“Labour surges to 6% lead”. Should we believe it?

Suddenly everyone – everyone, that is, who is obsessed with the ups and downs at Westminster – has got excited. Before today, eight national polls conducted since the Owen Paterson saga hit the headlines had been published. They all told the same story and none had received more than modest attention. They ranged from a Conservative lead of 3% to a Labour lead of 2%. Given sampling fluctuations, they agreed that the two main parties are level-pegging.

Then, wow, today, we have a ninth poll, this one, from ComRes for the Daily Mail,  reporting a six point Labour lead …. and the world of political soothsayers goes berserk.

One of the eternal verities of poll stories is that if one survey is out of line, then a) it receives more attention; and b) it may well be wrong. On this occasion, a) is certainly true; what about b)?

To make the point more formally, when a poll produces results that differ from other recent figures, there are three possible explanations. They are not mutually exclusive: two or all three may be true.

  • Is it a rogue? Sampling theory tells us that nineteen times out of twenty, a perfectly-conducted poll should be within 2-3 percentage points of the right figure for each party. One time in 20, the laws of chance mean that the errors could be greater than that. The ComRes poll shows Labour on 40%, the Tories on 34%. If it’s three points too high on Labour, and three points too low on the Conservatives, the true figure would be a tie: 37% each. If ComRes is really unlucky, the truth might be a narrow Tory lead.
  • Is it a blip? A poll might accurately report a brief, passing phase in public attitudes, in which the underlying state of public opinion is interrupted by a particular event. This causes voters without strong political loyalties – that’s a very high proportion of us – to give pollsters a different answer from that they would have given a week ago, and might give them next week. The various sleaze stories of recent days might be an occasion for such a blip.
  • Is it a trend? Sometimes the mood of the electorate changes, and stays changed. This happened after Black Wednesday in 1992 and after the financial crisis in 2008. In Scotland, the SNP leaped into a big lead after the 2014 independence referendum, and the party has stayed well ahead ever since. Maybe the sleaze stories have damaged the reputation of Boris Johnson and his party beyond repair.

Whether today’s poll reflects a rogue, a blip or a trend, or some mixture of them, will take time to become clear. My guess, and at this stage it can be no more than a guess, is that the first two factors are probably involved; I am not sure about the third. ComRes’s fieldwork (November 11-12) is close to that of the latest YouGov/Times poll (November 10-11; tie) and Redfield & Wilton (November 10, 2% Labour lead). It is possible that there has been a sharp jump in Labour’s lead in 24 hours; more likely that ComRes’s figures are slightly too generous to Labour and slightly too harsh on the Tories.

However, all recent polls show the race tightening since late October. It’s reasonable to suppose that this is a response to the sleaze stories. By definition this is so far a short-lived move – a blip. But will it solidify into a lasting trend? I don’t know. Ask me again in a month or two.

Meanwhile, Labour should not be too cock-a-hoop. The ComRes figures might not be quite right. But even if they are, a six-point lead for the opposition in mid-term when the Prime Minister has plainly had a bad ten days is nothing to crow about. Governments are in real mid-term trouble when they fall 15-20 points behind their opponents and stay there. Whatever the precise state of public opinion this weekend, The Conservatives are nowhere near that parlous state.