Why election polls can be completely wrong

Unlock the Editor’s Digest for free

Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.

Good morning. With Stephen Bush putting his feet up for a little longer, I’m your guest columnist today.

The big question for 2024: who is going to win the UK general election? (I think it’s likely to happen this autumn as I still can’t believe Rishi Sunak will go to the country in May.)

Of course, I can’t actually answer that question. But I can explain why the final result may look nothing like current opinion polls.

On the face of it, Sunak is in the last-chance saloon: his personal ratings are unimpressive, his Conservative party is unpopular and ill-disciplined after 13 years in power, and the country is still mired in a cost of living crisis with much higher mortgage rates than a few years ago.

And yet.

Morgan McSweeney, Sir Keir Starmer’s director of campaigns, gave a sobering briefing to the shadow cabinet ahead of Christmas. He highlighted various recent elections — abroad and in the UK — which defied almost all predictions when voters actually went to the ballot box.

I’ve got a copy of his slideshow: “How to Predict the Weather: The Polling Leads Edition”. It makes for fascinating reading.

Things can’t only get better

Of course, McSweeney is in the game of expectation management and trying to prevent a damaging surge in complacency within the Labour party.

He has cherry-picked some of the most eye-popping examples of elections where there have been major swings in the final weeks of the short campaigns.

But the core point is this: when people talk to pollsters at this point in the cycle they are usually making a comment about the ruling party and whether they are doing a good or bad job. The closer you get to a general election, the more they start to ask a *different* question: do I prefer the alternative(s), or not?

Remember: many, many voters do not follow politics very closely outside of election campaigns. There is still a large slug of undecided voters who could — just possibly — come to the aid of Sunak’s Conservatives.

Everyone remembers how, in the white heat of the election contest, Theresa May squandered a huge majority in 2017. They also know that Neil Kinnock’s Labour was far ahead in the polls in 1992 before that “lead” crumbled to dust.

But even the Labour landslide of May 1997 holds salient lessons for today.

Some polls put Tony Blair’s Labour party more than 30 points ahead of the Conservatives in May 1996, a year from the big election. Even a week before the vote, there were surveys putting Labour about 20 points ahead of the Tories. On the actual day Labour won by a much more modest 13-point margin.

It’s also easy to forget the polling patterns at the last general election. In 2019, Labour enjoyed an apparent advantage over the Conservatives for several months before May was ditched and new leader Boris Johnson led the party into a crushing election victory. 

The McSweeney slideshow concludes with a final comment saying: “Polls do not predict the future. Nobody has voted in the general election.”

Here are some of the most striking examples from the presentation:

Australia, May 2019:

Australia’s Liberal party won a “miracle” victory after the left-leaning Labor party had led almost all opinion polls for more than two years. The chart shows Labor with a lead of about 10 points in the autumn of 2018, before the two lines converge, with a sudden switch to the Liberals at the end of the campaign.

US, November 2016:

Analysts, bookmakers and journalists were left wrongfooted by Donald Trump’s presidential election victory for the Republican party in November 2016. Polls had consistently put Hillary Clinton ahead, with a lead of about eight points a month from ballot day. (Although there was one brief moment in September when the polls *did* put Trump ahead.)

Spain, July 2023:

If the Spanish polls were to be believed in the run-up to last July’s election, the centre-right PP was cruising for victory over the leftwing PSOE party with a consistent lead of about seven points. Instead, the final result was on a knife-edge, with 33.1 per cent for PP and 31.7 per cent for PSOE. Several months later the latter was able to form a coalition government with other smaller parties.

Germany, September 2021:

The story of the German election of 2021 is how Angela Merkel’s centre-right Christian Democrats (CDU) threw away a poll lead of 24 points over the left-leaning Social Democrats (SPD).

At one point in 2019, the SPD was only the fourth most popular party in the German polls, on a meagre 13 points. Yet Olaf Scholz’s party somehow emerged victorious in September 2021, confounding forecasters. The FT dubbed it “one of the most remarkable turnarounds in recent German political history”.

Norway, September 2017

Norway’s centre-right party won a historic re-election amid an economic upturn, shattering the hopes of the left-leaning opposition Labour party. Labour plunged from 40 per cent in the polls in 2015 — a lead of about 20 points — to a final election result of 27 per cent. That was its second-worst performance in nearly a century.

Now try this

When your job involves endless reading it’s sometimes hard to summon up enthusiasm for books.

That’s one reason I tend to veer towards novels which are 1) short and 2) clever, but not hard work.

During 2023 I enjoyed four classics from Graham Greene which all fitted the bill. (My favourite being The Quiet American.)

But the book I finished the year with, and which has lingered for days in my mind, was Act Of Oblivion by Robert Harris, one of my favourite authors (here’s the FT review).

Superficially this is an adventure yarn set in the aftermath of Oliver Cromwell’s death and the restoration of the monarchy under King Charles II.

A senior royalist is charged with tracking down the 59 individuals who signed the death warrant for Charles I. The chase takes him across the Atlantic to try to track down two of the final escapees across New England.

The book left me, a history graduate, wondering: why did I know so little about this extraordinary period of the English past?

Top stories today

• Cash builder | More than two-thirds of a UK government fund intended to jump-start housebuilding in England remains unspent more than six years after its launch, despite the chronic shortage of housing. As Jennifer Williams reports, only about 31 per cent of the pot has been spent to date.

• Labour force | Ahead of the general election, Labour has quadrupled its use of consultants, despite a promise by shadow chancellor Rachel Reves that she would introduce “tough new rules” to “slash government consultancy spending” in half.

• Unreformed | The rightwing Reform party is threatening to be a thorn in the prime minister’s side, capitalising on discontent with the Tories. Ahead of the party’s press conference in London today, Rafe Uddin goes behind the scenes.

• Lib Dem hopes | More than 2mn people in London and the South East are set to be pushed into a higher tax bracket — and the Liberal Democrats are hoping to capitalise on any subsequent discontent. New research underlines the electoral threat to the Conservatives in many of its heartland seats.

• Welsh defence | Vaughan Gething has defended plans for the biggest shake-up to council tax in the UK since the levy was introduced in 1993. The Welsh economy minister, a leading contender to become the next first minister, told the FT the current tax was “broadly regressive”.