Conservative Home

« Howard Flight: There are now less than two weeks to get out there and help save our country | Main | Fiona Bulmer: The left wing critics are wrong - Free schools will drive up standards in all schools »

Quentin Langley: How to read the polls

Picture 1 Quentin Langley read Politics under Colin Rallings and Michael Thrasher at the University of Plymouth and now teaches Public Relations and Political Communications at Cardiff University.

At first glance the polls seem extremely volatile and difficult to read.  At second glance, this is only partly so.  Nearly all the recent polls have shown the Conservatives at 35% +/-3 and Labour at 27%+/-3.  The Lib Dems, however, seem to be at 27% +/-7.  The consensus has them above Labour, and occasionally matching the Conservatives, but several polls, including the latest from MORI, have them in third place in the low 20s.

This raises two obvious questions.  Why the big variation in the Lib Dem vote?  And which pollster is likely to have it right?  This is not a new phenomenon.  In the run up to last year’s local and European elections, Populus was reporting Lib Dem numbers ten points, or 40%, below other pollsters.

Populus operates a slightly different methodology, though it is not intuitively obvious why this should produce a different result.  Indeed, unlike last year, it currently doesn’t.  Generally pollsters ask a fairly standard question: “if there were a general election tomorrow, which party would you vote for?  When an election has actually been called the question becomes “which party are you likely to support in the general election on the sixth of May?”  Populus shows voters a list of parties to choose from.  But why should this methodology produce a different result, and different only as to the support for the Lib Dems? 

My theory is this: about 10% of the electorate are irreconcileables.  They hate politicians, especially anyone they see as part of the establishment.  Given a choice of three candidates they will choose the Lib Dems.  Given a choice of four or more parties those same people will choose a fourth or fifth party.

Not only does this explanation, especially following the expenses scandal, smell right, it is also borne out by actual votes.  The Lib Dems consistently underperform in Scotland and Wales, where there is always a fourth party.  (In Scotland, they don’t do badly in seats, and would probably lose seats on a proportional electoral system, partly because you don’t need as much support to win a four way contest).  The real vindication of the Populus methodology and the Langley hypothesis explaining it lies in last year’s elections.  In the local elections the Lib Dems came second with support in the mid 20s.  In the Euro elections on the same day they came fourth trailing, their local support by more than 10 points.  In most local elections, of course, voters faced a choice of three parties, and typically five or more in the Euros.

It has to be noted, that Populus is not one of the pollsters reporting lower figures for the Lib Dems currently, so there may now be other factors at work.

What, then, are the polls telling us?

Conservative support seems to be steady compared with 2005.  If, as in the last four elections, actual support is in line with the highest figures recorded by opinion polls, then Conservative support is up by three points, or 10%, compared with 2005.  This would mean in a seat where Conservative support was 20% it would typically rise to 22% and where it was 40% it would typically rise to 44%.  If the brand has been detoxified, however, it may be that instances of ‘shy’ Conservatives are less than in the past, and polls might not be understating support as they have in the past.

Labour support has fallen by a third since 2005.  However, there could be ‘shy’ Labour supporters on this occasion.  If Labour support really has fallen by 10 or 12 points since 2005, this will be the biggest fall by any major party since 1983.  It is a much bigger decline than the Conservatives suffered in 1997, and comes after two successive elections of declining Labour support.  By contrast, the smaller Conservative decline in 1997 followed an increase in support in 1992, despite a decrease in the lead over Labour.

If that 10% are still saying Lib Dem to pollsters, but voting for a fourth or fifth party, Lib Dem support is steady compared with 2005.  Traditionally, Lib Dem support is slightly higher in the actual vote than in polls beforehand, but this has usually been explained by the increased exposure the party receives at election time.  I would suggest that, in the light of the debates, this is already factored into current polling.  If this is right, the main beneficiaries of Labour’s decline – in England at any rate – may be minor parties such as the Greens, UKIP and the BNP.  They are seriously competing in only handful of seats and may not win any at all.  They will benefit only in the sense of getting a slightly improved third or fourth place.

Polling is a snapshot.  Things could still change.

Projections from votes to seats are unreliable.   I have written on this before.  They usually assume a uniform swing (whereas I assume a proportional swing).  They take no account of tactical voting, which has counted against the Conservatives in recent elections, and may well count against Labour this time.  If the polling consensus is right, then the Lib Dems are in second place, and all the predictions beforehand that the Conservatives need 40% support and a margin of ten points for an overall majority are garbage.  Every one of those projections was predicated on the notion that Labour was in second place.  Labour benefits disproportionately from first past the post because of very concentrated support.  However, outside a few areas with strong personal votes, the Lib Dem vote is more evenly spread than the Conservative Party’s.  Even a small lead over the Lib Dems – comparable to Labour’s three point lead in 2005 – could easily deliver a Conservative majority.


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