It is worth your time following election opinion polls – here’s why

Every election opinion pollsters hear the same complaints about their work. “You can’t trust the polls, they are all over the place, and they got it wrong last time.” “I’ve never been polled.” “What can you tell from 2,000 people?” “They’re a plot to manipulate opinion: they are all owned by Tories.”

Every time, they try patiently to explain what they do and why they don’t always get it right. Thank goodness they don’t, by the way, because imagine how dull (or corrupt) politics would be if we knew the result of an election before anyone had voted.

Opinion polls are, of course, the worst way of finding out what people think, apart from any others. A well designed sample of 1,000 or 2,000 people is enough to construct a reasonably accurate picture of what all voters think. It needs to be adjusted so that it is representative – that is, more weight has to be given to answers from young women if there are too few of them in the sample.

And no, the polls are not “all over the place”. Different companies produce different estimates. Last time, two of them – Survation and YouGov MRP – were close to the result using very different methods, but we don’t know if they will repeat their success on Thursday.

Most of the polls this time do show a similar trend: the gap between Labour and Tories narrowing to start with, but now stable.

And no, the polls are not a “Tory plot”. All polling companies want a reputation for accuracy; if they tried to twist their findings for political reasons, they would soon be put out of business by a non-political competitor.

Yours,

John Rentoul

Chief political commentator