If you're ever at a loss for conversation amid a group of information-graphics professionals, bring up the topic of pie charts or proportional circles. Then stand back and watch the sparks fly. At the World Infographics Summit in Pamplona, Spain, this past week, the love/hate relationship with circular data presentations was a common theme.
Purists, such as Stephen Few (founder and principal, Perceptual Edge) and John Grimwade (graphics director, Conde Nast Traveler) are wary of the functionality of circle-based charts. Indeed, the human brain is more adept at comparing lengths than areas. For example, the largest wedge in this pie chart is clearly identifiable (figure 1). But what is the second largest category? And the third? A bar chart--even without a background grid--is easier to read.
Others, such as Alberto Cairo (director for infographics and multimedia at Época-Editora Globo), worry that bar and line charts have become too familiar, and risk being overlooked or dismissed too quickly by the reader. These standard visualization formats are indisputably elegant solutions, but information-graphics professionals should not rest on their haunches and rely too heavily on a form established in the late 1700s. Instead, we should push the boundaries and explore new ways of presenting the data in an effort to better engage the reader. Perhaps it was in this spirit that the proportional circle chart was born.
I have certainly been seduced by the format. Back in the July 2010 issue of Scientific American, I used proportional circles to provide perspective for the Deepwater Horizon oil spill (figure 2). I think it was an effective solution. But readers need to take a leap of faith when faced with these sorts of charts. Is the data properly represented in terms of relative areas of the circles? Or was the data used to inform the relative diameters of the circles? This representation would result in a dramatically different and misleading chart.
Part of the appeal in using proportional circle charts lies in their flexibility. The designer is released from using axis lines, and is given the freedom to use data points as design elements on the page. Although often visually pleasing, it may do a disservice to the data. Take this example (figure 3). The two central circles are the same size, but our perception is distorted by the surrounding elements. Chart makers can control exact measurements. But what about perceived measurements?
In my opinion, context is key. I would hope that my doctor relies on excruciatingly clear and straightforward--if boring--data displays. Yet readers of Scientific American magazine may be better served by more engaging and dynamic charts. The trick lies in approaching every chart thoughtfully, with the goal of allowing the data's story to outshine its container.