Making sense of the disparity in the polls

By Jonathan Bartley
April 6, 2010

The final pre-campaign opinion polls seem all over the place this morning with ICM and YouGov polls showing a six point difference in the Tory Lead over Labour. This indicates that this may not be an election easily called on the usual basis of a swing from Labour to Tories.

Bob Worcester from MORI always tells the media to look at the share of the vote rather than the lead (ie that there is a 2% margin of error around everyone's share of the vote and that is the crucial thing to look at). The disparity may also be explained by some as one or other poll being a 'rogue'. But we may well see more 'rogues' than usual this election. And even this doesn't perhaps tell the whole story. Lib Dems often cloud the picture significantly - something that polling organisations themselves understand. This factor could be far more significant this election, with TV debates presenting the Lib Dems on a more equal footing.

And there's more. It is true that an opinion poll reflects the balance of male and female, age distribution, income distribution, different regions etc... but crucially, most do not account for what is going on at constituency level. Most are also commissioned within a narrative of politics dominated by the big three parties, and so tend to ignore what else may be taking place.

Many of course take account of the situation in the 150 or so marginals which are key to determing who gets into Government, but even these do not (can not in a sample size of 1,000 or 2,000) drill down to local issues and other factors in any given constituency, when they are trying to get a national flavour.

Approximate forecasting of results is usually achieved by assuming that the swing in each individual constituency will be Normally-distributed around the mean. But here is the rub, because perhaps more than ever before, there will be significant local variation between constituencies.

A crucial element will be the 'other' category. Different polling organisations show different percentages for the category but what is interesting, is that it has tended to hold pretty steady. YouGov for example, have held it around 11-12 per cent for months now.

There are a small handful of obvious examples where things are not likely to relate closely to the national swing:

- The Greens may pick up Brighton Pavilion as their first Westminster seat and may come close in two others: Lewisham and Norwich South

- George Galloway’s Respect has a chance with two seats they are contesting in East London

- Respect could make a gain in Birmingham Hall Green

- Nick Griffin is contesting Barking and Dagenham

- John Bercow may lose his seat to Nigel Farage

- Richard Taylor may hold onto his seat in Wyre Forest

- Esther Rantzen may have an impact in Luton

But right around the country, independents and smaller parties are standing. Most have much less of a chance under first-past-the-post, but there will undoubtedly be ‘protest’ votes. Such votes are perhaps better seen, in theological terms, as votes of ‘witness’. Even where they don't have a chance, they may gain significant support and take support away from other candidates, skewing the results which one might otherwise try to predict on the basis of a national swing.

Our own poll, that we commissioned through ComRes last year suggested 78% believed independents should stand where MPs had behaved ‘unethically’, 63% believed British democracy would be strengthened if there were more independent MPs and 53% believed they would ‘seriously consider’ voting for an independent candidate at the next general election.

Then of course, there is the SNP and Plaid factor to throw into the mix in Scotland and Wales.

We shouldn't forget to add the European Election experience to the mix, either. The three big parties polled about 57 per cent of the vote, down from 64 per cent in 2005. And if Plaid and SNP and UKIP are added in, it still only comes to 76 per cent. That left 24 per cent for Green, BNP and assorted others, including independents. That is a significant seven per cent increase on 2005 which saw these sharing just 17 per cent of the vote.

The Euros were, of course, right off the back of the MP’s expenses scandal, it was a proportional list system, and on ‘European’ issues. But this doesn’t mean it won’t be some kind of indicator for what will happen this time. This is the first test for Westminster following MP’s expenses after all, and the anger hasn’t all subsided. Independents were also effectively excluded under the party list system in the Euros.

It’s likely that a record number of British voters will vote for the ‘other’ parties. Losing support to minor parties in marginals could be costly for the big three. Add to this the ongoing prospect of a hung Parliament, and the incentive to get behind 'minority' candidates who might be seen to 'make a difference' increases.

It is true that the anger over expenses has lessened and may not have such a significant effect that some initially predicted . It is there nevertheless, and there are still 148 MPs seeking re-election who had to repay more than £1000 pounds of misclaimed expenses. No one can be quite sure how that will play out.

The record number of retiring MPs may also skew the results. In most seats a sitting MP could be expected to command a degree of personal vote built up over the years.

The moral of the story? Prepare to see a few rogue polls and some significant variation in the numbers. But also be prepared for an election result which doesn't closely follow the polling predictions on the basis of a national 'swing'.

Although the views expressed in this article do not necessarily represent the views of Ekklesia, the article may reflect Ekklesia's values. If you use Ekklesia's news briefings please consider making a donation to sponsor Ekklesia's work here.