Gangster, huckster, hipster, trickster… now pollster enters the pejorative lexicon. We are committed to ensuring that our methodology is as accurate as possible, and our election review is ongoing.
Our eventual solution will be a durable, long-term improvement to the methodology, rather than a retrofitted analysis of what went wrong. There is unlikely to be one convenient explanation for why the polls were out.
As we go through each theory systematically, what strikes us already is that the Labour Party, if it can suppress the reflex to criticise the polls, could still learn a lot from the data. We cannot share the same sense of vindication as a Dan Hodges or a Philip Collins, but our data certainly build on their analysis.
Weaponising a single issue is too simplistic
The issues that defined the election campaign were the economy, the NHS, immigration, and the future of the Union.
Like many failed leaders before him, Ed Miliband picked the voter concern that was his party’s strongest policy area – the NHS – and tried to weaponise it (note to future leaders: “weaponise” is not voter-friendly terminology).
This ran into two problems. One was that it was predictable, so the Conservatives shut down the issue with set-piece attacks on Labour-run NHS Wales, by dropping ‘dead cats’ on the table, and finally by announcing a surprise, unfunded £8bn of additional healthcare spending. Indeed, much the same thing happened to UKIP on immigration. In the end, the Conservatives anticipated the immigration debate and went so hard on the issue in late 2014 that it had run its course come election time.
The second problem with weaponising the NHS was this: what voters say matters most is not always the strongest driver of how they actually vote. The NHS is a much-loved institution providing cradle-to-grave healthcare in times of great need. But between cradle and grave, our focus group discussions show that people are mostly driven by more prosaic concerns: work, money, family, social lives.
Issues can be sliced and diced in different ways. Rankings of specific issues tend to emphasise those which are emblematic of broader concerns – the NHS, immigrants, crime. The economy is a much more nebulous thing that flows into every area of life. Reforming the bedroom tax and introducing a mansion tax are not the ‘bigger picture’ policies voters expect on the economy.
Labour cannot win without the middle classes
The bedroom tax and mansion tax were also misguided because of the demographics of voter turnout.
As we reported before the election – and as implied by the results on the night – some demographics are considerably more likely to vote than others. The more affluent (ABC1) middle classes vote in much greater numbers than less affluent (C2DE) voters, and were breaking towards the Conservatives:
We are not arguing here for Blairism or centrism – that is a debate for the Labour Party and others to have. But any strategy focused on issues relevant only to the less affluent, the young and the disgruntled will fail because these groups tend not to vote in sufficient numbers to win general elections.
The next Labour leader will need to find an issue that cuts across classes and generations if he or she is to win an election. Aspiration, education and enterprise are worth revisiting. Of course, Labour will need a leader who can also talk to the working class voters it has lost to UKIP.
Where were the Miligrans?
We heard a lot about the Milifans but where were the Miligrans? Our pre-election aggregated figures showed that over-65s split 43% Conservative versus only 27% Labour (indeed, as more data become available we may eventually conclude that the gulf was even wider).
Even more worryingly for Labour, while the population in general would prefer a Labour-run NHS, the crucial over-65 age group are more likely to trust the Conservatives (42%) than Labour (26%) to manage it. As the age group with most exposure to NHS services, this should worry any leadership candidate planning to major on healthcare credentials.
The polls were wrong but they were not the problem
Voting intention methodologies need improving. But we should demolish one line doing the rounds: “Had the polls been more accurate we would have seen a different campaign.”
The argument is essentially that because the polls did not show the Tories ahead, the Tories surged ahead, so the polls should have shown the Tories ahead, which would have prevented the Tories from surging ahead. You might call this the Penrose stairsargument, after the optical illusion of an impossible staircase looping forever upwards and downwards:
Instead, there were plenty of giveaways in the polling data itself. As part of our post-election review, we have experimented with different ways of estimating the likely voting intention of undecided voters.
An interesting option is reassigning undecided voters to the party of the leader they think would make the best Prime Minister, rather than to the party with which they most closely identify (which is our historic approach to dealing with those reluctant to express a voting intention). This change is quite dramatic:
Obviously, this looks striking with the benefit of hindsight, but there are reasons why this may not be a long-term solution to voting intention polling. Retrofitted models that work for GE2015 may simply end up being context-specific. For example, the above adjustment may have led to a greater overestimation of Labour’s share during the Blair years – an error which would have exacerbated a pre-existent tendency for the polls to flatter Labour. As we keep saying, there will be no simple explanation for why the polls were inaccurate this time around.
But the conclusion Labour might want to draw is that many people who share the values Martin Freeman expressed in their party election broadcast were nonetheless unwilling to vote Labour. These are not “Shy Tories” or “Lazy Labour”. They are just, like most people, conflicted. The “Head Over Heart” voter, perhaps.