Could the polls be so wrong again?
Above is a picture of the famous speech by the then Labour leader, Neil Kinnock, at the Sheffield party rally a week before polling day on April 9th 1992. The polls were looking good for the party and it really did seem that the Tory government that had been in office since May 1979 was at last going to be beaten.
In the days leading up to the rally two polls from ICM had Labour with leads of 3% and 4%; Harris had a 6% lead while Mori in the Times was reporting a 7% margin.
On the Sunday before thee election NOP had Labour’s gap at 6%which by the eve of poll had been cut to 3%. ICM by then was down to level-pegging while Gallup found a 0.5% Tory lead – the only pollster to do so in the final week.
Even the exit polls taken on the day itself seemed to indicate that the Tory days were over. The TV results specials appeared to have been set up on the assumption that a Labour victory looked likely.
But by the time all the votes had been counted John Major was returned with an overall majority after the Tories secured a margin of nearly 8% of the overall national vote.
It’s in this context that the latest polls and the prospect of an October general election should be judged. Brown was reported to have been totally shattered by the outcome and this must still weigh on his mind. Of course, as we’ve discussed many times here polling appears to have got better since the 1992 disaster with new techniques designed to deal with what appears to be a systemic bias to Labour.
I think that the latest polls are in the right ball-park but I have one slight niggling doubt. The process of past vote weighting, whereby samples are partially adjusted in line with what respondents said they did last time, has never been tested in a situation like we have at the moment where the lead has changed hands three times in less than two years. Could that impact how people reply to the past vote question?
Sky News. I’m down to appear on Sky News this evening to discuss today’s developments. I’ll let you know when I have more details.