UK Elections 8: Opinion Polls

The Conservatives on 39%, 7 points ahead of Labour, or, on the same day, 37-35. What are we to make of this? Are the opinion polls so inaccurate that it is a complete waste of time reading them (and of money by the newspapers publishing them)?

The story is not one of complete stupidity. Pollsters first knew there was something wrong in the 1992 election. I remember driving back from the count (I was standing on behalf of The Anti-Federalist League - now UKIP - in Bristol). It was 9th April, exactly eìghteen years ago.

Everyone thought Neil Kinnock, the Labour leader, was going to win. But the story unfolded as we drove through the night and it was a clear, if not completely comfortable, win for John Major and the Conservatives. The pollsters put their heads together and began to worry.

In 1997 and 2001 there wasn’t much need to worry as the result was deemed obvious, and a landslide delivers inaccuracies of itself, although they noticed in 2001 a slight but definite overestimation of the Labour vote. In 2005 the trend continued. They identified the following problems:

1. Demographics. The boundaries for the constituencies are usually at least five years out of date. People have been moving out of Labour city strongholds to live in Tory, LibDem or marginal seats. In Labour marginals the Labour vote may not hold up but if they move to Tory marginals they might turn red.
2. As a result it needs fewer votes to hold a Labour seat than a Tory one.
3. Voting intentions. Interviewees are asked how likely they are to vote. It would seem that people who vote Labour are more likely to say they will vote and less likely actually to do it.
4. Confidence. If they feel certain their party is going to win they often don’t bother to vote.
5. Strength in the marginals. The Tories are widely thought to have pursued a progressive campaign, pushing themselves heavily in the seats they hope to gain. No one can be suire how well this is working.

The pollsters weight the vote they have taken from their sample of about 1,000 to cover these variables.

How influential these variables are likely to be in this election they are still pondering. How likely they will be in a future election, with for example a Conservative administration defending a weak or a strong lead, is even less certain.

So spare a thought for the pollsters. It is a job more difficult than it looks. I was going to add ‘but someone has to do it’ but I’m not even sure of that.

The Tim Hedges guide: If the Tories are on 39 or 40 and at least six points ahead of Labour, they will win a clear majority. Less than 6 points difference and less than 39% of the vote is likely to mean a hung parliament. If the two parties are level pegging Labour will gain a majority.