When well-meaning great-uncles ask me over the Christmas dinner table just what it is that I do for a living, and draw a blank at the words ‘data visualisation’, I’ll often resort to the phrase ‘fancy graphs’ to get my point across. And on a certain level, that’s exactly what data visualisations are. But I feel like this description fails to capture the additional dimension of creativity and innovation that data visualisation brings to the basic principles of charting: the ability to humanise data and tell a story.
Whilst the earliest charts that look and feel like the ones we use today were invented to represent economic data, the earliest examples of ‘data visualisation’ as I understand the term occurred in the field of epidemiology and public health. I think this difference is the human element of health and policy. Where economic data are, for the most part, predictable measures of quantity, value and exchange, data about people and their health and wellbeing require a more responsive and flexible approach that builds upon the foundations of charting to create meaningful, persuasive visual stories.
In one of the earliest example of data visualisation being used in this way, a physician named John Snow created a dot map of fatalities during a cholera outbreak in London during 1854. The arrangement of the dots on the map confirmed his suspicion that the water sourced from the public pump on Broad Street was in some way related to the high rate of cholera infection in the neighbourhood. Using statistics and geography to illustrate the connection between water quality and cholera, Snow is considered to be the founder of epidemiology. I love this example of data visualisation being used to advance public health even before the advent of germ theory because it demonstrates the way in which data visualisation can show strong correlations even when causation hasn’t been entirely determined, and lead to meaningful action. John Snow’s dot map isn’t the first example of its kind, but it was the first to accurately pinpoint and treat a public health risk using evidence-driven means.
Another famous example is Florence Nightingale’s 1857 Coxcomb diagram, a circular histogram that illustrated the number and causes of soldiers’ deaths during the Crimean war. The 12 segments of the coxcomb represent the 12 months of the year; Nightingale colour-coded the segments with red sections representing deaths caused by wounds, black for deaths from other causes and blue for deaths caused by a disease. She presented this diagram in a report to the British Secretary of State for War, making the argument that the conditions soldiers suffered in army hospitals were responsible for more deaths than battlefield wounds. Nightingale’s analysis and persuasive visualisation prompted Queen Victoria to set up a sanitary commission to improve conditions and the death rate fell, but just as importantly, Nightingale’s work set a precedent for the use of data visualisation to influence public policy.
Today, data visualisation is a core component of evidence-based decision-making in government. In my work for the Social Investment Unit, I’ve used data visualisation to map the relationship between MSD’s assessment of need and applicants’ accommodation in social housing, and the Government’s ability to account for the Social Vote at the level of the individual across multiple agencies. At the Ministry of Education, I used a Sankey diagram to show the relationship between school leavers’ NCEA qualifications and their eventual education outcomes. Decision-makers used these diagrams to understand how improving the targeting of government spending could improve peoples’ lives, and return social value over time.
What makes these data visualisations more than just ‘fancy graphs’? I think it’s partly because they are bespoke designs that adhere to the basic principles of graphing information without being restricted by them. In my last blog, I touched briefly on William Playfair’s contribution to chart-making: line charts, area charts, pie charts and bar charts. These are all standard formats that we are familiar with and can read at a glance. There’s nothing wrong with them; they’re highly effective formats for displaying information. But good data visualisation goes beyond simply plotting a chart; it starts with the context of the information and creates an optimal presentation of the data using a combination of charting elements (e.g. height, width, area, slope and angle) to create as much meaning from the data as possible.
In short, the most important factors for good data visualisation are creativity and innovation. If there’s no chart type that makes the best of your data and tells its most meaningful story; invent your own!
The standard set of charts that are available for you to use in your Excel toolbar are there because they’re the gold standard of data presentation methods: versatile, easy to read, and effective. The difference between a graph and a data visualisation is that a data visualisation isn’t necessarily a versatile format, it’s often a situationally specific format. This allows a data visualisation to transcend the numbers and tell a story.
I think the best historical example of storytelling with data is Charles Minard’s Sankey diagram of Napoleon’s invasion of Russia in 1812. Minard, who was the Inspector General of Bridges and Roads in Paris, had used statistical diagrams during his engineering career and continued to experiment with data visualisation in his retirement. In his now-famous Sankey diagram, he displays six factors in two dimensions: the number of troops in Napoleon’s army, the distance they travelled, the temperature, latitude, longitude direction of travel and time. The scientist Étienne-Jules Marey wrote that Mindard’s diagram ‘defies the pen of the historian in its brutal eloquence’.
It doesn’t take much searching online to find contemporary data visualisations that evoke similar sentiments. This map of migrant deaths over time by Curran Kelleher and this time-lapse of gun deaths in the US are incredibly powerful examples. It’s a well-known adage that a picture is worth a thousand words, and these data visualisations are more than just pictures. Although they’re created with more sophisticated technologies than Charles Minard’s paper and ink, these data visualisations are similar in that they are creative and innovative applications of combined charting elements to engage us emotionally with data, leveraging the ability to interpret these elements that have grown universal over the centuries since Descartes developed the Cartesian plane.
Next week I’ll have a look at the intersection of datavis and journalism, and the way in which they complement one another in the interests of quality public information.
Data is beautiful – Sarah
Sarah blogs about how data can be made aesthetic as well as informative.
Want to read more? Try A brief (and incomplete) history of data visualisation, part 1 or more from Sarah.
We run regular business intelligence courses in both Wellington and Auckland.