Pie charts are the smiley faces of data visualization: Overused. A little too cute. At worst, a substitute for thinking.
So we’ve given them up. Permanently.
We crunch a lot of data here, and we want the conclusions to be crystal clear. Pie charts just don’t make the cut. They make comparisons visually difficult and defy sensible labeling methods.
Instead, we’ve moved on to heat maps and mathematically accurate funnels, stacked area charts and some striking XY charts.
If you must pie, read on for some don’ts and dos.
What not to do #1
We wish pie charts were as yummy as their name suggests. Instead, they are a mushy mess not even improved by ice cream.
This post is an analysis of how pie charts can be misleading, with a handful of ghastly examples that highlight the charts’ visual murkiness. Many of the defaults in common software programs set the stage for disaster: the first example in this post shows how data looks distorted in a standard pie chart because of the eye’s difficulty judging relative area and perceiving 3-D models.
See for yourself.
What not to do #2
But people do try. Some nice folks from the California Institute for Telecommunications and Information attempted to correct the limitations of the basic pie chart by adding functions—specifically, they addressed the eternal challenge of showing a time series.
(It has to be said that, although we appreciate this blog’s critique of pie chart readability, they may have missed the memo on blog readability: the light blue type on a white background is a killer.)
If a pie chart can do anything well, it has to be pie, right? Wrong.
This pie pie chart, which shows Americans’ favorite varieties pie, gets points for topicality and mouthwatering design, but it still utterly fails to make scientific sense.
See, the chart has a pesky asterisk, and we all know that asterisks don’t belong on pies. This asterisk tells us that the total percentage of the slices adds up to more than 100 because respondents to the survey ranked their favorites rather than chose outright.
Until someone can figure out how to get me more than a hundred percent pie per pie at Thanksgiving, I don’t want that nonsense anywhere near my statistical pie, thank you very much.
What to do
At Spark, we’re thinking beyond the pie. We’re thinking about the donuts, parfaits, and waffle charts of data visualization. They are not as easy as your basic pie chart, but they tell a crystal-clear story.