Britain Predicts: Who would win if an election was held tomorrow?

Country/area: United Kingdom

Organisation: The New Statesman

Organisation size: Big

Publication date: 3/9/2021

Credit: Ben Walker, Michael Goodier and Josh Rayman

Biography: Ben Walker is a data journalist at the New Statesman.

Michael Goodier is a data journalist at GlobalData & New Statesman Media Group.

Josh Rayman is a visual journalist at GlobalData & New Statesman Media Group.

Project description:

Britain Predicts attempts to forecast how an election would pan out according to the polls if held today. Our unique model recognises that the UK doesn’t often vote or swing as one, and that there are people in some constituencies who are more susceptible to changing their vote than those living elsewhere.

Impact reached:

Britain Predicts now sits as one of the UK’s most popular election calculators.

Techniques/technologies used:

The model was written from scratch based on the documentation of the Strong Transition Model from Electoral Calculus (electoralcalculus.co.uk) and other work from The New Statesman done in previous elections. The calculator was written in JavaScript so it could be sent across to the client allowing for a more tactile approach as opposed to sending the data across to a server to calculate and return a result.

The graphical control was built with a sense of play in mind, with a big circular slider component that is designed to encourage repeat usage. This also removed the issues we had in previous election projects, where each slider affects the other sliders (to sum up to 100%). The slider makes it clear how modifying one party impacts the others.

The project was built with Svelte and Layercake (d3.js), and the data is updated using an AWS Lambda scraper that polls a Google Spreadsheet, so the default poll values can be updated daily without having to redeploy the project.

What was the hardest part of this project?

Like any other model, our model has limitations and assumptions. We had to make some tough calls, such as the decision not to link the Reform Party to its predecessor, the Brexit Party, which can lead to some unusual results such as Reform gaining seats in Scotland if they are given a high enough share. We also had to split the model up to treat Scotland and Wales separately, as they have single-nation parties (the SNP and Plaid Cymru).

What can others learn from this project?

The project builds on previous work to improve our collective understanding of how people vote in elections and how we can predict the results of those elections. It also highlights the uneven distribution of seats in parliament, with some parties being over-represented while others get fewer seats.

To contextualise changes in polling figures, journalists can also use the interactive part of their calculator to input their voting scenarios and see how many seats each party would get.

Project links: