Polls, maps and graphs
Despite the fact that a change of government after 14 years is imminent, the 2024 General Election campaign has felt a bit anticlimactic. This is largely due to the result being a foregone conclusion. Labour’s unassailable poll lead has been unaffected by the calling of the election back in May, where historically most poll leads narrow the closer we get to polling day. The campaigns have been a bit lacklustre, unless your name is Ed Davey, with a series of debates that haven’t done much to shift the dial. Any speculation surrounding the results have focused more on how big Labour’s majority will be, and which prominent Conservative will find themselves out of a job following July 4th.
In the absence of a competitive and nail-biting election, there has been increased coverage of the polling that has taken place over the past six weeks, to fill the vacuum left by the lack of contentious policy debates.
Days before the election, Cratus Group were delighted to host a pre-election webinar with pollsters Focaldata to explore their own data, as well as the trends and indicators to look out for.
So what did we learn?
What is an MRP model?
MRP stands for Multi-level Regression and Post-stratification which just emphasises how sophisticated and advanced the UK’s polling and research methods market is. Think of this model like a large sheet that takes into account your age, gender, voting history and other variables like ethnicity and the number of cars you have – before multiplying it to generate a prediction for each of those archetype individuals in every constituency in the country. This collates an enormous sample of data to expose relationships between people’s characteristics and their voting intentions. This data is then used at the constituency level to predict the outcomes of seats based on the concentration of the various types of voters who live there.
The level of marginality in this election is unprecedented
A seat is considered marginal when there is a 5% gap between the incumbent and the closest challenger, meaning a 2.5% swing would be enough for the challenger to claim victory. At this election, almost half of the seats (52) that the Conservatives are predicted to win are considered marginal. If they were to lose every single one of these marginals, then we would be looking at a Labour majority of around 330 seats. This helps to explain why the projected Labour majority ranges from anywhere between 150 to 300, depending on the outlet, as these marginals are simply too difficult to predict. 17% of the seats available are being considered too difficult to predict as they sit within the margin of error. Nobody can reliably estimate just how big Labour’s victory will be, as so much hinges on these extremely tight races.
Labour’s ‘sandcastle’ majority
In this era of hyper-politics, voting has become more transactional and we are seeing more and more voters lending their ballot to a political party only to withdraw this support at the next election. The Conservatives built an impressive coalition of voters during their four successive electoral victories, acquiring a different type of voter at each election: akin to Thanos collecting the Infinity Stones that allowed him to wipe out half of the universe. In 2010, we saw Cameron win the trust of middle class professionals, before the low-educated voters in the South were picked up in 2015 along with West Country liberals in 2017 and finally the ex-Labour post-industrial votes in 2019. Labour is set to speedrun this process in one electoral cycle and pick up a loose coalition of transactional voters in what will be a large majority, but one that could be washed away just as quickly as it was constructed.
If you found this interesting, you can watch the webinar in full here to learn more about Focaldata’s methods, and how tomorrow’s election will play out in practice.
