Tuesday, 7 October 2014

Labour's mistake: thinking of the voters like lego

We left you last week with a question: could the Conservatives get away with a totally uncosted load of tax cutting bribes? And the answer, now as it has ever been, is this: yes, they probably can.

Regular readers will know that Public Policy and the Past has always thought Labour's poll numbers were extremely underwhelming. Even when they've led in the polls, they've never been able to pull away. Now they're actually behind the Conservatives with some pollsters - though, we should note, not with others. The mix of a recovering economy and their lead over Labour on questions of Prime Ministerial leadership seems decisive - at least insofar as being returned as the largest party after the May General Election. Could the Conservatives even gain an overall majority? Well, that's probably beyond them, though they have an outside shot at making it to the 323 or so seats they need for a working majority.

To understand what has happened, and why things have changed so quickly after Labour seemed to be crawling ahead just a few short weeks ago, we have to look deeply into the way the present Labour leadership sees the world.

The hope of Labour's leader, Ed Miliband (above), as well as the analysis of his advisers, has always been that left-leaning Liberal Democrat defectors who had voted for Nick Clegg's party in 2010 would come over to them en masse. Disgusted at coalition cuts and austerity, as well as at Liberal Democrat pledge-shredding on issues such as university tuition fees, they'd vote Labour whatever happened. Large numbers of Liberal Democrat switchers to Labour, of around a third of the smaller party's 2010 voter base, have always seemed to bear that out. Close observers of opinion polls have termed this Labour's 'firewall', boosting Labour's vote well into the mid- or even high-30s and ensuring that the Conservatives could never pull away enough to wrest decisive control of any contest. According to this way of counting the world, seats such as Lancaster, Broxtowe and Warwickshire North would all fall to Labour as Liberal Democrat deserters flocked to their banner. They didn't need many Conservative switchers. They had the soft left that the Liberal Democrats had captured over Iraq and the economics of 'New Labour'. They could move slightly to the left.

Now this theory was simple. It was elegant. It was backed up by hard, cold numbers. It seemed to make sense. It stuck.

Until now, that is.

Now those numbers of Liberal Democrat to Labour 'switchers' are on some reckonings only in their teens, and they're falling towards the number of Liberal Democrats who are actually thinking of voting Conservative, partly because they'd like a few thousand quid extra in their back pockets, thank you very much.

But what is the deeper reason behind this mismatch between perception and reality? It's down to the fact that the electorate is made up of solid blocs that one can define ideologically is wrong. And the whole thing's a bit patronising, to be honest. Voters don't see themselves spread out along a left-right spectrum, and they certainly don't see themselves 'moving' their allegiance from one party to another, like platoons of foot soldiers. There is no such thing as settled lego-style bricks of voters who are loyal to their 'home', whether new or or long-standing. Citizens are 'churning' around all the time - from Liberal Demoract to Labour, sure, but also from Liberal Democrat to Conservative, from non-voting to Conservative, and back and forth between the Conservatives and the United Kingdom Independence Party, as well as from Labour to Ukip. Failing to understand that 'churn', and what it tells us about a fundamentally unstable and volatile electorate that might change its mind (and back again) at any time, is one of Labour's biggest conceptual errors of this Parliament.

Labour's Liberal Democrat 'firewall' might still hold - it looks likely to do so in the North of England, for instance, where the toxicity of Clegg and Co. seems likely (for instance) to hand Labour Thursday's Heywood and Middleton by-election on a plate. But elsewhere? It seems much more questionable. And there will go the whole election - for Labour, the victim of structural and schematic thinking in an age of the individual.


  1. Polling is not a neutral profession, a great deal of assumptions have to be made about how to interpret the raw data due to the small sample sizes. Just as the You Gov poll which showed YES ahead with two weeks to go was ultimately revealed to be an outlier, the three polls conducted by You Gov immediately after the Conservative conference show similar otlier properties.

    Almost all raw data from polling by all companies shows the Conservative vote is down by 10 - 20%, Liberal Democrat down 65% and Labour up 25 to 35% from 2010.

    The conservatives have done a great deal in this parliament to alienate large sections of society. Polling amongst the 18-24 group shows almost overwhelming opposition to the Coalition Parties in that age group. The only age group positively supporting the Conservative Party on a consistent basis is the over 65 age group where there is a 10% lead in support over Labour.

    It appears that polls consistently weight against Labour support and use models to reduce the current apparent support.

    I think that at the next election Labour will win a larger majority than expected and by a larger margin of the popular vote than expected. The media narrative is of failure and despair for Labour, but that is all it is. The poor record of the current administration cannot be obscured from the people whom it has affected. They are sufficient to turn the tide.

  2. I forgot to mention that the latest You Gov already has Labour ahead again by 2 points and the raw data looks more in line with their usual results. Unfortunately polling is very prone to sample error.

    You gov has a large sample pool of people registered with them, but people to some extent self-select ( you may be invited to complete a poll but not do so within the time allowed), In order to poll 2130 people, 3000 may be invited to complete the poll, allowing for non-completion of the survey.

    Whilst YouGov may have invited the correct proportion of self-identified respondents to complete the survey (as based on 2010 VI), the actual response rate will probably not correspond.

    As you are not always sampling the same group and to some extent the group is self-selecting strange results can crop up. Polls are not analogous to the whole population despite weighting to attempt balanced demographics. Unfortunately the press does not care to represent this basic shortfall in the polling process and instead treats the mariginal polling sample as a proxy for the whole nation, hence Labour can be six points clear on Thursday (based on polling conducted on 01st October 2014) and have been clearly ahead by four/ five points for nearly three years, then suddenly the tables turn, solely on the basis of one speech, so that the lead vanishes and the Conservatives are ahead by 1 point on Friday (Based on polling conducted on 02nd October 2014). This is not credible, the great man theory of history was discounted a long time ago, and Cameron is not anyones idea of a great man.