Why a box?
Consider this box plot. What dos the box actually get you? What is the ratio of data expression to ink?
This boxplot is great for analysis …
… but this is cleaner, leaner and perhaps clearer for a presentation.
Or consider this example:
Fig.1: Percent delays by city and airline
Principle 4. Pictures are words
Consider these sparklines - they are graphical words.
In R we can put more data into a sparkline if we want to. Here we see IQR and min/max graphed onto a sparkline:
Tufte invented the sparkline - a small graph that shows the upward and downward trajectory of some variable in a small, concise graphical statement. For Tufte, the sparkline functions just as a word does - small, discrete representation meant to express a single - though complex - idea.
Principle 5. Get off the grid
In the graph below, the lines are the same as in a line graph. But the y axis has been obliterated - instead, it is occupied by the names of airlines on either side with their traffic shown at various points. It is as if the sparkline words had been arranged in a paragraph. It is a very small change but a highly effective one:
