The Data are Alright… and the puns are … intended.
(Note: yesterday was March 14, or 3/14, otherwise known as Pi Day!)
I wasn’t able to blog yesterday (or even eat pie!) because I was hosting my first ever webinar with the brilliant folks at the Working Families Party. (Shout-out to new readers!) but I loved seeing all of the posts on social media honoring the sanctity of pi(e) in all of its forms. With that out of the way, I feel compelled to take a position on a HIGHLY controversial issue in the data visualization blog-o-sphere: namely, pie charts.
You see, pie charts are maligned by statisticians the world over. Some people even say pie charts are the “comic sans” of data viz! Yikes! The trolling of pie charts is intense & this treatise and this blog post are particularly emphatic. The earliest criticism I found goes back to Charts and Graphs, first published in 1923 by economist Karl G. Karsten.
Why do critics think pie charts are “as bad as Nickelback”?
- It is hard for our brains to interpret and compare proportions
- Too many colors are confusing (or worse, misleading)
- Pie charts use a lot of room on the page
- Your eyes do extra work jumping from the legend to the chart
- Depending on the order of your segments, readers will draw different conclusions
- 3D / foreshortened pie charts are EXTRA bad. Avoid at all costs.
- Hating pie charts is a fad; critics want to be cool.
The case for pie charts
Maybe you’re thinking I should finally get a-round (hehe) to making my point about pie charts. I don’t blame you. I think pie charts are a good tool for changemakers. Please don’t turn me into mincemeat!
Data viz experts recommend using pie charts under two conditions:
- The slices sum to a meaningful whole (h/t Randal Olson)
- There are 3 or fewer “slices” of the pie
In at two party system like we have, for better (or more accurately…) for worse, general election data can work really well as pie chart (this is less true for primaries where there are a lot of candidates in the field). The data should always sum up to 100% of the votes cast. Most of the votes are cast for 2 (or 3) candidates. Voila! An excellent use case for pie charts!
Moreover, pie charts are a simple way to communicate meaning without having to put as much time into the X and Y axis of a bar/line graph. Sure, horizontal bar graphs might be easier on the eyes, and vertical bar graphs are great at showing comparisons over time, but sometimes, it’s better not to re-invent the “wheel.” If a pie chart can get the job done, I say, get the job done and go back to basebuilding!
Finally, for many of us, pie charts are in our comfort zone (if not, check out this tutorial!). On one hand, I do want to push us to try more innovative data viz techniques. On the other hand, I think there’s nothing wrong with using the skills we already have. I firmly believe that data and spreadsheets are for everyone! There’s no benefit to being stubborn about pie charts unless they are (unintentionally) confusing or misleading. If you use the two tips mentioned above, you’ll be well on your way to making pie charts that tell a clear story.
I might have to eat my words, but for now, I’m pro-pie. And that’s all there is to it!
Here are some goofy pie charts for your entertainment