POLITICS & INSTITUTIONS - ECONOMY

Keeping data visualisation honest



Better serving citizens in the EU was the main focus of the first edition of EU DataViz, which brought together hundreds of participants to learn more about technology and trends, join thematic sessions and see data visualisation across a wide range of applications.  Shutterstock

Better serving citizens in the EU was the main focus of the first edition of EU DataViz, which brought together hundreds of participants to learn more about technology and trends, join thematic sessions and see data visualisation across a wide range of applications.  Shutterstock

Award-winning expert in data visualisation and former visuals desk lead at the Guardian, Xaquín Veira González, was in Luxembourg on Tuesday for the first EU DataViz conference, providing tips on how to stay ethical with data visuals, which are often subject to manipulation and bias. 

González--who has also worked at National Geographic, New York Times and El Mundo--kicked off his presentation with a concrete example which went viral in 2016 which showed Pablo Casado, People’s Party leader in Spain (formerly its vice secretary general of communication), showing a graphic on Teleivision Española, the data for which made it appear as if social spending had been on the rise since 2011, year after year. It’s something González called “the overworked intern mistake”, but a basic one: “the visuals and numbers don’t match”. 

The party even published the graphic on its social media--which launched a cascade of replies and reworking of the graphic, so that the data and numbers did actually match. The party eventually corrected its mistake, but others responded in turn--providing, for example, graphics with inflation factored in, presenting a more realistic view of the situation. 

As González explained, the creator of the original data visualisation ignored a “basic charting rule”, that the chart should have started at base zero. It might be one thing to manipulate data for an advertisement, but it’s quite another when it comes to public discourse. “If a visualisation is meant to inform the public, it should be honest,” he added. 

Informing--and better serving citizens in the EU--was the overall focus of the first edition of EU DataViz, which brought together hundreds of participants--mainly from the public sector but also researchers, journalists, academia and the like--to learn more about technology and trends, join thematic sessions and see data visualisation across a wide range of applications. 

Among González’ takeaways were that, in order to avoid misunderstandings or bias in data visualisation, it helps to be aware of the pitfalls across the lifecycle--from collection through analysis, to the visualisation itself--and, of course, the final story the data is helping to tell. Question, for example, whether correlations being made are legitimate, be transparent and be aware of the visuals aligning with words in the story, he argues. 

It can be fine to get creative, but not just for creativity’s sake: always remember there’s a story to tell. In another set of examples, González compared a well done example, “Iraq’s bloody toll” chart by Simon Scarr, which captured the lives lost over nearly 9 years of American military engagement in the country. The visualisation has high impact, the use of red reminiscent of dripping blood. Reuters produced a similar, albeit poorly annotated, visualisation, leaving the reader confused with the white space as to whether deaths by guns rose or dropped significantly after the enactment of Florida’s “Stand Your Ground” law. 

To keep up with data visualisation, González recommends Junk Charts or, for a more humourous take on visuals (but one which González warns may take you “down a rabbit hole”), Tyler Vigen’s site on spurious correlations