The opinion polls were wrong in 2015 – that’s to say, none of those conducted immediately before the poll found the seven point Conservative lead which it produced. The British Polling Council duly conducted an inquiry which, to cut a long story short, found that firms’ samples contained too few Tories. It said that “the 2015 poll were some of the most inaccurate since election polling first began in the UK in 1945”.
This background has made it easy for partisans of the parties in this election to mock poll findings that they don’t like – though it should be noted that we are still at the stage at which polls, in the phrase of ConservativeHome’s proprietor, are “snapshots and not predictions”. Polling day is still a week away and, although many postal votes will have been cast by now, there is no reason to expect next Thursday’s result to replicate findings obtained today.
So if the most recent polls are “wrong”, this can only mean that they are inaccurate snapshots of what’s happening now, not what will happen next week. Furthermore, such a claim would be complicated by the fact they are currently varying widely. On Sunday, ICM and ComRes found a 12 point Tory lead, Opinium a 10 point one, YouGov a seven point one and ORB a six point one.
It is therefore possible that when these recent polls were conducted the Conservative lead was bigger than 12 points; smaller than six points; that there wasn’t a Tory lead at all, or that it was a number between six and 12 points not picked up by the pollsters – all allowing for margins of error. So much for a statement of the obvious. But there is a futher complication.
Even if one of those findings was “right”, that isn’t the end of the matter. What will count most on election day is not how many votes are cast for each party, but how those votes are distributed across all constituencies. It will do the Conservatives and Labour no good if each pile up votes in seats that they already hold. Their aim must be to use their votes efficiently – in other words, to ensure that they perform well in their target seats.
This is exactly what Lynton Crosby et al managed to achieve in 2015. And it is what they will be seeking to repeat next Thursday. Which are these? We don’t know – because the Tory campaign, very sensibly, is not telling anyone. Mark Wallace’s best calculation for this site last month was that the Conservatives are targeting seats with Labour majorities of up to 8000, but CCHQ’s plan may well have changed since then.
All in all, then, neither presume when a good poll for your party comes out, nor despair if a bad one is issued instead – and not just because we’re still at the snapshot-not-prediction stage. Our best guess is that the polls have indeed picked up some movement to Labour and some away from the Conservatives since the manifesto launches, but that Theresa May is still on course for an increased majority come June 9.