By Jessica Frank-Keyes
Polls often narrow as the date of an election draws near. If the gap shrinks this time round, will it mean Labour won’t be able to secure the thumping majority it desires? Jessica Frank-Keyes takes a closer look
As the clock ticked towards the end of this Parliament, and the inevitable general election, there was one graph that had come to embody this political moment more than any other.
Two lines: one red, shooting upward then flatlining up high; and one blue, a sharp drop, and then, similarly fixed in place, bumping along – if not the bottom – then the graph’s lower half.
I’m describing, of course, Labour’s consistently resilient poll lead. A lead that prompted acres of newsprint – in this paper too – speculating about the election date. Some pollsters, including YouGov, put it at more than 20 points, while Opinium records a gap of around 14.
Surely, received Westminster wisdom went, Rishi Sunak wouldn’t go to the country when voters appear to be giving him such a clear message about his chances?
Well, we’re now well into that campaign. Holidays have been cancelled, gaffes are aplenty, and July 4 is inching ever closer.
And when it comes to the polls, the crucial question on everyone’s lips is where those two lines will go next. Labour do not need to secure that full 20-point lead to win a majority, and some even argue those around Starmer would welcome a narrowing, in an effort to avoid complacency.
Enter Lord Robert Hayward, a former MP, who pinpointed the existence of ‘shy Tories’ ahead of the 1992 election, who spoke out this week warning Labour’s dominance could be overstated.
He said this was down to how undecided voters are recorded, and the results of the recent local council elections suggested 2019 Conservative voters – who are now ‘don’t knows’ – will both vote in “fairly large numbers” and are likely to once again back the Tories.
“About 33 years on, I am yet again convinced that a statistical bias exists in the polls,” he told the Guardian.
So, could the polls really be getting it wrong? Chris Hopkins, political research director at Savanta, isn’t so sure.
“The idea at this point that there is evidence to suggest there will be polling error is just wrong,” he told me.
The last time there was major polling error was in 2015 – when pollsters failed to predict the Conservative majority at the general election, which saw David Cameron elected to No10.
But since then, experts and data wonks from the British Polling Council (BPC) and Market Research Society (MRS) have looked at what went wrong.
According to a briefing from the House of Commons library, they found significant factors were “simply that the polls had too many Labour voters in their samples, [and] too few Conservative voters, and that the weighting schemes failed to correct for this”.
Hopkins reckons this kind of error “isn’t really happening now… there’s too much of a lead and too much reliance on other metrics for the polls to be substantially wrong”.
He adds: “Ultimately if there is an error, then it’s an industry wide one at this point.
“When we’ve had industry-wide polling errors before, there’ve been eight pollsters, and there are 30 now, so I just can’t see it.” That said: “I might end up with egg on my face on July 5!”
So, if the polls aren’t wrong – or overlooking existing Conservative support – could voters be about to start changing, or making up, their minds?
I asked Kieran Pedley, from Ipsos, who told me he is still predicting a narrowing of the polls.
“On the one hand, can things narrow? Yes,” he said. “But at the same time, most campaigns don’t tend to shift the dial very much.”
What Pedley did also stress was the crucial importance of undecided voters. “There’s definitely scope for things to narrow in the sense of 2019 Conservative voters who aren’t sure to ultimately plump for the Conservatives again.
“But I think what I would stress is it’s not just going to happen… these people are undecided for a reason.
“There needs to be a sense of the Conservatives giving them something to vote for… and they’re going to have to work very, very hard in his campaign to do that.”
His prediction then?
“My assumption since the start of the year is that there’ll be a narrowing, because people start paying attention to politics – most normal people don’t on a day-to-day basis.”
“My assumption since the start of the year is that there’ll be a narrowing, because people start paying attention to politics – most normal people don’t on a day-to-day basis.”
This will certainly be music to CCHQ’s ears. But this doesn’t mean polls will narrow enough for the Conservatives to win – and not everyone’s in agreement.
Hopkins, for one, isn’t convinced.
“I just can’t see it,” he says. “Labour have led on every policy area for a long time, including the economy. Keir Starmer has led on best Prime Minister with us since May last year.
“And it’s been a sizable lead, it’s gone from four points to 12… in a head to head that’s massive. I can’t overstate that.”
One thing we can glean from the polls is the nature of the election campaign yet to come. The Tories, trailing behind, will be fighting hard, with everything to gain, resulting in a splatter of policy proposals in a bid to reclaim ground.
Labour, on the other hand, will be doing its best impression of a waterfowl. Paddling desperately beneath the water as it aims to glide effortlessly along the surface. Everything to lose, and light on policy for fear of upsetting the apple cart.
The latest round of polls landed in hack’s inboxes late last night, with Survation and Redfield & Wilton putting Labour on a 23-point lead, while JL Partners had Starmer’s party on just 12 per cent, down from 15 per cent in early May.
However, experts stress we should view polls in the round rather than look at individual samples – and JLP themselves noted their result would still equal a “categorical” Labour victory.
Disputes about polling accuracy, and whether or not the gap will shrink, or grow, will continue right up until the ballot boxes open.
And until that 10pm exit poll lands, truthfully, your guess is as good as mine.
