As my colleague Adam Ludlow has noted, telephone polls have consistently shown the Conservatives ahead of Labour since the turn of the year. But in the past week or so, ComRes polls have reverted to the dead heat we were showing last year. What has changed?
The problem of reported behaviour
Reported behaviour is a big challenge for pollsters. Ask someone if they recycle regularly, donate money to charity, or pay their bills on time, and you may not get an entirely accurate answer.
Some of this can be attributed to the interviewer effect – the desire to portray oneself favourably to another person – but some is also due to self-deception and poor powers of prediction. Even online survey participants, without an interviewer present, are often excessively optimistic about their own abilities and virtues.
Like many civic acts, we know that people struggle to attach a realistic probability to their likelihood to vote. Around 90% of survey respondents say they are more likely to vote than not vote. Three quarters say they are certain to vote.
And yet average turnout has been just under two thirds (62%) of eligible voters over the last three general elections. We know, though, that some people are more likely to vote than others.
Older people vote
The importance of the “retirement age” demographic can at times be overstated, but over 65s account for about 10m people (22% of the total electorate). They are also significantly more likely to vote.
This skew means that more than a quarter (26%) of all votes cast at a typical general election are cast by people aged 65 and above. By contrast, 18-24 year olds probably account for fewer than 8% of total votes cast. Such disparities are important when you look at party breakdown by age:
Base: Aggregated voting intention across 5 waves, 27th March to 30th April, 18-24 n = 373; 25-34 n = 530; 35-44 n = 658; 45-54 n = 802; 55-64 n = 732; 65+ n = 1075.
Wealthier people vote
Social class also predicts likelihood to vote. The ten seats in Great Britain with the highest turnout in 2010 averaged 76% turnout. They returned six Conservative MPs, three Lib Dem MPs, and only one Labour MP. They tended to be leafy, middle class constituencies like Richmond Park and Winchester.
The lowest turnout seats in 2010 included Manchester Central, Birmingham Ladywood, and Glasgow North East – deprived, inner city constituencies. Across the ten lowest turnout seats, nine Labour MPs were returned. Turnout averaged just 49%.
In seats which are consistently wealthy or deprived, these turnout disparities have little impact on the makeup of the Commons (a high-turnout seat and a low-turnout seat both return one MP). But in seats with a mixture of more affluent and less affluent neighbourhoods, it could tip the balance in a tight race.
Labour will therefore be concerned that they trail behind the Conservatives among every social grade except the least affluent DEs:
Base: Aggregated voting intention across 5 waves, 27th March to 30th April, AB n = 1227; C1 n = 1313; C2 n = 717; DE n = 913.
Rallings and Thrasher state in their report on participation at the 2010 general election that “there was remarkable uniformity across Great Britain in the turnout of postal voters with over 83% returning their ballots in each nation”.
It has been suggested that one of the main aims of the Labour ground campaign has been to capitalise on higher postal turnout, by registering their supporters to vote by post. Indeed, some polls have suggested that Labour are beating their opponents in terms of raw contacts on the doorstep
The only query here would be whether Labour canvassers can confidently identify which people to register for postal votes, given the aforementioned problems with reported behaviour. If you think voters mislead pollsters, imagine how inaccurate canvassing records are.
Home owners are more likely than renters to vote. Little surprise, then, that George Osborne has favoured schemes like Help to Buy and Inheritance Tax reforms. Again, turnout favours the Conservatives:
Base: Aggregated voting intention across 5 waves, 27th March to 30th April, Owned outright n = 1821; Owned (mortgage / loan) n = 1477; Rented from council n = 222; Rented from housing association n = 186; Rented privately n = 359.
So can Labour win?
A quick caveat: many of the factors outlined above are themselves correlated (e.g. age is linked with home ownership, which correlates with affluence). The effect of each individual factor on propensity to vote may not be as strong as implied.
Nonetheless, the recent rise in Labour support in ComRes polls has been driven mainly by younger voters, who traditionally have not voted en masse. The proportion of voters aged 18-24 saying they are certain to vote has risen from 36% in our first poll of the campaign (28th-29th March) to 56% in our most recent poll (29th-30th April). Among 25-34 year olds, there has been a jump in the same period from 46% to 71% now saying they are certain to vote.
This surge raises a few questions. The deadline for voter registration (which is now at the individual, rather than household level) has passed. Similar deadlines for postal vote and proxy vote registration have also been missed.
Perhaps inspired by Labour’s superiority on issues important to young adults, its strong digital campaign, and endorsements from figures like Russell Brand and Martin Freeman, younger people have finally woken up to the need to vote. But have they woken up too late?
If – and it is a very big “if” – Labour can get out the youth vote and other first-time voters in anywhere approaching the numbers currently suggested, Ed Miliband may well make it into 10 Downing Street. It would be an historic achievement.