I have some devastating news for any circle loving freaks out there: bubble graphs and pie charts may look more visually appealing than bar graphs, but our minds find them harder to read. Adjunct professor of statistics Kaiser Fung has declared that bubble charts are hopeless, and many information designers agree. This is why:
When comparing the circles above, it’s not immediately obvious what their size difference is, whereas the small bar is more clearly 2/3rds of the big one. Bars require us to do one thing: compare one dimension, their length. Circles require us to compare the multiple dimensions that makes their area. As I explained in my first post, our brains are lazy, and asking them to interpret circles takes more effort.There is also confusion about whether we are comparing circle heights (diameter) or areas (anyone remember πr^2?). It is conventional to compare areas, but as computer graphics programs change the values of the circles diameter (seen above in red) when you scale an object by percentage rather than the area (seen above in green), designers can end up making bubble charts that display data in terms of diameters rather than area.Business information designer, Stephen Few, explains that our perceptions of angles in pie charts can be just as unreliable. In the pie chart on the left can you work out which colour represents 1/4? It becomes more obvious when I rotate the chart around, as seen on the right pie chart above. Again, we find working out angles more complicated than seeing lengths, unless they appear in a familiar position.Following the idea that simpler is easier, designer Edward Tufte tried to simplify data display even further, by reducing conventional bar graphs to a series of lines, which resulted in this rather confusing graph:
This demonstrates how familiarity can also play a large role in our ability to understand; although the above graph has been simplified to lines it’s meaning is not made clearer. While this technique may have saved ink back when print media ruled our world, psychologists Chabris and Kosslyn explain that our mind reads the connected lines as several separate elements of information, rather than the one single solid visual element it sees with a bar. Like with the circles, this adds more elements to understand and therefore increases the effort required to read it. Chabris and Kosslyn also discuss how making bars 3D has the same effect of introducing extra elements that our brains get distracted by. So if you want your audience to see an accurate comparison of values: a flat bar graph imposes the least strain on our brains.
However, circles are capable of providing viewers with a general overview of data, for example, quantities displayed across locations on a map, like the image above. And it turns out, we aesthetically do prefer curved shapes rather than angled shapes like bars. As Scientific American’s Jen Christiansen points out, there are times when engaging your audience, while providing a general overview of the data, is the main objective. In these cases it can be appropriate to use more complicated forms of data visualisation to capture people’s attention, especially when we don’t need to know precise values. After all, if you can’t capture people’s attention, then they won’t read the graph.