78  Small Multiples

William S. Cleveland describes constructing graphs as an exercise in visual encoding. The reader must use visual decoding to interpret the graph, which he refers to as the process of graphical perception. In this section I’ll introduce you to some design aspects of graphical perception that aid in producing statistical plots.

Edward Tufte introduced the idea of small multiples as a way of presenting sub-sets of a data set in comparable formats. The concept relies on keeping the design of related graphs identical so that the user can focus on changes in the data.

Figure 78.1: Six overlapping trend lines on a single plot are overwhelming for the viewer.

Using small multiples (?fig-datingFacets), each sub-set is plotted in its own identical plotting space. We refer to each sub-plot as a facet.1 We can more easily distinguish the six trend lines, but making direct comparisons is a bit more difficult. In this case color is redundant, so all points have the same color.

  • 1 Note, this is very distinct from plotting six separate plots!

    • Small multiples separate each sub-set into its own plot, making it easier to see individual trends.