THERE WAS no shortage of predictions about how Britain’s general election would pan out on July 4th. An enormous amount of data gathered over the six weeks of the campaign—144 national polls in all, surveying a total of 622,000 people—pointed to one outcome: a large Labour landslide. That outcome duly materialised: Labour’s seat haul of 411 seats gives it a majority of 172 in the new Parliament. Even so, the polling firms did not exactly cover themselves in glory.

An average of 17 voting-intention polls conducted immediately before election day suggested that Labour would enjoy an 18-percentage-point lead over the Conservatives. But once the ballots had been counted, Labour’s share of the vote was just 10.3 points ahead of the Tories’. Underestimating the Conservative Party’s support and over-estimating Labour’s is an age-old problem for the polling industry. But this is the biggest miss since the election in 1992. Jane Green, a professor of politics at Oxford University and president of the British Polling Council, says it is “too soon to know why” the pollsters were off target. Even so, theories abound.

Polling firms have a torrid time anticipating who will actually bother to vote. Turnout of just 60% was the lowest since 2001. New voter-ID laws that require proof of identity to cast a ballot may have had a small effect on turnout (at local elections in May 2023, 0.25% of people who tried to vote were prevented from doing so). But that is likely to have mattered much less than muted enthusiasm for Sir Keir Starmer, Britain’s new prime minister, and a high degree of confidence about the eventual result. These factors may have resulted in younger people who told pollsters they would vote Labour turning out in lower numbers than expected (or casting their votes for other parties).