Time to separate the noise from the signal. Polling headlines often pick out the noise—dramatic changes that might be short-term, or reflect sampling fluctuations, or both—rather than the signal: the underlying state of public opinion, which conveys more significant messages.
Two polls in recent days point in different directions. YouGov’s poll in Saturday’s Times reported a “collapse” in the Tory lead in less than a week, from 13 to 8 per cent. On Sunday, Opinium’s poll for the Observer said the Conservative lead was holding steady at 8 per cent. Cue the old cliché, “polls apart.”
Sampling theory tells us not to be surprised by this. There is a margin of error of two or three points in the support for a party. The margin of error for the gap between two parties is almost double that. And the margin of error for a change in the gap between two parties is roughly half as much again. My guess is that the Tory lead has fallen a bit in the past fortnight. This is confirmed by three more surveys published since the weekend, but they disagree by how much.
Meanwhile, let’s tip-toe away from the noise and look for the signal. Here, separate YouGov data, available on its website but seldom reported in the media, tell an important story. (Memo to my excellent former YouGov colleagues: your tracking data are wonderful, but hard to navigate. Maybe that’s why they receive far less attention than they deserve.)
The following table lays out the way voters view the two main parties. It shows how they rate on six different aspects of their overall reputation. These are what political scientists call the “valence” qualities that generally determine whether a party is electable or not.
Those results, from shortly before the dramas of “Freedom Day,” show something we have not seen since the general election. Both parties now have terrible reputations. As the last line of the table shows, the average “valence” scores of the two parties are dreadful: minus 31 for the Conservatives and minus 25 for Labour. Arithmetically, Labour had a “valence” lead of six points in early July; but all this means is that the Tories were fractionally deeper in the mire than Labour. Most dramatically, just one voter in six believed either party was “in touch.”
We last saw such plague-on-both-your-houses figures in the autumn of 2019, when Jeremy Corbyn led the Labour Party and Boris Johnson was being battered by parliament and the Supreme Court. But since the election, and until this month, only one of the two parties has had dreadful ratings at any given time: not both.
In April last year, just after Johnson left hospital and before Keir Starmer had made an impression on the public as Labour’s new leader, the overall valence scores were: Conservative plus 8, Labour minus 20. Two months later, as the deaths from Covid mounted, and Starmer was seen to hold Johnson to account at Prime Minister’s Questions (as well as lance the boil of antisemitism in Labour’s ranks), the Tory valence score dropped to minus 20, while Labour recovered to net zero.
Labour retained its valence lead over the Tories into last autumn, when the opposition also overtook in voting intention. This spring, the Tories returned to the lead, as the vaccine rollout gathered pace. It looked as if the government was regaining its popularity.
The valence figures, however, show that things are not that simple. In May—when the Conservative voting intention lead had stretched to double-digits and the party won the Hartlepool by-election—the Tories’ valence score was minus 14. To be sure, that was an improvement on minus 20 the previous June; but not a big one. The larger change was Labour’s decline over the same period, from zero to minus 29. Since May, the Tory valence deficit has more than doubled, to minus 31, while Labour’s has barely changed. Neither party’s brand image is remotely near where it wishes to be.
When we delve into the figures, it appears that the vaccine roll-out made the government seem stronger and more competent than it did last winter. However, over the same period, there were sharp falls in the numbers of voters crediting the party with being trustworthy, or wanting to do the right thing rather than “serving its own interests.” Labour, meanwhile, has a truly calamitous reputation. By huge margins voters regard it as weak, out of touch and incompetent. The latest figures, shown in the table, are not quite as bad as they were in May, but they are still awful.
What all this suggests is that, even when the Conservative Party enjoyed big leads, its support was more fragile than it looked, and owed more to Labour’s weakness than its own performance. Perhaps the events of the past week have exposed that fragility. Polls over the next few days should clarify the matter.
However, as long as Labour’s valence score remains so low, Johnson can feel confident about the next election. A poor government is always likely to win when most voters think the alternative is rubbish. This is especially true when a nation has experienced such turbulence. Floating voters may well decide to keep a hold of nurse, however fumbling, for fear of choosing someone worse. (This, incidentally, might explain Opinium’s latest finding that the Tories continue to hold a comfortable lead over Labour, even as the government’s rating for handling the pandemic plunges sharply.)
But what if Labour and Starmer restore their reputation? The latest monthly Ipsos-MORI political monitor charts the fortunes of new opposition leaders since 1980. It shows that most of them enjoyed an early honeymoon which faded over the following 12-18 months. Starmer is in the middle of the pack. He is doing worse than Tony Blair, better than Michael Foot. What matters is what happens next. David Cameron’s worst rating in 2007 was much the same as Starmer’s is today—yet he recovered to become prime minister three years later.
In short, the next election is wide open. It will be won by the party that can climb most effectively out of the hole it’s in. My advice to those thinking of betting on the result? Don’t.
This blog was first published by Prospect