Pie charts have historically been one of the most widely used displays for data visualization. However, you'll find they're often the butt of jokes and the subject of articles like these:
Data visualizers generally despise the pie chart and claim they are a poor representation of data. We're here to give you the low-down on pie charts and when it may be necessary to use them.
In general, the size of circles can be hard to differentiate. See the example below:
It's difficult to tell the difference between First Class and Second Class - they look nearly identical, but the data aren't - second class is actually 17% larger! If you don't believe us, check out the numbers for yourself:
|
|
Area |
|
Same Day |
804 |
|
First Class |
1,886 |
|
Second Class |
2,206 |
|
Standard Class |
6,082 |
There is no one way to read a pie chart- take for example the stats below for how people judge the values shown:
- 50 percent of people use the outer arc to make proportional judgments
- 25 percent use area
- 25 percent use the inner arc or angle
It's impossible to know which method your audience will use to get insight out of your pie charts. This can lead to more questions than answers - which is the opposite of what good data vizzers are trying to do!
Pie charts can display a quick overview of your data set, yet it's not at all detailed and often doesn't represent subtle differences well.
Take the chart below. Which category has the greatest number of sales?
The chart makes it very hard to see since the difference of sales may not be that great. However, if your company wants to know which category is bring in the most sales for them, you need to be able to be able to give them an accurate answer.
For this data set, a bar graph would be a better representation, since the eye can most easily see distribution by height.
The bar graph can also be sorted to help the eye comprehend size hierarchy, making it much easier to answer your questions.
By now you may be asking, "Is there ever a scenerio where a pie chart would be useful?"
Here are some examples of when it may make sense to use them:
...and no other "part-of-whole" chart, like Marimekko charts or stacked bar graphs are getting through to the consumer. This is can be crucial when you're trying to drive adoption of your dashboards - what's the point of building a dashboard for a consumer, if they aren't going to use it? If the reasons above aren't compelling enough to drive adoption of other chart types, use pies.
When you have more than three categories in your pie chart, it becomes particularly difficult to gauge size. However, it is easier to read if there are 2 or 3 categories, especially if they are widely different values and you include a label. If it's more than 3 categories, we recommend always using an alternate chart type.