June 10, 2017

Quick data visualisations – taking inspiration from election graphs

By Paul Laughlin

quick data visualisationOne of the understandable concerns raised, when I train analysts, is how they can produce quick data visualisations.

After we study the design principles and options to consider, it can seem that to do a quality job takes longer than they normally have. Many analysts complain that timescales are squeezed and corporate life increasingly frenetic, as leaders just want ‘good enough’ answers and want them now.

However, experienced data visualisers can manage to produce effective data visualisations rapidly. Not every engaging graph needs to be a design project, that has gone through extensive refinement. During our month focussing on sharing data visualisation resources, I hope a more pragmatic look at rapid versions might help too.

So, to provide some examples of data visualisations produced within time constraints, I thought we’d look at the analysis of UK election results. What did media companies manage to commission & publish within 24 hours of data being available?

From overnight election coverage to maps for the nation

Let’s start with the BBC coverage. As well as the interactive graphics used during the election coverage overnight, this morning they published an engaging set of maps to explain the results. In this post they include national maps, basic interactivity (selection & filters), as well as zoomed maps for Scotland & London.

Not a diversity of mapping types (probably given the timescale), but good to see use of a density map with binning into hexagons – to accurately show number of constituencies.

This is a clear example of how some basic mapping variants, with basic interactive features (or use of side-by-side comparisons), can be very effective visuals to get out quickly:

Election 2017: The result in maps and charts

The Conservatives remain the largest party in the House of Commons on 318 seats, not quite the 326 needed to win an outright majority. Theresa May’s party lost 13 seats, while Jeremy Corbyn’s Labour gained 30. Select the “Changed seats” button below to see how the UK’s political geography has changed overnight.


Considering different attributes of voters & their behaviour through graphs

Even more rapidly published, was this set of charts from the Financial Times. Given my emphasis on the power of scatter plots, when used correctly, it is good to see their use to present all constituency data points in several of these graphs. Together with a couple of other chart types, to show migration of voter allegiance & effective targeting of marginal seats.

This is an impressive collection to have published in the time period. I think the plot of labour gains in marginal constituency is a particularly effective visualisation.

Election 2017: how the UK voted in 7 charts

It is impossible to discuss the dynamics at play in this election without considering the fallout from the EU referendum. Brexit may have become less prominent in the latter weeks of campaigning, but it still looms large over the political landscape and will undoubtedly have played on many voters’ minds at the ballot box on Thursday.

Unfortunately the full article is limited to FT subscribers, so here are some of the charts I thought worked well.

quick data visualisation 1


quick data visualisation 2


quick data visualisation 3

Statistical analysis is also aided by quick data visualisations

Finally, I’ll turn to a statistical expert. Andrew Gelman regularly blogs on statistical methods, correcting academic papers & generally working to improve standards in the industry.

In this piece he usefully turns to what is often a news staple after election results. The tabloid story is normally that the ‘pollsters got it wrong again‘. However, as Andrew rightly shows in this short post, they actually had a fairly good night. Forecasts were not far off this time & polls had picked up the ever closing gap.

It is interesting to see that even the use of two basic scatter plots, plus use of colour, can convey useful information and truthfully display the full data available.

Statistical Modeling, Causal Inference, and Social Science

The Conservative party, led by Theresa May, defeated the Labour party, led by Jeremy Corbyn. The Conservative party got 42% of the vote, Labour got 40% of the vote, and all the other parties received 18% between them. The Conservatives ended up with 51.5% of the two-party vote, just a bit less than Hillary Clinton’s share last November.

Don’t give up on producing quick data visualisations

I hope some of those examples inspired you. In a month when I’m sharing a range of data visualisation sources & resources, I hope you’ll be encouraged to use them.

As you can see from some of the examples above, even a simply scatter plot graphic, produced quickly after an event, can help inform the debate. Try your hand at quick data visualisations, you may be surprised how much they help.

When decisions need to be taken quickly it makes sense to enable people to use their fastest reasoning (and studies show that is visual not cognitive).