Inaccurate opinion polls are what got us into this mess in the first place | Media


There’s one dread descant to this shambles. The exit poll was good enough, but what about the opinion polls that preceded it? Another general disaster. And a disaster of particular resonance because these polls shape coverage of election campaigns. They seem to tell journalists and sundry experts what’s happening. But what if it isn’t, just as in 2010, 2015, 2016? Then the whole story is wrong.

It was wrong seven years ago when poll after poll overestimated a Lib Dem surge – 33% of the total vote at one moment – and saw “Cleggmania” slide to fewer seats and a smaller share than 2005 on the night.

It was wrong two years ago when the BBC’s David Cowling monitored 92 polls over the six weeks of the campaign. Some 46 of them suggested Labour leads. “None of the 92 accurately predicted the 7% lead the Conservatives would actually achieve.” The British Polling Council launched an investigation-cum-inquest.

It was wrong last year at referendum crunch time. The last two polls before the count showed Britain remaining by 55% to 45% and 48% to 42%. No wonder Nigel Farage went to bed glum and woke up happy. It was wrong again on Thursday. Some of the polls on 7 June delivered Tory leads of seven points, 10 points, 12 points and 13 points. No wonder David Dimbleby’s jaw dropped when he opened the exit report.

And see how the polls seem to underpin decisions as well as headlines. Would David Cameron have promised a European referendum if he hadn’t thought another Lib Dem coalition, ruling it out, was on the cards? Would Theresa May – wallowing in 23% polling leads in April – have walked the 8 June plank? Would her manifesto have included threats to pension increases, winter fuel allowances and the rest if she hadn’t thought a few lost pensioners were dispensable?

Of course, one or two pollsters did better this time round. Survation and new-method YouGov may take a bow. But foraging for truth in a world where only an overall average of all the polls seems safe is a mug’s game.

There are explanations on offer. There always are. We’ve had “shy Tories”, very late swings, computer voting versus the telephone call. Now we’ve got likelihood of young people turning up at the polling booth. But the underlying problem is that something is always wrong – or wrong enough to affect perceptions and decisions.

What if those two polls with 12- and 13-point leads had turned out to be right? Then that would reflect their own properly scientific samples, duly massaged, yet the story itself would have been completely different. And thus the same question may be asked about poll results weeks or months before an election, the findings that influence policy packages. Why should they be any more solid?

There are arguments about snapshots not being predictions, to be sure. Duly registered. But countless snapshots make the whole picture, and that picture keeps falling off the wall.

There must be a better media way. No more Dimbleby jaws dropping on the night. It won’t come from canvassers on the doorsteps: they were as wrong as the pollsters. It has to come by stepping back, clearing minds, realising regional differences and finding the story for ourselves: with any phone to a pollster firmly switched off.

Source