Elections sometimes look very different afterwards. I remember sitting on the roof of BBC Television Centre watching the dawn break on the morning after the 1992 election, with Peter Kellner, who was then a journalist who knew a lot about opinion polls, rather than the president of an opinion polling company, YouGov.
I had been his back-up for the election results programme – I was assigned by the BBC to help crunch his numbers and speak into his earpiece if he needed information while on air (he didn’t). My main memory of that night was of scrabbling under his desk trying to connect a loose cable to his laptop while The Unexpected intruded on the smooth preparations for the programme to go live at 10pm. With about 10 minutes to go before the polls closed and David Dimbleby introduced the programme, a rush of late information from the exit poll changed the prediction from Labour as the largest party in a hung parliament to the Conservatives with a small majority.
For the first 90 minutes of the programme, everyone working on it struggled with the exit poll numbers and tried to work out why they didn’t show what everyone had assumed they would show for weeks before. For the whole of the election campaign, the polls had suggested that Neil Kinnock would be prime minister, possibly in alliance with the Liberal Democrats, a new party under a new leader, Paddy Ashdown. Then the results started to trickle in and it became clear that the exit poll was right.
As Kellner and I stared at the rising sun and wondered what Labour had to do to win after four consecutive defeats (meanwhile Tony Blair, the party’s employment spokesman, was answering that question down the line from the BBC studio in Newcastle), we knew that the opinion polls had got it wrong. It wasn’t until weeks later that we understood more about why. Three things had happened. One was that the polls’ samples were too Labour, having failed to keep up with social change over the previous decade. Another was a late swing: people changed their minds in the last day or two. And the final problem was that some people, who became known as “shy Tories”, told pollsters that they didn’t know how they would vote and then voted Conservative.
In pictures: Experts' predictions for the General Election - 04/04/15
In pictures: Experts' predictions for the General Election - 04/04/15
1/10 Andrew Hawkins (ComRes)
“My position has moved: no party can win a majority now. I have also shifted in favour of the Conservatives winning more seats than Labour. That, however, assumes that the current Tory momentum is maintained and that they don’t do anything daft or careless between now and polling day. But the underlying pattern is distinctly in their favour.” (In January he predicted Labour would be the largest party, possibly with a small majority.)
2/10 Joe Twyman (YouGov)
“Probably: a ‘well hung parliament’. Possibly: Conservatives winning most votes and seats, thanks, in part, to SNP gains at Labour’s expense. Speculation: Conservatives unable to form another coalition, not having enough seats with just the Lib Dems, but Labour better placed with SNP and Lib Dems – albeit informally.” (In January Twyman said: “Gun to my head? Labour minority government.”)
3/10 Ben Page (Ipsos MORI)
“Stuck in ‘too close to call’ mode still, made harder by the way votes translate into seats in Parliament. If the parties remain neck and neck, Labour might just end up with more seats, but not a majority. We still have weeks of campaign to go and no clear picture for the marginals, where the polling that is being done suggests a lot of local variations that have plenty of potential to surprise us in May.” (Last time Page said it was a “mug’s game” to make predictions four months before an election.)
4/10 Rick Nye (Populus)
“Since January, the Conservatives have clearly improved on the polls relative to Labour to the point where I’d expect the Conservatives to win the most seats as well as the most votes 7 May. What’s less clear is whether the Conservatives would be able to form a government. (In January Nye expected a hung parliament in which Labour would win most seats but not necessarily most votes.)
5/10 Nick Moon (GfK)
“Something would need to change dramatically for there to be any chance of a one-party majority government. My guess: the Tories will be largest party, but some way short of forming even a two-party coalition. A Labour minority government seems most likely, but I won’t be putting money on it.” (Prediction unchanged since January.)
6/10 Damian Lyons Lowe (Survation)
“On Survation’s public polling, Ed Miliband remains the person most likely to form the next government. However, he’s far from the workable majority figure required . Friday 8 May will remain a day of deals and discussions with other parties to form the next government.” (In January he expected Labour to be the largest party in a hung parliament, by 40-50 seats over the Conservatives.)
