From Business Intelligence
Wikipedia: Michelson and Morley's data on the speed of light. This traditional box plot rightly eschews gridlines and color-coding in plotting the speed light experiments and adds a red line to mark actual speed. It could be improved by using either Tufte method in the following examples, and also by removing the frame and tick marks.
Created entirely using the spreadsheet available at Tufte Charts in Excel. This is a Tufte-redesigned box plot that eliminates the box and the markers, using instead simple lines and dots along with whitespace to distinguish between elements. This contains all the information of the traditional box plot in a simpler display.
Plot with offset interquartile ranges. This is another Tufte redesign that uses slightly offset lines to distinguish the middle quartiles from the outer quartiles. This particular implementation should eliminate the gridlines and the frame. Gridlines are normally extraneous, but in this case they may distract from the offset lines and are thus a bigger concern here.
Also known as a box-and-whisker chart. This visualization shows the distribution of values in the data according to the upper, median and lower quartiles. Use this to accurately depict data that has skew and outliers.
- There are many ways in which box plots can reduce non-data ink.
- Instead of using an actual box and whiskers, place a dot in the middle of the median and use plain lines to show the upper and lower quartiles (see example 2).
- Another possibility is to use plain lines instead of boxes and whiskers, and offset the quartiles (see example 3).
- Grid lines should be avoided.
- Avoid using color. In a typical graph, there is no reason to "color-code" the boxes. As this page demonstrates, category names are adequate.
- Learn more about box plots at Wikipedia.
- Information & Visualization has a good blog post discussing ways to design and use box plots.
- A Tufte-style box plot can be created using the spreadsheet available from Juice Analytics.