The first thing some people say when I tell them I am in market research is “Oh, opinion polls?”. Well, no, the bulk of MR has nothing to do with them, but it doesn’t stop the wider industry getting the collective jitters when an election comes around. 1992 is etched on every pollster’s soul, even beyond any personal disappointment. The polls fucked up, comprehensively and enormously, with a strong Labour lead – stronger than they’re reporting now, actually – going in to the last day.

Since ’92 polls have been adjusted to take into account people’s apparent reticence to admit voting Tory. But are people lying to pollsters this time? And if so, what are they lying about? The reason people lied in 1992 was – according to the ‘experts’ – that voting for the devil they knew was something a little shameful, or selfish, or lazy. Does that reason still apply to the Tories? Maybe – but it might just as easily fit the Labour vote. My own hunch is that the “don’t know” element has been overstated this time out – a lot of public “don’t know” voters do in fact know where they’re likely to put their X, it’s just they don’t want to confirm it because their vote will be a reluctant one.

In 1997 I watched the election with a Labour-supporting researcher who was convinced that a ghastly polling error had taken place and the majority would be tiny, if it existed at all. As it is the polls still overrated the labour share of vote, as they seem to every time, but they underestimated tactical voting. I’m hoping that this time they’ve finally got the Labour share about right, but I’m still a little worried.