Making Readable Graphics
What makes a readable graphic? This is a tricky question, because the answer changes depending on who you ask. Many turn to the data-ink ratio, as described by Tufte:
The larger the share of a graphic’s ink devoted to data, the better.
People tend to treat this as a hard rule. However, Tufte gives himself wiggle room in the rest of the data-ink description:
The principle has a great many consequences for graphical editing and design. The principle makes good sense and generates reasonable graphical advice—for perhaps two-thirds of all statistical graphics. For the others, the ratio is ill-defined or is just not appropriate.
The ratio is good for two-thirds. For the rest: ill-defined. That leaves a whole one-third for which maybe, possibly we should strive for something else? Does that mean we can bring in the “chartjunk” sometimes?
Well. It depends.
If visualization was nothing but unbreakable rules, then we could just let the computer do our jobs for us. Bow down to your robot overlords.
Data is more fluid than that though, because it is a representation of something in the real world, and as we all know, life can be complicated. Visualization is a way to represent that fluidity in the data.
There are visualization rules that you cannot break. They involve the technical aspects of how a chart is constructed.
As we talked about in the previous section, established chart types like a bar chart or pie chart use specific visual encodings. The bar chart uses bar length to represent values, so if you truncate the axis you truncate the value. The values of a pie chart must sum to 100 percent, because the sum of the parts always add up to the whole.
For these sort of things, if you break the visual encoding, you use the chart wrong, which means you mess up how readers interpret the data.
However, a few times every year, a discussion starts around the fuzzier aspects of chart design.
“The baseline always needs to start at zero.”
“Pie charts are terrible. Never use them.”
“A bar chart would’ve been better.”
The rebuttals are always similar in flavor:
“What about the sort of dataset that doesn’t include zero?”
“People know how to read the chart type and it’s fine for this very specific dataset.”
“You are dead inside.”
The debates always end the same with everyone agreeing to disagree. It’s a grand ol’ time.
In practice, when visualizing data for an audience, there are always factors to consider that can conflict with visual efficiency. You decide what tradeoffs you are willing to make and move on. Sometimes you make the right choice and sometimes it doesn’t work. It’s part of the process.
With that out of the way, let’s go back to the original question: What makes a readable chart?
A readable chart is one that provides clarity. A readable chart is one that aims to remove confusion. A readable chart is one with a clear purpose. A readable chart is one that uses a visual encoding that makes sense for the context of a dataset. A readable chart is one with a clear direction for how to interpret.
In the sections that follow, we look for ways to do this. We’ll lean more heavily on exercises so that you can better learn how the design principles work in practice.
If there’s any confusion at all, feel free to post to the forums for help. I also encourage you to post your work in the forums, as feedback helps you improve quicker.
