Conservative Diary

« Cameron is right to focus on Day One in Downing Street | Main | Ipsos-Mori poll: "Late surge" in marginals now points to a narrow Conservative majority »

Do bank holiday polls underestimate Conservative support?

As Nick Wood pointed out yesterday, the momentum has very much been with the David Cameron and the Conservative campaign since the final Leaders' Debate last Thursday.

So many have been perplexed by the polls in this morning's papers showing a small reduction in the Tory voteshare.

A discussion has ensued in the comments section of Stephan Shakespeare's post this morning as to the reliability of polls taken over bank holiday weekends: are they less reliable because it is harder to make contact with some people, thereby underestimating Conservative support?

Ben Page from Ipsos Mori sums up the issue thus:

"The theory is that they are slightly less reliable as lots of people are away. But this is hard to prove one way or the other and there is no definitive evidence at all - but there were some rogues in the otherwise accurate poll series in the last election campaign."

Stephan Shakespeare of YouGov gives this take on this morning's poll movement:

"There's been a clear trend of strengthening of support for the Conservatives in the past week and the 1% move down last night [in the YouGov poll] should be taken as nothing in the slightest bit significant, statistically it's meaningless. There could indeed be a minor effect from the bank holiday - certain demographic types may be marginally less likely to be at home when the telephone rings, and this might introduce a sliver of bias against the Tories - but the bigger point is all the movements are too small to suggest any kind of trend reversal."

Meanwhile, Andrew Hawkins of ComRes says that bank holiday weather is more likely to be a crucial factor rather than it being a bank holiday weekend per se:

"A sunny bank holiday is more likely to result in a dodgy sample, as you end up with more Labour voters in your sample because it is genuinely more dificult to get hold of Conservatives. That's been the experience in the past. So the samples sometimes need a health warning - but if the poll is weighted properly it shouldn't really make too much difference. There is a difference between a sunny bank holiday weekend and a foul freezing one, so there shouldn't be a huge difference to the outcome of this weekend's polls. In any case, most of the changes are within the margin of error."

In the meantime, poll-watchers can soak up the latest prediction from PoliticsHome's Poll Centre - that the Conservatives are 28 seats short of a majority.

We now await a new Ipsos Mori poll conducted solely in the all-important marginal constituencies, which will be published later this afternoon.

Jonathan Isaby


You must be logged in using Intense Debate, Wordpress, Twitter or Facebook to comment.