Close panel

Close panel

Close panel

Close panel

Visual arts 26 Oct 2016

A love of maps should mean using fewer to illustrate data better

As any visitor to my house will be able to tell, I love maps. Our living room is full of them. A Soviet map of Portsmouth (complete with Cyrillic labels) gazes across the room at old maps of South America and Italy. A large Mercator world map occupies an entire wall while Charles Minard’s celebrated map of Napoleon’s ill-fated Moscow campaign hangs in the hallway.

Judging from the reaction, I am not alone. And with the US election imminent, maps full of data will soon be hard to avoid as news organisations attempt to paint a picture of the political lie of the land.

Sometimes one map is not enough. Last year, for a Financial Times article focusing on Ukip’s challenges after the UK general election, my colleague Chris Campbell worked on a map showing the party’s second place finishes. The map was designed in the cartogram form often used for political maps – rather than prioritising geographical accuracy, it represents the constituencies as squares shown equal in size to reflect their equal value in the House of Commons. Those shaded purple show where Ukip finished second.

Having completed the map, Chris realised that this, by itself, did not provide enough information because a second-place finish is ambiguous – for example, it could be seen as either a good thing for Ukip if they were expecting to finish third or a bad thing if they were hoping to win the seat.

Chris extended the graphic to include additional maps of Ukip’s first and third place finishes. This revealed a lot more geographical information – the solitary win in Clacton, swaths of third-place finishes across much of the UK, a general absence in Scotland – but still some lingering questions. For example, what about the London-shaped hole in Ukip results? And how did the party’s performance compare with others, particularly the Liberal Democrats, Britain’s traditional third party?

So Chris tried a full set of first, second and third place maps for all the main UK-wide political parties – 12 maps in total. Suddenly the final graphic became compelling enough for it to assume central position in the story. Readers were invited to linger on the maps and discover any number of geographical patterns in the election results themselves. Maps in “discovery mode” like this are often irresistible.

“Only when specific locations or geographical patterns in data are more important to the reader than anything else

As well as political maps, we can also use topographical maps for local stories. Earlier this year, Steve Bernard, another colleague on the FT graphics desk, produced a striking animated map to illustrate what might happen the next time a big earthquake hits Japan – this rapidly became the FTs most-watched video on Facebook, highlighting the broad appeal of maps when used to tell a compelling story.

However, maps are not the solution to every visualisation problem with geographical data. As the FT Visual Vocabulary highlights, “spatial” is just one of many possible relationships in data and maps and should be used “only when specific locations or geographical patterns in data are more important to the reader than anything else”. Ignore this advice and maps may end up leaving readers lost.

Recently, some painstaking research by Laura Noonan, the FT’s investment banking correspondent, produced a data set of top banks’ presence in eight cities lining up to take London’s role as the EU’s banking centre. Nearly 5,000 entities, each allocated to one of the eight cities.

When it came to visualising this data, early suggestions revolved around an interactive map – hover over a city to discover more information about the banks in the city.

However, Martin Stabe, the FT’s head of interactive news, has already warned about defaulting to interactive graphics. And putting this information on to a map would presume (incorrectly, I hope) that our readers either did not know where Madrid, Paris, Dublin or Amsterdam were – or that the distance between the cities was of importance to the story (it is not).

Instead, we focused on patterns in the data we wanted to draw out and identified two key aspects: the strength of the general banking presence in each city and the strength of specific banks (Deutsche Bank, Barclays etc) across all of the cities.

It was this latter requirement that ended discussions of a map, because readers would have to memorise bank activity in locations and scan from city to city. Instead, we looked to another part of the Visual Vocabulary, and discussed a simple grid design for displaying the relationship between banks and cities.

The draft grid aimed to show the highest level of banking presence in each city. For example, a bank having a fully formed subsidiary in the city would be shown on the grid as a higher level of presence than just having a branch or nothing at all. Sorting the data – by the highest level of presence – in both rows and columns helped to highlight the key patterns on the grid.

I showed the draft to a few people – a useful suggestion from some was to add the data for London itself, which would provide useful context about how the “challenger cities” matched up. Feedback from colleagues on evolving graphics is valuable; it can validate your design decisions – and they may well see something that you have missed.

Final polish included adding some detail on how to read the chart (columns relate to cities; rows relate to banks) so readers would know how to interpret it.

To twitter

The gridded version is far more effective at communicating those patterns in the data we needed to draw out

Compare the final version of the graphic against how the same information would have looked presented on a map. As much as I like looking at maps, the gridded version is far more effective at communicating those patterns in the data we needed to draw out. In particular, look at how hard it is to identify the trends in a specific bank on the map.

Similar information on insurers was compiled by Oliver Ralph, the FT’s insurance correspondent, which allowed us to reuse the grid design with that data too. Having learnt how to read the first chart, we felt readers would find the second one even easier to digest.

We at the FT are not alone in being sceptical about producing maps in volume. Matthew Ericson, associate editor at the New York Times, wrote a few years ago about maps that he felt should not be maps.

So as lovers of maps, we are keen to create beautiful ones whenever they offer a crucial addition. Truly appreciating them, however, means not defaulting to a map just because you can. Like a lot of things in the world of data visualisation, the right way to use them is to follow the mantra “fewer, but better”.


For more articles from the Financial Times/ FT.com please register here

Source: Alan Smith. 2016. ‘A love of maps should mean using fewer to illustrate data better’. Financial Times / FT.com. October 20, 2016 Used under licence from the Financial Times. © The Financial Times Limited 2016.

All Rights Reserved. Articles sourced from the Financial Times have been referenced and are used under licence from The Financial Times Limited. These articles remain the copyright of The Financial Times Limited and were originally published in 2016. All rights reserved. “FT” and “Financial Times” are trade marks of The Financial Times Limited. The Financial Times Limited has not endorsed, verified or been involved in the creation of the information provided from other sources in this publication, and is not responsible or liable for its accuracy, completeness or content.

Other interesting stories