Pie Chart vs. Donut Chart: Showdown in the Ring
In this, the most eatingest of seasons, I present to you the classic quandary of our time: pies or donuts?
Here, of course, our topic has more to do with data than the culinary arts. Namely, which is superior: pie charts or donut charts?
A quick aside — I know that for some of you, the answer is c) none of the above. That’s fine and you’re welcome to carry on with your pastry-free graphing, but some folks really do love their circular graphs. As for me, I don’t feel strongly either way. If you can make an effective pie or donut chart, that’s awesome. Of course, I’m also a marine biologist who doesn’t care whether you say “sea star” or “starfish.”
You’ve probably seen a pie chart before — big circle, cut into pieces, can’t miss it. A donut chart is essentially the same thing, except that it has a somewhat smaller circular cutout in the middle, turning the filled pie into a hollow donut.
That may seem like a fairly minor difference, but something important happens upon Timbit removal that illuminates the complexities of the human brain and visual perception.
How to look at a circle
Go ahead and look at the pie chart below. Notice how you look at it — chances are, your eyes go right to the center and (at least at first) you view the pie chart in its entirety.
Because pie charts are filled in, you view them as a whole; you see the circle and judge the pieces according to their areas.
And here’s a donut chart. Because donut charts are hollowed out, there is no central point to attract your attention. Where do your eyes go instead?
If you’re like most people, your eyes travel around the circumference and judge each piece according to its length. As a result, you can also think of a donut chart as being a stacked bar graph that has been curled around on itself.
Area vs. length
Why does it matter whether you read something by area or by length? Well, our brains process these two properties differently — and we’re only good at one of them.
Humans are exceptionally good at judging linear distances. If you want to throw something at a tree in the distance, that gray matter of yours will run all the calculations to get the right trajectory and fulfill your weird, anti-tree vendetta. You can easily tell if something is taller, wider, longer, shorter; single-dimension comparisons are pretty straightforward.
Areas aren’t so easy. Unless the difference is blatantly obvious, it may take you a moment to compare objects of different sizes. And if you’re anything like me, you’ll cheat and use linear distances anyway (check height and width, then guesstimate the rest).
The lesser of two evils?
Donut charts still have their drawbacks, of course — while easier to read, they’re still not the greatest for comparisons within the same graph and they’re downright terrible for comparisons between two different graphs.
Remember how I said I’d show you some of my old, crappy graphs? FEAST YOUR EYES:
I was trying to show that the daily percentage of something (I can’t tell you what for privacy reasons) increased after the October break. I thought I’d be all avant-garde and go nuts with the donuts. That was a BAD decision.
The one thing this monstrosity does do well is illustrate how hard it is to compare between donuts. For example, which is larger — the blue segment on 11/15 or the one on 11/16? It’s really hard to tell.
What could I have done with this data instead?
Well, since I wanted a before-and-after comparison, I could have used a bar graph.
This bar graph makes it much easier to see how dramatically the percentages rose after the October break.
Of course, the graph above still invites the reader to look at each date individually. I added two annotations for the “before” and “after” averages to draw attention to the changing trend, but the individual bars can still be distracting.
If I instead wanted readers to focus ONLY on the change in averages, I could have just done a simple graph comparing the two numbers:
Simple, yet effective.
It isn’t flashy, but it gets right to the point. It would have been the most efficient, effective way for me to convey my message.
Oh well. Live and learn. And cringe.
No more pie charts?
I hate to write off an entire category of graph, but more often than not there’s a better option than the pie.
When a reader sees a circle subdivided into sections, they’re going to want to compare section sizes. Even if a comparison isn’t your goal and has nothing to do with the graph’s core message, comparing sizes is sort of automatic behavior when you see a pie chart.
If you do want the reader to compare segments, other types of graphs are better (bar and donut, especially). If you don’t want a comparison, that automatic behavior makes the pie graph the wrong choice anyway.
Aside from stylistic reasons (using a coin, planet, or round dessert item as the graph background), I really can’t think of a good reason to use the pie chart. Can you?
Originally published at hypsypops.com on December 6, 2016.