People don’t seem to get that pie charts are awful. The basic point is that the human eye cannot measure areas as well as it can measure lengths. In order to show this I’ve done some exercises in my workshops – I drew a pie chart and a bar chart and asked the class to estimate the number associated with a particular sector of the pie / particular bar of the bar graph, and (luckily I must point out, since the class was small) the error in measurement from the class for the pie was more than that of the bar, so I could convince them.
At yet another class I was teaching last weekend, I got another idea to show why pie charts aren’t particularly useful and I thought I’ll share that here.
Check out this pie chart. Let’s say this represents the quantities of various fruits I’ve consumed in the last one year (numbers pulled out of thin air, using the rand() function in Excel). Look at this chart and tell me which fruit I’ve consumed the most and which the least.
You might say that adding data labels might help solve this problem, but my question is if you must add data labels, why have a visualization at all in the first place? Also note the other problem with the pie chart – you need to keep referring to the legend (at least in the default version that Excel offers. I haven’t been able to figure out how to put the category labels next to the pies itself).
How would you better represent this data? Consider this bar chart, in that case, again made using Excel (with a few tweaks to reduce the “ink”). Here, you can clearly see the relative sizes of the quantities of consumption of various fruits, and easily figure out for yourself that my favourite fruit is Orange and least favourite is Grape.
The basic idea I’m trying to explain is this – there is little that a pie chart can show (apart from proportions, maybe) that a bar chart cannot. And even if you want to show proportions, you can do one additional step of calculating proportions and plotting that in a bar, instead of putting it in a pie.