"Nothing about a trend ensures its continuation – but beware..."We’ve had a massive nine opinion polls, on the eve of the election. That’s almost double the number we had in 2005.
The Conservative lead varies from four points (TNS BMRB) to 12 (Angus Reid), although the shares of the votes being predicted for each of the three parties are more stable.
The Conservatives shares range from 33-37, although all but one are 36+/-1.
The Labour shares ranges from 24-29, although all but one are 28+/-1.
And the Lib Dem shares range from 26-29, although all but two are 28+/-1.
None of these polls – assuming a uniform national swing – would give the Conservatives a majority of seats, although they’d be pretty close on some. We suspect anyway, that there won’t be a uniform swing, and all the indications are that the Conservatives are doing better in the Con-Lab marginals they need to win. But maybe not by quite enough to win outright.
But here’s a trivia question. In the last four elections, there have been a total of 21 similar polls. How many have under-stated the Labour vote? Answer: just one.
At the last election, four of the polling companies over-stated Labour’s performance, whilst one, NOP, got it spot on. In 2001, all six over-stated Labour’s performance, as did five companies in 1992. In 1997, four companies over-stated Labour’s share of the vote. Just one, ICM in 1997, has a polling company under-stated Labour’s eventual performance. Nothing about a trend ensures its continuation – but beware...
The BBC/ITN exit poll will be out at 10pm.
Professor Philip Cowley
Posts
All the polls showing the highest forecast Labour share have a low share for "Others". That suggest that the relevant pollsters may have been less successful at winkling out intentions to vote for other parties among supposed Labour voters.
ReplyDelete