33.4 Classifying graphics

Graphics may be classified in different ways and a number of terms are used. Data graphics generally refers to displays of raw data. Model graphics implies displays associated with models, such as plots of predictions or residuals, plots of parameter estimates or of statistics for comparing models. Multivariate graphics includes displays showing many variables such as parallel coordinate plots and mosaicplots, small multiples including facetings, and dimension reduction displays such as biplots and multidimensional scaling plots. Interactive graphics describes displays which can be directly manipulated: queried, sorted, reformatted, linked across multiple displays. Dynamic graphics means displays which can be set in motion, such as rotating plots and animated plots. There is some overlap between interactive and dynamic graphics, but the emphasis with the former is on interaction and the emphasis with the latter is on dynamics. And there are further terms mentioned in the literature.

It is not just the technical aspects and the data background that affect the use of graphics. The setting is important. This can range from a publication of a research project in a specialist field to an item on the television news about the progress of an epidemic. In some circumstances a graphic may be paid close attention, in others the graphic may be competing for attention. Sometimes much must be explained, sometimes it is safe to assume substantive knowledge on the part of the graphic’s viewers. Graphics are not an easy matter.

Main points

• Not all graphics that look the same are the same. Context matters.
• Ensembles of plots require coherence of look, theme, alignment, scales, and formatting.
• Graphics can be changed in many ways and describing why a change has been made is more enlightening than describing how it has been made.