online.
We all do it to some capacity, and it's an easy way to win social points. Twitter, in particular, is a finely tuned tool for rewarding division. The most opinionated, controversial, or divisive comments are the ones that draw the eyes, while more moderate, nuanced, and thoughtful posts tend to drift away into oblivion. I say this to say that earlier today DataViz royalty Edward Tufte posted about the uselessness of pie charts and donut charts. Now, I'm not saying that pie charts and donut charts are the be-all-end-all of parts-to-a-whole reporting. However, framing them as charts meant for airheads is also missing the point a bit.
First off, let me come to Tufte’s defense. I can see where he's coming from.
Pie charts and donut charts are often used very poorly.
If you've been in the business world, much less the data visualization game for any period of time, you have no doubt seen dozens of poor examples of pie charts. Charts with 50 wedges or more, a thousand labels, or no demonstrable value.
This example shows sales (via the SuperStore dataset) by State. Even without numerical values spelled out, you can see the advantages of bars over the pie. Individual states are easier to compare (Oh, Texas is less than half of California), and you have a more complete picture without the smallest values disappearing entirely.
It doesn’t work because this isn’t a use case for a Pie Chart.
The weaknesses of pie charts are many.
In the same amount of space that it takes to show a pie chart. You can easily have a horizontal bar chart that is not only easier for a reader to tell the difference in sizes between the bar lengths but does not require a slew of labels radiating around the chart or an extensive key in order to pull any value.
But, in this weakness, we are actually seeing the strength of the pie chart.
Pie charts are highly effective at showing outsized populations.
Revisiting our sales pie from above see that California and New York account for more than 25% of all sales. Ant the top 5 states (CA, NY, TX, WA, PA) account for half of all sales.
That’s a valuable insight.
Though this chart as a while may not be the “best” option, we learned something from it.
This pie chart of the lyrics to “Baby Shark” effectively shows us that more than half of the lyrics in the song are the “Doo Doo” chorus that your kids won’t stop singing.
What about this pie showing the Net Worth of Africa?
We could display all continents together, showing the relative net worths, but that often blows out very small values (as we saw with our Sales in the SuperStore example). By showing ONLY the portion of Net Worth of African Nations vs the rest of the world, the contrast is stark and profound.
What’s the Point?
Pie Charts are very useful tools depending on WHAT you’re trying to show. There is rarely a one-size-fits-all answer.
That’s why analysts exist. We apply expertise in terms of understanding our audience, the business, and the application of visual tools to help lead our audience to a greater understanding of their data.
This means “always” and “never” rules sometimes need to be disregarded in favor of novel or non-best practice approaches.
And while a pie chart often isn’t the best answer, it sometimes is.
I also like the way that multiple pies can be sized (by area) to show the totals while the wedges show a breakdown. This can be done with stacked bars too, but the good thing about pies is the areas can be compared in multiple directions - such as on a map, or in a matrix.
Amen to all of this! I especially love the Africa viz. It's a very powerful use of a pie chart for a very serious topic.