Boris Johnson would lose an election held today. Despite holding a steady four-point lead in recent polls, the Conservatives would lose around 70 seats. They would remain the largest party in parliament, but with too few MPs to stay in office. And just as their huge gains in Labour’s traditional “Red Wall” seats delivered Boris Johnson’s big majority last December, it is a seven per cent swing to Labour in these seats – almost double the national average – that would force the Tories into opposition. Keir Starmer would probably become prime minister, albeit at the head of a minority or coalition government. The main figures are shown in the chart below.
No, I have not commissioned a special poll, peered into a crystal ball or consulted Mystic Meg; and yes, you are right, none of the recent poll stories in the media point to that result. However, the data are in the public domain. They can be crunched by anyone with a little time and an appetite for playing with Excel.
The data come from Opinium’s polls for the Observer. Each week they question 2,000 people throughout Britain. As well as providing figures for the main subgroups, such as age, gender, region and past vote, they give results for different kinds of constituency. The key group comprises the 54 seats in England and Wales that the Tories gained from Labour. Anyone can look at these any week.
However, only 200 or so respondents each week live in one of these seats. The margin of error is enormous. Nobody should pay much attention to a single week’s figures from such a small sub-group – and, as far as I know, nobody does.
How, then, do I conclude that Johnson is in so much trouble? Here’s how.
First, party loyalties have been unusually stable for the past month. The last four Opinium surveys have reported overall Conservative leads of five, four, four and four points. This makes it possible to combine the data from all four polls to produce an overall sample of 8,000. This reduces the margin of error in subgroups of individual surveys.
Second, Opinium asks people not only how they would vote today but how they voted last December. In the seats gained by the Tories, respondents across Opinium’s four latest surveys recall voting: Conservative 48%, Labour 38%. This is close to the actual total vote from the 54 seats: Conservative 47%, Labour 39%.
Third, by comparing how people voted last December with how the same people would vote today, we can measure the swing in each group of seats. This shows that over the four weeks, respondents in the Red Wall seats are substantially more likely to say they have returned to Labour than respondents elsewhere.
Fourth, although the current voting figures in the Red Wall seats vary a fair amount from poll to poll, as you would expect, what matters more is the way support has changed; and here the figures are remarkably constant. In all four polls the swing to Labour is 7%, plus or minus one. Consistency is not proof of accuracy, but it is reassuring. Had the swings in the four weeks been, say, four, five, six and 13%, the average would still have been seven, but I would have been more nervous about that average.
If we apply a seven-point swing to the 54 seats that the Tories gained from Labour, then 45 of them would now revert to Labour. That is not all. Opinium also reports an above-average swing in other seats that the Tories gained from Labour in the previous decade. Twenty five of them would fall to Labour, bringing the total to 70.
This would reduce the number of Tory MPs to 295, while Labour would climb to 272. Changes in the fortunes of the Liberal Democrats (11 seats last December) and SNP (48) might affects these figures a little, but probably not much. In reality, even if Opinium’s data is absolutely right, swings will vary from seat to seat. But the chances are that these would broadly cancel out. The Tories would save some seats on a below-average red-wall swing – but lose others on an above average swing.
To retain power, the Conservatives need around 315 seats. If they fall short of that, then the combined total of Labour, SNP, Lib Dem and the smaller anti-Conservative parties would be enough to outvote them, even if they do a deal with the Democratic Unionists. On Opinium’s figures, the Tories would fall well short of this. They would be out.
Now to the qualifications. There will not be an election any time soon; and past governments have recovered from far worse poll figures than these. A snapshot of the Grand National as the horses reach Becher’s Brook for the first time is not much use for predicting the final winner.
Second, Labour still has real problems. Even though Keir Starmer outscores Boris Johnson handsomely on competence, honesty and “having the nation’s best interests at heart”, more people trust the Conservatives more on the economy, and by a margin of 47-36%, voters do not think “Labour is ready to form the next government”. Its current reputation can be summed up as ‘nice leader, shame about the party’.
Oddly, that news is not as bad for Labour as it might seem. It is barely three months since Jeremy Corbyn stood down as leader. It will take time for Starmer to convince millions of voters that his party really has changed. If and when he does, Labour’s support could rise significantly.
Meanwhile, the big story is not just that the Tories have seen a 26 point lead crumble in three months – the most precipitate collapse in government support since the winter of discontent 41 years ago – but that voters in the Red Wall seats are leading the way. Last December, Johnson acknowledged that many voters in Labour’s former heartlands had only “lent” the Tories their support. Seven months later, given the chance, many of them would take it back.
This analysis first appeared in The New European https://www.theneweuropean.co.uk/top-stories/peter-kellner-on-boris-johnson-and-keir-starmer-polling-1-6746756