Pie Chart

A Brief History

The pie chart was probably first used by William Playfair, in a book published in 1801 called The Statistical Breviary. In it, he used circles of different sizes to represent the areas of the various countries in Europe and how they changed over time.

Figure 1 Pie Chart Based on the Data in Table 1

[Page 1031]The circles were colored to indicate whether the country was a maritime nation (green) or not (red), and some were divided into slices to reflect the ethnic backgrounds of the residents. In a previous book, Playfair had used overline charts and line charts for the first time. However, possibly due to his reputation as a shady businessman, pie charts were not adopted in the United Kingdom for nearly 100 years, although they were better received in Europe.

Variations

In an exploded pie chart, one segment is separated from the rest. This can be done either to highlight that segment as deserving of special attention, or because that slice is so small that it would otherwise be overlooked. In a polar area diagram, all of the segments have the same angle, and differ from each other in terms of how far each slice extends from the center. Although this type of chart is often attributed to Florence Nightingale (and, indeed, is sometimes referred to as a Nightingale rose diagram), it was likely first used in 1843 by Léon Lalanne.

Problems with Pie Charts

Despite their ubiquity in newspapers, magazines, and business presentations, pie charts are rarely used in the sciences. They are pleasing to look at, but it is difficult to draw conclusions about the numbers that they represent. One major problem is the interpretation of the area of the individual segments. Viewers can easily discriminate among lines of different lengths, as in over-line charts and histograms. They are less accurate, though, in discerning differences among areas and angles. Doubling the length of a line makes it look twice as long, but doubling an area makes it appear 70% larger. If the slices of a pie chart differ from each other only slightly, people have much more difficulty rank-ordering them than if the data were presented as overlines. Viewers also have difficulty interpreting angles; they tend to underestimate acute angles and overestimate obtuse ones. Furthermore, the same angles are seen as different if they are oriented differently (e.g., the slice extending vertically versus horizontally from the center).

...