The interactive team was at the heart of the Times’ and the Sunday Times’ political coverage before, during and after the election. What follows is a snapshot of some of the visual work we are most proud of. Our mammoth and unprecedented turnout project looked at voter interest in every constituency going back to 1918. Our stunning ‘ring map’ of the seats which always predict the general election told a thousand narratives of how political power has . Finally, our post-election charts – some of which went viral on the Friday morning – formed part of a highly engaging analysis
The work of the interactive team consistently ranked among the most-read and most-engaged piece of the election. All three of the pieces below had very high engagement among our readers, who praised the complexity and clarity of our work in the comments. Our concentric circle ‘ring map‘ (for want of a better term) which featured in the ‘seats that always pick a winner’ piece was praised for the number of fascinating narratives held within it. For example, readers were able to explore the country’s swing seats which constantly changed hands at each election, or see which seats Labour had won in 2005 but never since, thus showing how far away the party is from a sizeable majority. For the charts built on the morning after the election, being able to script them beforehand meant we could finish and share the charts before the results had even been declared. This gave us an early-morning viral hit when the Times Red Box editor shared the “arrow-swing” map. Finally, our work across the election also raised the profile of our team internally, which has resulted in a huge increase in commissions and requests to work with us since.
Our election coverage helped us to develop some truly innovative tools and techniques. Much of our data analysis for the different pieces was done using R, the standard programming language of the Times interactive team.
However, we also built a new responsive graphics tool for the election which allowed us to upload a set of images of different sizes. This would produce embed codes for our CMS which would serve readers different sized images depending on the size of the reader’s screen. This proved to be a far less buggy approach than using Ai2html, as labels don’t jump around unexpectedly on static images.
Preparing for the morning after the election, we wrote R scripts that would automatically download the election data and produce a set of complex visuals with ggplot in the Times style that we wouldn’t have been able to do in Datawrapper: things like arrow swing maps, histograms and “brick” charts, as well as several that we didn’t end up using. We then wrote a set of functions which woulid export these charts as different sized images for various screen sizes, meaning we could upload directly from ggplot to our responsive graphics tool. The function had a built-in faceting option, which meant that two charts side by side on the desktop image would be stacked vertically for mobile, complete with appropriate text sizes and labels.
All of these developments meant we could speed up the production of quite complex visuals. The rest of the visuals were a combination of Datawrapper charts and bespoke D3 interactives, which we used to add personalisation to the stories.
What was the hardest part of this project?
The turnout project was particularly challenging. We had to factor in hundreds of boundary changes into our analysis, while digitised maps featuring constituency boundaries back to 1918 are not readily available. This meant we had to do a lengthy process of manually checking which historic constituencies best corresponded with the current ones, and weight the average turnout proportionally across the constituencies it straddled. As we later found out, some constituencies changed boundaries without being renamed, making it even harder to spot changes. If we were to do this again, we might want to consider working more closely with the House of Commons library to see if there’s any way they can digitise historic election maps for us. Nevertheless, the result was a unique piece of research unlike any of its kind which we hope others will be able to build on in the future.
What can others learn from this project?
We hope that others take some inspiration from the visuals we’ve produced throughouot this election, just as we’ve taken inspiration from others before us: we couldn’t have done the red wall “brick” chart in the results piece without the work of twitter map legend Alasdair Rae, for example. On the technology side, we think we’ve found a reliable and quick way of getting graphics built with R and other tools onto our website as responsive visuals that others might want to emulate. Finally, we think the team has set a clear example of how to successfully cooperate with a large newsroom of busy journalists. On the morning after the election, the interactive team became the focal point as busy reporters and editors looked to put the Tories’ win into context.