Newly-published research backs up our assertions that donut pie charts work fine, that square pie charts work great, and that visual weight matters more than you think.
Robert Kosara, a senior research scientist at Tableau Software, recently posted new research to his YouTube channel. If you’re a data geek, it’s well worth 10 minutes of your time. If you’re not, the key takeaway is that people read by pie charts by area or arc length, not by the angle of each slice.
What does that mean?
1. You should make square pie charts. Kosara doesn’t make this claim in the video; he just says that people are more likely to read pie charts by area than by angle. But if you don’t need to show people a pie slice angle, why wouldn’t you use a format that better activates people’s brains, improves comprehension, and offers more scope for detail?
2. Those Infogram pie charts are as dastardly as we thought. Remember that awful Infogram pie chart in which the visual weight of the largest slice was 3.5x bigger than it should have been? Our complaint isn’t just theoretical: Kosara’s study shows that this type of chart really does introduce significant bias, as shown below.
Chart showing pie charts that artificially enlarge one slice introduce significant bias. Source: eagereyes YouTube channel.
Again, I recommend you watch the video. But even if you don’t, rest assured that donut pie charts work fine and square pie charts work great.
Want good data delivered to your inbox? Subscribe here.