Thinking about Visualisation and Design
by John McAuley, visualization manager at Novartis.
When thinking about design, visualisation falls into one of three camps. First, there’s the scientific camp in which visualisation is used to assist in scientific discovery and enable the presentation of research findings. Typically, this is the sort of visualisation found in research publications.
Error bars are integral, although recently found less so in a perception study at Viz 2014, and colour is kept to an absolute minimum, black, white, shades of grey. The more hip may use something like ggplot2, a great R package by Hadley Wickham, but in general the design lacks an aesthetic appeal and interactivity is rarely supported. We could flippantly reduce this camp down to trend (or line) and bar charts, often with log scales and error bars.
Second, there is the infographic camp.
While not necessarily new (some of the original and still most discussed visualisations, such as Minard’s Napoleon’s march, would today be labelled as an infographic), infographics have become a hugely popular way to reduce large quantities of information into concise representations that can be easily shared with a broad and potentially less analytic community. Design is heavily applied – representation is not reusable – colour is used to make for an aesthetic appeal (as opposed to enabling deeper insight) and space supports a clear presentational form. Infographics tend to have a narrative bent, allowing readers in a more conversational tone to grasp the complexity of a dataset or delve into a difficult phenomenon.
For the majority, Infographics are static, designed by graphic designers – in conjunction with a domain expert – and shared on social media or through publication. Often an infographic is used to support an argument, to persuade the reader that a particular product will solve their problems, that a political party has all the answers or that a particular policy will be the ruination of all we hold dear. What is interesting here is the representational approach that designers take when trying to persuade the reader towards a particular end. The, let’s say, more gaudy and less precision-orientated approach (described by Vande Moere as “visual embellishments”) that some infographic designers apply is often criticized in the visualisation community as “chart junk” having a low “data-to-ink” ratio and serving of little practical purpose.
However, when we think about this sort of visualisation we should not apply the same critical framework we would when addressing scientific visualisation and instead consider the goals and objectives of the designer. Some research on this topic has shown that gaudy and arguably silly images have improved recall, which, if wishing to persuade somebody towards a particular decision, could be considered as a legitimate design objective.
The last and final camp, I consider information visualisation.
Unlike scientific visualisation, information visualisation is not based solely on the analysis or presentation of scientific data (or data obtained through the application of the scientific method!) but can also address a range of different data sources and data types – from social data presented using an ambient display to text corpora graphed as a semantic network of interrelated concepts – yet the goal remains generally to inform as opposed to persuade the reader.
Typically, the elements of design are used within this context – space is used in line with Gestalt theory, as a way to group similar items for example, and colour is applied to support perception and improve cognition.
However, increasingly researchers and designers are investigating how design can be applied in different ways. Vande Moere, for example, has sought to conceptualise the design space in terms of utility, creativity and soundness – in which a visualisation is considered along the lines of utility, reliability and robustness – while labs such as Density Design have developed a notable aesthetic approach in their design of complex visual displays. Advances in computing power and the development of interactive frameworks such as Prefuse, Flare and of course D3 have illustrated the potential that interactivity can play in engaging people with visualisation while also enabling a deeper level of insight. Although work by Jeff Heer has suggested that interactivity remains a “double edged” sword, leading potentially to erroneous insight, there remains much scope to increase the application of design yet still, as Tufte states, “show the data”.