On Sunday 19th October I went along to the “Introduction to infographics and data visualisation” (Guardian Masterclass). The main aim was to explore different approaches to communicating complex data to a range of audiences. For example, as a researcher I hope to eventually publish my analysis in a peer-review journal to share my findings with an academic audience. However, of equal importance is to communicate these findings to a more general audience, and I think infographics are a great mean to that end. The day turned out to be a great, informative, with a good balance of theory and practical advice.
As humans, as are primarily visual creatures. Visual instructions make us 3 times more likely to complete a task, compared to written instructions alone. (For example, imagine trying to assemble Ikea flat pack furniture reading only text versus the pictures, arrows and diagrams). The combination of images (which provide an immediate emotional impact) combined with insight and information through text, help us to learn. As such, data visualisation is a form of storytelling.
The main ‘schools of thought’ in visualising data are Tufte and McCandless. The Tufte approach is best when your audience is technical or expert, and this approach prioritises data quality over design. In a sense, every pixel should correspond to something in the dataset, otherwise it is just chart junk. McCandless on the other hand prioritises getting the maximum number of people interested in the data, who would previously not have been interested. Therefore for McCandless design is king, and slight data rounding/manipulation is ok if it helps to get the main message across.
A good example is this famous infographic “Billion Dollar O Gram” by McCandless (click to enlarge). The actual data almost certainly doesn’t fit into such nice linear boxes. In fact, at the bottom is a note “some slight visual cheating to make things fit”. I imagine this would be deemed entirely inappropriate by Tufte followers. However the key question to deciding upon your approach (Tufte v McCanless) will be: who is your audience?
We were introduced to a fool-proof process which (apparently) works no matter what the project, story or client. I say apparently, as I haven’t put it to the test as yet, but it does intuitively make a lot of sense. The process involves 4 steps, which must be followed in a consecutive order.
Collaboration and teamwork are important. The ‘masters’ noted how you are unlikely to find someone with the whole skill-set, and the best work tends to be the product of a team effort. (Although it can be useful to expand your skills to learn and appreciate different parts of the process).
Remember the rule of 7. (Or as I remember it from my undergrad classes 7+/-2.) In the 1950s George A Millar reported that 7 is the optimal number for human memory, afterwhich information overload tends to occur.
Another master highlighted that society is increasingly being bombarded with volumes of data and we have an increasing demand for information. However, low levels of statistical literacy in general population is often a barrier to good decisions. Therefore although aesthetics are appealing, the outcome of informing decision making (and therefore enhancing understanding) should not be forgotten.
It can be helpful to think about the characteristics you wish to expose in your data visualisation. E.g. correlation, change overtime, magnitude, distribution, ranking, uncertainty, deviation, spatial, part to whole… etc this fits in with the story building part of the process.
How to make it all beautiful… the job of design is form and function combined. If you fail to communicate, you fail to design. The job is visual story telling. The final product should have visual clarity and visual persuasion (Scott McClough).