Are pie charts good visualization choices? Well, pie charts have their own problems as well. Pie charts are too rowdy. The issue with the part chart is that they’re designed to represent different percentages. You could say, “Oh, well 37% of people use this toothpaste”, “22% of people use that toothpaste”. There’s no consistent axis where you can actually line up that 27-37% and clearly see which one is larger. Sometimes it’s very difficult to distinguish that.
I know Tufte was very anti-pie charts, and I don’t quite take a position quite as strongly as his. I should throw that out at least, but nothing that would be enhanced by making it 3D. All 3D does is makes it a little prettier. So, people do that because they think it makes their graph more eye-popping. It has non-data-ink. It’s distracting. It draws attention away from the data that you’re actually trying to show.
There is another ratio I want to talk about which is data-density ratio. Data-density is the number of data points or data entries in your graph, the area of the graphic, and this is something you’d like to maximize within reason in a graphic. And you can do this by increasing the data/ink ratio. Well, you can do that by eliminating all the metadata, the redundant data, and all the chart junk, and so on.
|#1 Ranking: Read how InetSoft was rated #1 for user adoption in G2's user survey-based index
But you can also increase this ratio by increasing the amount of data in the graphic. The human body is capable of seeing a lot of data on a graph. It’s better to show that information as long as it’s well presented. So for example, you could have multiple series shown in the same graph and show them each in a different color.
You could have multiple graphs next to each other. It’s possible to shrink a graph down very, very small and have one really effective way of showing more than one dimension. Take a 2D graph and repeat that graph a dozen times across a page. So, shrinking it all down to a very small level and just having one variable change between each one. You could have for example, temperature over time, but change in the concentration of some chemical element in each one of these pictures.
You’d actually show a third-dimension in these two-dimensional graphs. So, by doing, you’re adding an awful lot of data into a small area of these graphs and increasing the data-density. And these make for very effective graphs. This is one of the reasons that the Napoleon march graph was so effective because so much data was packed into such a small space.
Maintain That Graphical Integrity
Let’s just summarize here. We want to make sure that we’re not lying with our graphs. Maintain that graphical integrity. We want to maximize the data-ink ratio within reason. We want to maximize the data density within reason. We want to avoid chart junk and metadata, and gridlines, vibrations, ducks.
We want to label our data well and make sure the text that we’re putting on there represents the data and blends in well with the data. And the big takeaway from this is just think about these things. It’s so easy with a tool like Excel or PowerPoint to just fire up a wizard, click “next”, “next”, “next” and throw a graphic on there. And if that graphic is not really doing something for your users, if it’s not doing a good job of explaining the data to your users, then what’s the point? Sometimes it’s easier to just show the data.