Chapter 98. Tip #13: Eliminate Chartjunk (But Not Graphics)
Chartjunk is a term coined by Edward Tufte in his book, The Visual Display of Quantitative Information. In it, Tufte defines chartjunk as “non-data-ink or redundant data-ink.” My interpretation of chartjunk is that it is any design element that is counterproductive, actually detracting from a data visualization rather than adding value to it. Chartjunk can be as brash as charts that are made to fit within a graphic, with the data almost a complementary element to an illustration (see the work of Nigel Holmes). It can also be as subtle as a three-dimensional bar or extra gridlines. Calling nonessential elements “chartjunk” is a fair criticism, but the same person who invented the term also said the following:
“Only a picture can carry such a volume of data in such a small space.”—Edward Tufte
As with most debates, the truth probably lies somewhere in the middle. When the design becomes the primary purpose of a data visualization, it can become distracting and even misleading. Neglect design completely and a data visualization can be unengaging. In the last tip, I explained the importance of balancing data and design in data visualization. The tasteful use of graphics is one way to do that.
Chartjunk is not always composed of graphics. There are several examples of “non-data-ink” outside of graphics, including extra gridlines, shading, and three-dimensional marks. In the same way, graphics do not always create chartjunk. ...