A low turnout, difficulties in polling ethnic minority voters, a surge in tactical voting and a late swing to the Greens are being examined by pollsters, after an election campaign in which Labour’s vote share was widely overstated in the polls.
It was perhaps the most polled election in British political history, with surveys consistently showing a Labour landslide and devastating losses for the Conservatives. While that is the overall result voters delivered on 4 July, Labour secured a lower vote share than polls had predicted.
The polling average for Labour going into the election stood at about 39%, while Keir Starmer’s party actually secured 33.7% across the nation.
A detailed dissection of the discrepancy is planned, with a meeting in early September of British Polling Council members, the body representing pollsters, for a “sharing lessons” exercise.
However, pollsters already point to a number of small factors that might have combined to create the disparity. “It could just be one of those occasions where all the errors are small, but they all went one way,” said Peter Kellner, the veteran pollster.
The main factor pollsters cite is the relatively low turnout, which might well have disproportionately affected projections of Labour’s vote share. The final Opinium poll for the Observer gave Labour a vote share of 40%. The company had already changed its models in an attempt to take account of a tendency for Labour’s vote share to be overstated.
“The thing that really wrong-foots polling is turnout,” said Adam Drummond, research director and partner at Opinium. “We had turnout of about 60% – 7 points down on last time. It could be that voters saw what the overall picture was and voted to get their desired results accordingly.
“Therefore, in places where it was a tight Labour-Tory contest, they voted Labour, while in places where it was probably a safe Labour seat and not in doubt, they felt able to vote in other ways, or not at all. Low turnout does seem to have disproportionately affected one party.”
Another problem appears to be the success of independent candidates, who focused on concerns about Labour’s position on the war in Gaza.
Several figures said that the polling industry simply did not have good enough surveys on the behaviour of ethnic minority voters, overstating the numbers in which they would vote for Labour.
Luke Tryl, the executive director of the More in Common public opinion and polling company, raised the issue at a conference last month.
“I just don’t think our samples across the industry of [Black, Asian and minority ethnic] voters are good enough,” he said. “We saw it a little bit in the West Midlands mayoral race. We all underestimated the share of the vote of [independent candidate] Akhmed Yakoob, who was standing on a more pro-Gaza, independent platform. He got more than twice as much as anyone predicted, so there’s clearly something there.”
A late swing is also being examined, amid signs that some voters may have switched from Labour to the Greens in the final days. “It does seem from all of the evidence that there was a late breakaway from Labour among some of its voters, or some people not voting,” said Tryl.
The degree of tactical voting, apparently deployed by a record number of voters at the election, might also have further chipped away at Labour’s vote share. Voters initially saying they would back Labour could have opted instead to support the Liberal Democrats in order to remove the Conservative in their area.
“In 1997, the Tories lost 30 seats to tactical voting – 20 to Labour and 10 to the Lib Dems,” said Kellner. “My guess is that, when we look at this in detail, it was a much higher figure of 60, 70 or 80 seats [in 2024]. Whether the tactical voting websites played a role, I don’t know – we didn’t have that in 1997. But the degree of tactical voting was clearly huge.”
