The Visual Display of Quantitative Information

As a student, I’d get The Visual Display of Quantitative Information from the university library just to look at it. I always wanted to own this book. It is one of the most beautiful printed products I’ve held in my hands for a long time. You would strategically place this book on your coffee table to impress your nerdy friends. The typography, the layout, and figures — everything is incredibly elegant. Not only does it have the look and feel of an old book, it also smells old. The edition I purchased was printed in August 2009. In my imagination, it has been sitting in Edward Tufte‘s garage for the last 12 years, acquiring the distinct smell you’d expect from the antiquarian bookshop. The smell is not unpleasant; it adds to the experience.

The Visual Display of Quantitative Information is a guide to making successful data visualisations by increasing the density of information in a graphic. You do this by increasing the data-ink ratio, the ratio of ink used to portray data to the amount of ink used for auxiliary information, such as scales and labels, or purely for decoration. From here, everything else follows: Remove all decoration, lower the number of ticks and labels displayed in scales unless they add information, and enhance the readability of the chart. Show the data in the most minimal way possible. Tufte makes his point by example. He starts with well-designed graphics and goes through various reduction cycles to end up with a better graphic.

Other suggestions punctually add elements the enhance a graphic with additional information. Using range-frames and quartile plots instead of scale bars is a great idea to convey more information with little additional graphical overhead. To this day, these ideas are niche and never made it into the mainstream — probably one reason why Tufte has such an ardent following.

It’s a thin line between improving a chart and just being pedantic. Reducing box plots to a dot representing the mean and white space to indicate quartiles because it uses less ink is pushing it too far. Passing over established conventions makes the graphic harder to read. When I see a typical box plot, one with an actual box to depict the two central quartiles, I know how to read it. In Tufte‘s example, I have to go the extra mile to figure out what the chart says. In this case, more is better.

I don’t share Tufte’s distaste for patterns and moire effects — he calls it junk. I have a soft spot for monochromatic charts and maps. This map by Tom MacWright is beautiful, and I have experimented with black-and-white SVG patterns myself. Apparently, monochromatic patterns are „an illusion, eye-straining quality that contaminates the entire graphic.” Well then, I may have to reconsider my opinions.

It’s a good book that is fun to read and pleasant to look at. If you’re making data visualisations every day, then you won’t learn that much new. That’s ok — Tufte wrote The Visual Display of Quantitative Information in the early 1980s, a time when “those who produce graphics for mass publication are trained exclusively in the fine arts.”