Wednesday, 5 February 2014
Don't panic when you read the polls
The recent news of shock UK opinion polls showing the Labour Party only one or two points ahead of the govenring Conservative Party, and a sharp swing towards a 'yes' vote on independence in Scotland, sent commentators scurrying for the superlatives. 'Gamechanging'; 'massive'; 'huge': these were the sort of words bandied about.
Except that they weren't so huge. Labour soon posted some nine or ten point opinion poll leads less than a week after their polling Black Monday, and the most recent Scottish independence polls show nothing like the rapid change that showed up in the last ICM survey.
That's not to say things aren't changing, of course. Labour's UK lead is very, very, very slowly deflating - from double digits a year ago to medium single digits today (above, you can see UK trends since 2010). At this rate the next General Election will be close in terms of seats: something that readers of 'Public Policy and the Past' always knew anyway. And in Scotland? Well, there's a gradual change there, too, with 'yes' climbing and 'no' holding relatively steady. But 'no' was so far ahead - by over thirty percentage points in a few polls during last summer - that some change just had to take place. There was no way that such a landslide was ever deliverable, especially as Scots are haunted by the idea of another five years of UK Conservative government they never voted for in the first place.
But the numbers are changing gradually. It's inherently unlikely that in a week or a month or two, five to ten per cent of voters are going to say 'that nice Mr Cameron, what I really want is five more years of him', or 'you know what, I always thought that Scotland should be an independent country'. Things don't work like that. Parties (and referenda campaigns) draw their numbers from the don't vote, won't vote, might vote either way camps - small, wary and sometimes all over the place, as is their very nature. Throw in the way pollsters change their raw numbers all the time these days via various weightings, and you're looking at a complex picture in which you can't put all that much store by one set of numbers.
So here's a tip: don't get too excited when you read that you very much like (or dislike). Look at lots of polls. Look at the trendlines. And take the long view.
Things are going to get very hot over the next fifteen months: don't get your own self too hot under the collar, will you?