30 Setting the scene
“To find a form that accommodates the mess, that is the task of the artist now”
Drawing graphics involves choosing the variables to look at, deciding what raw data and statistics to display, and selecting plot types. Initial plots may benefit from changing the scales or transforming the variables. It might be useful to exclude some of the cases. It could be helpful to add models. All of these actions are concerned with the content of the graphic.
With graphics for exploration, analysts should have background knowledge about the data and the main aim is to make structural features stand out. Exploratory graphics need little text of any kind, possibly just variable labels and simple scaling. With graphics for presentation, more information is needed. Few viewers will have enough background information, so context must be provided to make the graphic understandable. Guidelines of a journal or organisation may demand particular formatting standards. Ideally graphics should be polished as well as informative. Whether graphics are drawn for exploration or presentation, it is sensible to avoid overcrowding within a plot and to avoid overcrowding of plots together.
The content itself, the data objects like points and lines, can be reformatted. Sizes, colouring, shapes, ordering can be changed, sometimes dependent on dataset variables. The framing of the content could be extended and amended: the axes, tickmarks, gridlines, background, text (title, caption, labels, legends, annotations), spacing, size, and aspect ratio. Colouring, size, width, typeface apply to many of these. Most of this kind of formatting is for refining graphics for presentation, but improving the look of graphics helps exploratory work too.
Exploratory graphics involves drawing large numbers of graphics (this cannot be said often enough). Software is not as supportive as it should be in handling many graphics. They need to be ordered, aligned, and arranged. Scales and colouring should be consistent. Graphics texts tend to give advice on drawing individual graphics, but much advice is also needed for drawing and organising sets of graphics.
Recognising information in data graphics is easier if the graphics are well displayed. It is best if they are carefully laid out with plenty of space, both within and around them, and that they are appropriately formatted. When actors prepare to put on a play, they rehearse the words first, the content, and then add actions, while the stage designer, light manager, costume manager, and other specialists organise how best to display the content and set the scene. It is all brought together for the performances.
There are two kinds of graphical scene setting, for exploration and for presentation. Although they are not distinct, it is helpful to think of them separately. Exploration benefits from flexible layouts and plenty of space, concentrating on overviews more than details. Scales and labels may be the only text needed. Formatting should be supportive and unobtrusive, ideally just using defaults, so that little, if anything, needs to be changed. Exploration is open and informal. Presentation requires precise layouts of specified sizes, using space efficiently, with detailed formatting and explanatory text including titles, captions, legends, annotations, accompanying comments and more. Presentation is closed and formal.
Exploration is at most a small group activity, involving only a few people who know the aims and background of the project. For presentation the audience is potentially huge and their interests, knowledge, and experience will vary. This chapter discusses four aspects of graphical scene setting to make the information in graphics stand out: scaling, formatting, space, and layout.