7/10 Michelle Harrison (TNS)
“It’s less a case of who wins but who can scrape over the line. Labour polls better on the NHS; the Tories poll better on the economy. Can any claim additional territory from the other over the remaining weeks? Probably not. But our polls show that the public thinks the Tories will be the largest party. In the absence of a firm lead, I’ll go with the wisdom of crowds.”
8/10 James Endersby (Opinium Research)
“Despite the recent weekly statistical ties, we’ve witnessed the faint whispers of movement in the air and a slow, unsteady and shaky sway towards the Tories. How this shift plays out over the coming weeks obviously depends on a huge number of factors. My call, if this holds fast and momentum gathers: Conservatives 288, Labour 267, SNP 45, Lib Dems 24, Plaid Cymru 3, Ukip 3, Greens 2.” (In January he put Labour on 320 seats.)
9/10 Martin Boon (ICM)
“I’m tempted to say: how should I know? I’m just a pollster. But I feel that Miliband may just have raised himself from the grave, so I’ll add a couple to where I had them before. Everyone else largely becalmed although, I see the Greens disappearing from view and Ukip sliding a touch. Tories 34 per cent, Lab 32 per cent, Lib Dems 14 per cent, Ukip 12 per cent. I don’t trust any academic model that translates vote shares into seats, so a seat projection from this is a pure and simple guess, which is Labour to be touching 300 seats with the Tories just behind.” (In January Boon predicted Labour on 290 seats.)
10/10 Lord Ashcroft (Lord Ashcroft Polls)
He refuses to make predictions. “My polls are snapshots, not predictions.”
The first was easy to fix, and the pollsters fixed it. The second they can do nothing about, although forecasters have tried to predict this election assuming that there will be the same swing back in the Government’s favour that there has been, on average, in the past. Most forecasts, though, suggest that the Conservatives will still fall just short of the number of seats that they need to keep David Cameron in Downing Street. (Oddly enough, the opinion pollsters, when asked to predict the outcome in our Poll of Pollsters, say exactly the same thing.)
There is the more general problem of opinion polls failing to pick up behavioural changes. In the Scottish referendum, for example, the final polls suggested that the No vote would win by a margin of five percentage points, but the result was that No won by 11 points (a case of the “shy Noes”?). If there were a six-point difference between the opinion polls and the result of the general election, that could mean either Cameron or Miliband winning a majority outright. I wouldn’t expect the opinion polls to be that far out: a referendum is a one-off event and strange things are going on in Scotland at the moment.
At the last general election, the final polls suggested that the Tories would win by a margin of eight points rather than the actual seven. In the only other big recent test of opinion polls, the London mayoral election of 2012, Boris Johnson’s margin of victory over Ken Livingstone was similarly a little over-estimated. That ought to worry the Conservatives.
But there is the third problem that the pollsters faced in 1992, an election in many ways like today’s, in which a Tory prime minister sought re-election against a Labour Party that he said was too left-wing. It is possible that the shy Tories are back. There may be a slice of British society that thinks it is not done to admit to voting Conservative. They feel under social pressure to conform with liberal attitudes towards public spending, welfare and immigration, but in the privacy of the polling booth may vote differently.
We have no way of knowing whether that is true this time until the exit poll is published at 10pm on 7 May. But the return of the shy Tories is now Cameron’s only hope. He thought that when the voters saw the whites of Miliband’s eyes they would recoil in horror, but it hasn’t happened. Instead, Miliband last week forced the Conservatives to defend non-dom status – a tax perk for the rich. As one senior civil servant commented to me, “even if God and the Queen came out for the Tories, it won’t stop this being Labour’s week”.
Who knows if we will be watching the dawn break on 8 May to discover that the shy Tories have, against all expectations, put David Cameron back in Downing Street?
The Independent has got together with May2015.com to produce a poll of polls that produces the most up-to-date data in as close to real time as is possible.
Click the buttons below to explore how the main parties' fortunes have changed:Reuse content