2020

UK General Election 2019

Category: Best data-driven reporting (small and large newsrooms)

Country/area: United Kingdom

Organisation: The Economist Newspaper

Organisation size: Big

Publication date: 10 Jan 2019

Credit: James Fransham, Martín González, Evan Hensleigh, Matt McLean, G Elliott Morris, Dave McKelvey, Dan Rosenheck, Marie Segger, Alex Selby-Boothroyd

Project description:

The Economist’s data-driven coverage of the December UK general election. This included our own poll aggregator; detailed analysis of YouGov’s “MRP” model of the election to look at the effect a surge of support towards the Liberal Democrats might on the election outcome; an innovated series of dynamic election maps; and election night live coverage, including a live forecasting model that updated through the night; plus additional data-led stories.

Impact reached:

We set out to provide as much insight and analysis of the snap UK general election as possible. We wanted to publish data-driven, visually stimulating stories that worked well in print and online, and that got noticed among the torrent of election coverage. Internally, we began from first principals: what information do we have at hand, and what is unlikely to be found elsewhere. We wrote a piece on the accuracy of polls that nodded to innovation in pollsters methods (MRP methods). We commissioned five high-quality constituency-level telephone polls with Survation, to inform on local-level reporting. Finally, we created a national poll tracker; innovative dynamics maps of historic electoral behaviour, and an election-night forecast model. Our coverage was lauded both internally and externally. 

Techniques/technologies used:

As with much of The Economist’s data team’s project we used a wide variety of tools. Our data journalists rely on R and Rstudio, furnished by a suite of different packages. Our visual journalists work with R and Adobe Illustrator, along with some proprietary visualisation tools. Our front-end visualisers mainly use D3.js. We use cutting-edge and robust statistical techniques to make the best use of data, such as our own MRP analysis for the “Graphic detail” page. 

What was the hardest part of this project?

While a UK election had been rumoured for sometime our expectation was that it was likely to happen in the New Year. For that reason the timing and resource allocation was not wonderful, and we had to scramble a little bit more than we would have liked. Aligning the interests of different mediums, the tensions between print and online, both in scope and timing of what is possible always brings challenges. Yet, for example, our print-edition “Graphic detail” which made use of our internal MRP modelling, along with existing pollster analysis, helped inform our editorial endorsement of the Liberal Democrats. 

What can others learn from this project?

The project combined data-driven reporting from different perspectives. The local constituency polls alongside the national poll-tracker provided interesting insights. For the poll tracker, we scraped and hand-input national polls in order to analyse voting intentions by gender, education and by region. The project shows how these different prisms of reporting can be combined with innovative data elements—and deliver informative coverage of a pivotal political event in Europe. 

Project links:

projects.economist.com/uk-elections/2019/general-election-results

www.economist.com/britain/2019/11/07/how-britains-pollsters-have-changed-their-methods

www.economist.com/britain/2019/11/07/the-conservatives-are-struggling-to-win-a-crucial-midlands-marginal

www.economist.com/graphic-detail/2019/12/07/voting-lib-dem-could-hurt-the-tories-as-much-as-labour

medium.economist.com/forecasting-britains-election-in-real-time-bfcb8d395fa2

www.economist.com/graphic-detail/2019/12/13/britain-votes-resoundingly-for-boris-johnson