Data visualisations: why facts don’t speak for themselves

William Allen calls for a robust debate of how data are presented.

Whether on the news, in magazines, or through social media, we increasingly see data communicated through visualisations, which represent and present data to facilitate understanding. These may be static, unchanging charts or interactive animations and websites. In 2008, Chris Anderson proclaimed the ‘end of theory’ in Wired Magazine, arguing that ‘with enough data, the numbers speak for themselves’. It might also be tempting to think that visualisations, like the data underlying them, also are self-evident, clear, or objective windows onto reality. But is this the case?

Making visualisations involves lots of different people and processes, and this influences the data they communicate. The creative and technical work is done by designers, who have different kinds of training, motivations, and backgrounds that influence how they do their jobs. Meanwhile, they may work on behalf of clients in the media, non-governmental organisations, or governments themselves. These organisations may also hold different values and have agendas for how their output gets used. Visualisations also have many purposes that influence their content, which can enable viewers to ‘read’ specific values, or ‘feel’ particular emotions.

What this means is that data visualisations can be persuasive, even political. They give particular views of data and the phenomena those data represent, sometimes making reality seem more neat and simple than it actually is by excluding aspects or whole populations, intentionally or not. They also can tell and frame stories, leading viewers to particular conclusions.

In 2011, Simon Scarr produced a visualisation called Iraq’s Bloody Toll for the South China Morning Post, seen below. It shows the number of civilian deaths on a monthly basis from March 2003, when US-led forces invaded Iraq, through to December 2011. Unlike typical bar charts however, the data were plotted upside down. The bars were also coloured deep red. The suggested visual imagery is unmistakable: when combined with the evocative title, this visualisation has a clear message about the human lives lost over that period. In an interview, he later reflected that ‘a few people in the newsroom suggested splashing drops of blood on the page but we resisted all unnecessary artistic touches’.

“Iraq’s Bloody Toll”, a visualisation designed by Simon Scarr appearing in the South China Morning Post in 2011.

After it was published, the visualisation stirred up some debate among visualisers. Was it misleading? Did the unusual orientation introduce confusion rather than clarity? The fact that this graphic created such conversations about the role of visualisers underlines how they are not neutral; they are also used to persuade. Visualisations depend on their purpose and intended audience. Therefore, some are therefore more effective at achieving their outcomes than others.

This is perhaps best illustrated by a 2016 response to Simon’s work, produced by Andy Cotgreave, another visualiser. He turned the graphic upside down to reflect the conventional orientation of bar charts, changed the colour to blue, and renamed the piece. His version is seen below and titled Iraq: Deaths on the Decline It suggests another narrative through only these three changes.

“Iraq: Deaths on the Decline”, a visualisation by Andy Cotgreave, appearing at ComputerWorld.com in 2016.

So, what can studying data visualisations contribute to our understanding of free and open debate, particularly in journalistic settings? Increasingly, the public gets information about many topics through visual forms. If data and representations of data are not neutral, instead offering particular windows or viewpoints, then we need to be conscious and aware of these messages.

Furthermore, we should welcome debate about the data including how they were collected, what might be missing or overlooked and how their visual depiction affects the interpretation of the data. Reflecting on the role of data in society, Cotgreave remarks ‘let our data express opinions and viewpoint. Then let’s have mature conversations about the data, not dangerous dismissals…My interpretation of the data might not agree with yours, and that’s ok. This isn’t lying with statistics: it’s having a data-driven conversation’.

Enabling and facilitating public conversations about data visualisations is an important part of sustaining and promoting debate and is therefore an essential part of free speech.

William Allen is a 2017 Dahrendorf Scholar, investigating the ways that data visualisations about migration communicate ideas of freedom. He is a Research Officer with the Centre on Migration, Policy, and Society (COMPAS) at the University of Oxford.

Read more:

Leave a comment in any language

Highlights

Swipe left to browse all of the highlights.


Free Speech Debate is a research project of the Dahrendorf Programme for the Study of Freedom at St Antony's College in the University of Oxford. www.freespeechdebate.ox.ac.uk

The University of Oxford