I suspect that “pie charts suck” is the data visualization equivalent of “Hello World”. It’s the first step towards enlightenment, the moment when people become aware that data visualization is more than picking a chart from a library and a few random cool and colorful effects. We should recognize the importance of pie charts in this rite of passage.
Hopefully, at some point most people will come full circle (pun intended) and realize that pie charts don’t actually suck that much, and that there is a place for them in some cases.
Between those two moments in time, pie charts enjoy a really bad reputation. A continuous flow of bad examples makes sure that this reputation is well-deserved. But why? My current answer: yes, there are a few issues with this chart type, but often how people design it, how they manage the data and how they fail to craft their message are the main culprits.
Litmus test for pie charts: slices in a pie chart must total 100% (OK, between 99.9% and 101.1% to account for rounding), and that’s not negotiable. All charts that fail this are either wrong, humorous, silly, ignorant or purposefully manipulative. Also, your data must allow for some type of aggregation: if the total doesn’t make sense then you can’t use pie charts (for example, you can’t sum unemployment rates). I will not discuss here charts that fail these two basic rules.
I hope you like my little pie chart above. I was able to squeeze into it the most common errors, among them:
There are many issues with this chart, but we have 28 slices, so there isn’t much we can do about it, right? Not really. Check the next lesson for alternatives.
P.S. Not exemplified here but also important: don’t compare pie charts, because the concept of part-of-a-whole doesn’t extend beyond the single whole. When you have two or more charts and want to compare slice A in each of them you’re no longer comparing to a whole, so you should use other chart types.