Reading a pie chart without being fooled by it

A pie chart can be trusted for one specific job: showing how a single whole is divided into parts, such as the share of a household budget that goes to food, housing, transport and everything else. The circle stands for the total, and each slice's size stands for its share of that total. Read that way, a pie is a fair picture. Most of the trouble comes from using a pie for something it does poorly, or from a chart whose slices do not actually add up to the whole it claims to show.

Before you draw any conclusion, check that the slices add up. If the labels give percentages, they should total 100 percent, although rounding can leave a total of 99 or 101 percent. A total far from that is a warning sign. It may mean respondents could pick more than one answer, so the chart is really showing mentions rather than shares of one whole. In that case a pie is the wrong format, and a simple bar chart would show the counts more honestly.

Check what the whole represents

The whole matters as much as the slices. A pie titled "spending by category" could describe one household, a city budget or a survey of a few dozen people, and those are very different things. The title or caption should say which one it is, along with the number of cases behind it. Consider a made-up example: if a pie is built from 20 responses, a slice of 35 percent stands for seven people. It looks as authoritative as a pie built from thousands of cases, but it is far less certain.

Next, look at what the slices measure. Some pies show shares of a count, such as the number of people who chose each option. Others show shares of money, such as spending. These are not interchangeable. A category that holds 30 percent of spending may involve far fewer people than its slice suggests if those people spend more on average. The label should name the measure, and the measure should match the claim in the text.

Also watch for two pies placed side by side. Two circles of the same size can hide the fact that each shows a different total. A slice can shrink even though the amount behind it grew, if the whole grew faster, and a slice can grow while its amount fell, if the other categories fell further. Look for the totals or the underlying counts, and compare shares only when both pies describe the same whole.

Too many slices and the tiny ones

Each extra slice makes a pie harder to read. Thin slices are difficult to compare because their edges sit close together, and their labels crowd into a small space. Many designers keep a pie to about five slices. When a chart has a dozen categories, a common and reasonable step is to group the smallest under an "other" heading. That is fair when the chart says what the group contains. It is misleading when the "other" slice hides a category that matters to the story being told.

Slice order also shapes what the eye notices. A chart that starts the largest slice at twelve o'clock and runs the rest in descending order is easier to read than one with slices scattered at random. Pay attention when the slice that supports a headline sits exactly where the eye lands first. Order is a design choice. It should not change whether the numbers are correct, but it can change what seems important at a glance.

Colors deserve a check as well. Charts that reuse the same palette for different categories can make one slice look like another across two pies. Read the legend every time, and do not assume that a particular color always means the same category.

Comparisons are where pie charts fail

Chart designers generally agree that people judge lengths along a shared scale more accurately than they judge angles or areas. A pie chart asks the reader to compare angles. Two slices of 22 percent and 25 percent are hard to tell apart by eye. The problem grows when the slices are not next to each other: an 18 percent slice and a 21 percent slice on opposite sides of the circle are very hard to rank, because the eye has no shared edge to measure them against. If your goal is to rank categories or to see a modest gap, a bar chart with values printed beside the bars will serve you better.

Three-dimensional pies make the problem worse. A tilted slice at the front of the circle looks larger than an identical slice at the back, because perspective enlarges what is closer. Exploded slices and thick donut rings are a different matter: they do not tilt the picture, but pulling slices apart or hollowing out the center removes the shared edge or center that helps the eye compare them. When a chart uses 3D effects, look for the underlying figures in the source, or picture the slices as flat bars before you judge them.

When you want to compare the same categories across two groups or two years, a 100 percent stacked bar chart or a set of plain bars usually works better than two pies. Each bar starts from the same baseline, so the lengths can be compared directly. A pie remains a fair choice for a single whole with a few clearly different parts. In that case the checks above are usually enough to read it correctly. The rule of thumb is simple: if you cannot tell which slice is bigger without measuring, the chart is asking more of the circle than it can give.