Data Visualisation for Drunks

Two months ago, I wrote an article reflecting on how data visualisation is not primarily about creating charts, but should instead focus on effective communication through visual means. Today, I want to share one of my favourite principles for designing impactful visualisations: eliminate complexity. Or as Kelly Johnson, the aeronautical engineer who headed up Lockheed’s Skunk Works, exhorted his team: KISS. Keep it stupid simple.

This was not an arbitrary challenge. Kelly’s elite engineers were developing the world’s most advanced aeroplanes (the SR-71 Blackbird among them), which would have to be serviceable out in the theatre of war with standard tools and basic training. While not all of us are realising our visualisations for deployment in combat situations, the lesson still applies: you cannot control the environment in which your work will be used. While the physical tools might not matter in the case of data visualisations, mental tools certainly do.

After all, it takes energy to interpret data — and we know jobs are becoming increasingly fragmented. With frequent task switching and its associated penalties (see No Task Left Behind? by Mark et al. 2005) we are seeing an impact on employees’ ability to focus effectively. Moreover, high cognitive load means that people are less likely to sit down and fully digest information. From a societal perspective, there are other factors to consider, such as the impact that poverty has on decision making capabilities (see Poverty Impedes Cognitive Function by Mani et al. 2013) or a lack of general analytical literacy.

What is particularly important in is the audience’s ability to understand and contextualise the information presented.

Jet lag is another example. As Williamson & Feyer noted in a study, even mild sleep deprivation — i.e. going 18 hours without sleep — led to a performance impact equal to a 0.05% blood alcohol concentration (BAC). To put this into context, this means a transatlantic flight can impact employees’ performance as much as two alcoholic drinks! Worryingly, the study found that prolonged sleep deprivation increased these symptoms to an equivalent BAC of 0.1%. This is the threshold for loss of motor control and impaired speech. At this point, focused data-driven decision-making has long fallen by the wayside.

As analysts, we tend to spend a lot of time ruminating over what information we want to present and how to best communicate it. But perhaps we should consider what state our audience will be in when they are presented with the data visualisations. Instead of creativity, we should be focusing on a visualisation’s interpretability, particularly given the potential cognitive constraints imposed on our audience due to stress, sleep deprivation, or a lack of analytical literacy. While you can never guarantee success, keeping it simple helps prevent your audiences from drawing the wrong conclusions.

— Ryan