Why do we visualize data? Do data visualizations aim to inform audiences effectively? Or do they simply aim to catch people’s eye, providing the just gist of the data? This is a question which has been hotly debated by some of the leading authors in the field of data visualization recently.
Data visualization is a spectrum, determined by your data, your objectives and your audience. Whichever visualization style you choose, it should be the most effective one for the purpose at hand. Most of the time people are trying to share detailed insights into their data in the quickest and most efficient way. This approach is a great starting premise.
However, to limit data visualization to just being used for this purpose is stifling and ignores a vast array of other perfectly valid reasons we might visualize data.
Consider Stefanie Posavec, one of many incredibly creative people who use data visualization to make art. My favorite are her wearable data objects, designed in collaboration with Miriam Quick. One of these is a Perspex necklace representing a week’s worth of data from air quality sensors in Sheffield, UK. Wearing the necklace, the wearer can literally feel how air quality changes in a one week period (the big red spike in air pollution was Bonfire Night).
Is it data, visualized? Yes. Is it the most effective way of showing the data?
I asked Stefanie about the purpose of her work. “We were exploring new approaches to making data memorable and accessible,” she told me. “For people who don’t care about data, which will make the most impact, the wearable or a bar chart?”
What about the importance of informing people?
“I’m always trying to inform my audience, but the level of information can vary from the gist to something more detailed and in-depth,” she said.
Examples from data art reveal the reasons we need to be flexible in our definition of data visualization and our approach to visualizing our own data.
The purpose of a visualization will also determine the extent to which you should inform effectively. Are you creating an operational dashboard which will be viewed daily by managers? Will those managers need to see, in an instant, which areas need attention? Will they need to interact, drill down and see more details in the data to make actionable insights? If that’s the case, then wearable art is clearly the worst thing you could do! You will instead be focussing on crisp, clear visuals, such as bar charts and trend lines.
However, operational dashboards are not the only reason we use data visualization in our organisations. Like Stefanie, sometimes we’re only communicating the gist of the data. Sometimes it’s more important to make someone engage with the overall message rather than the minutiae. In this case, you can consider moving away from the “functional” end of the visualization spectrum to the more “beautiful” end. I’ve written about this in the past, and the books you can read to help learn more.
Maybe you’re going to be doing a data-driven presentation for your managers. You’re going to be in a room, with a projector, showing and describing your insights. You’ll want to focus on just the relevant numbers, not all the numbers. You probably don’t need to label everything as accurately as you would for a printed or emailed visualization. You might also want to simplify the charts too. Each of those choices reduces the effective information transfer of the chart as a standalone piece of work, but that’s fine in a context where you will add the missing pieces through your own commentary. Before you judge others’ work, you should consider the context for which it was designed.
Consider Charles Minard’s amazing 1869 chart depicting Napoleon’s disastrous 1812 Moscow campaign. This is a masterpiece but would it be effective in a presentation about the effect on Napoleon’s army? No, it would take too long to explain it.
Charles Minard's depiction of Napoleon's 1812 Russian Campaign
How about the example below (from Jorge Camoes)? It’s a simplistic reduction of a complex military campaign to single fact, but in the right context, maybe that’s all you need?
Remember -- data visualization offers a spectrum of choices. Your role is to choose the one that best fits your data and your audience. This still allows for creative, art-like pieces of work in your organisation.
This post was inspired by a fascinating blog by Stephen Few, asking “What is the best response to bad practices?” It’s a great question and generated a fascinating thread of comments. I wholly recommend you read it.