Chartjunk, a term first coined by Edward Tufte, refers to all those elements in a chart that add no further value. Chartjunk comes in many shapes and size – redundant labels, unnecessary gridlines, over-the-top use of colors and everything that a ‘creative’ mind can think of. However what should be kept in mind is that this term also has a strong contextual meaning. An element that may appear as chartjunk in one graph may serve a useful function in another and vice-versa. Hence a broad all-encompassing classification of an element as chartjunk should be avoided.
Chartjunk can be easy to spot for the trained eye and once you’ve had an initiation, it will typically be one of the first few things that you will tend to notice when information is presented to you graphically. Let’s take the example of the chart shown below. Admittedly it is a pretty readable chart.
However on closer inspection, we are able to identify a number of elements that can be eliminated without taking anything (that matters) away from the chart.
Gridlines : The gridlines are self evident. However in cases where the amount of data presented in large, gridlines can always be given lighter shades so as to allow the focus to remain on the information being presented.
Borders : Normally borders will not really have an impact on the data being presented. Under ‘normal’ circumstances, they would not interact with the data or the visuals. Removing them will save precious screen ‘real-estate’.
Background Colors : If anything, background colors will in most cases, tend to decrease the contrast and delay the discovery of the message being conveyed. However, once in a while we come across graphics that are exceptions. In these cases, the background helps to paint a story that the compliments the narrative and the data being presented.
Other elements that can act as chartjunk can include legends when they are self evident, minor-tickmarks and data tables. However as I said, chartjunk is contextual and one size does not fit all.
Here’s our chart without the accompanying chartjunk.
Guess how much area did we save by eliminating chartjunk?
A full 45% !
(that was an approximation but brings out the point)