Pie Charts are bad and 3D Pie Charts are very bad.

Hi, everybody,

As a data analyst in a large organization, I read about how bad is to use pie charts and in particular 3D pie charts. However, the managers whom I work for really like pie charts, especially 3D ones. I usually label each wedge with the actual value and wonder if this is good enough to alleviate the problems raised in this guideline. These managers are not that stupid. Their instinct may tell them or us something. I thus wonder if pie charts have some merits that may explain why many people like these.

Source: https://www.stevefenton.co.uk/2009/04/pie-charts-are-bad/


Hi Peter,

  1. Google: “Financial Times, test, The science behind good charts”. I recommend you do the test and you will never use pie charts again.

  2. You could save the pie chart, only in these cases. When Pie Charts Are Okay (Seriously): Guidelines for Using Pie and Donut Charts

  3. 3D graphics only provide decoration, does not provide information.

  4. I also recommend this paper: “Graphical Perception: Theory, Experimentation, and Application to the Development of Graphical Methods. William S. Cleveland and Robert McGill. Journal of the American Statistical Association. Vol. 79, No. 387 (Sep., 1984), pp. 531-554.”

Best regards,


Hi Peter. I’ve done research on pie charts and have written about them on my blog. First, we need to separate pie charts from 3D pie charts here. With a small number of slices, and for data that has a meaningful part-to-whole relationship, they work just fine. There’s a bit more on this on my website and I also gave a talk on some of my research a few years ago (search for “Unloved” on my website to find it, I can only post two links here as a new user).

3D pie charts are a different matter. 3D charts generally introduce lots of distortions that aren’t related to the data at all, and produce large errors in reading data. I have a forthcoming short paper at VIS this year where I used the distortion created by 3D pie charts to investigate the perceptual mechanism used to read the charts. Figures 1 and 2 in the paper should give you a sense of the distortion, and this is using parallel projection – perspective projection is a lot worse.

As to why people like them, my personal theory is that we inherently like round, complete shapes better than the jagged bar charts that usually work much better.

So in short, I’d recommend dropping the 3D and just going with 2D pies – assuming you don’t have a lot of data points in each chart and the data fits the chart. Adding the values is a good idea, but it’s really a cop-out. What purpose are the charts serving if people are just reading the numbers?

As an alternative to pie charts I started using treemaps recently. They allow for more data dimensions without much additional clutter, seem to be able to fit more data on the same chart, and seem to make it easier to pick up differences in area. Pie charts seem fine for me with datasets that have less than 5 five categories that I want to render at the same time. For more I would choose another plot type.