Reflections on the election polls – and creating a Conservative Party people need not be shy about supporting

  • 12 May, 2015
  • Politics
  • Polling

This is the text of a speech I gave last night at the Post-Election Conference, jointly hosted by Conservative Home, the TaxPayers’ Alliance, Business for Britain and the Institute of Economic Affairs.

Good evening. You might think it’s a bit much to kick off this post-election conference with, of all people, a pollster. If so, I can understand your scepticism. After last week I’m not sure whether it’s worse to be a pollster or a Liberal Democrat. But as I will explain, it would be a mistake to dismiss the polls out of hand, or to think there is nothing we can learn from them.

But before I go any further, let me congratulate David Cameron and his campaign team – and all the candidates who defied the polls – for what was undeniably an absolutely remarkable result. It was all the more remarkable for being so unexpected.

Not only did no pollster anticipate it, nor did any forecaster, academic, commentator, pundit, journalist or party insider. I believe the highest public prediction – made by Harry Cole, the Sage of Guido – was for 297 Conservative seats. The Tory leadership were not expecting this result, and Ed Miliband had drafted a victory speech. Even the exit poll, which left everyone in the political world open-mouthed in disbelief, understated the scale of the Conservative triumph.

Though I published a weekly national poll and surveys of 167 constituencies, I never made a prediction. There were two main reasons for this. First, the possibility of a late change of heart, and second, the fact that polls themselves can affect the outcome in a constituency as people see which way things are going and change their vote accordingly. Indeed even on the night, speaking at a private event, I declined to be drawn. (I said instead that the British people “are good judges of character and they don’t believe the unbelievable. Whatever they have decided today, I’m sure they knew what they were doing”.)

But as you would expect, since Friday I have looked long and hard at my polling, especially that conducted in the last few weeks and months. It would be nonsense, and an insult to your intelligence, to claim my research detected what was about to happen across the board. But I can point to some successes.

First of all, Scotland. My snapshots identified the correct winner in 25 of the 29 seats I surveyed in 2015. The average difference between each party’s vote share in my poll and the share they received on the day was 3.3%. Of the ten Scottish polls conducted in April, the correct winner was identified in seven and the average difference in vote share was just 2.2%, well inside the normal margin of error. It was this polling that confirmed suspicions of an SNP surge and ultimately gave rise to the fears of a Labour-SNP pact that helped swing the election.

In England and Wales, the constituency polls I conducted during the campaign had an average “error” of just 3%, and identified the right winner seven times out of ten.

A number of the seat-by-seat polls did not match the eventual result – not that they were meant to be predictions, as I may have mentioned. But at the risk of undermining my reputation for modesty, some of them were bang on. The Tory margin of victory in Battersea, the Greens’ challenge in Bristol West, the tight race between the SNP and the Conservatives in Michael Moore’s Berwickshire seat, the outcome in the former three-way marginal of Bristol North West and the narrow Labour victory in Wirral West, to pick a few, were on the money.

It is also worth remembering some of the criticisms of my constituency polls that were made before the election. By not naming candidates, I was told, I was understating the importance of incumbency and therefore seriously underestimating the number of Lib Dem MPs who would be re-elected. So much for that.

At the same time, I was told, having methods designed to take some account of shy Tories meant I was understating UKIP in their target seats and would be embarrassed by my polls putting the Tories ahead in Thanet South and Rochester & Strood. Indeed Mr Farage accused me of indulging in “voodoo” to concoct a Tory lead in his constituency. But the Tories won both seats, and in both seats UKIP received the same share on Thursday as they had in my polls.

The weekly Ashcroft National Poll raised some eyebrows for having found only one Labour lead in 2015, and for putting Labour on just 30% in the last three weeks before polling day. The six-point Conservative lead at the end of April, dismissed at the time as an outlier, looks rather different in retrospect.

But there is no getting away from the fact that, in common with other pollsters, my research did not anticipate Thursday’s result and, in particular, the scale of the Conservative vote in England and Wales. There are several possible reasons for this.

One is that there was a late swing. This would have to have been very late indeed, too late even to be captured by the final polls completed the day before. It could be that on the day itself, significant numbers of people decided with the security of an economy going in the right direction and the more credible Prime Minister. The risk, as they saw it, of a weak Labour government propped up by the SNP probably also played a part. Indeed the very closeness of the polls published in the last couple of days before the election, and on the day itself, may have made people think again whether they really wanted the change they were likely to get.

Another factor at play could be the micro-targeting key groups of voters by the Tories in critical seats, effective enough to make a difference at a local level but in numbers too small to show up in national polls. But this would not explain the discrepancy between the final national polls and the vote shares the parties ultimately achieved.

Next is the possibility that people’s voting behaviour has changed in response to the new political landscape, with a coalition government and the emergence of new parties – and that these changes have made it more difficult to create politically as well as demographically representative samples than the pollsters had fully grasped. People’s previous voting behaviour has become a less reliable guide to what they might do in the future.

Finally there is the suspicion that people were not always entirely straightforward when interviewed by pollsters. Some people almost certainly gave themselves the benefit of the doubt when asked how likely they were to turn up and vote: the polls indicated a much higher level of expected turn-out than actually came to pass.

And, of course, some people will not have stated, or admitted, which party they really intended to vote for. This is a problem the pollsters have encountered before, in 1992, and the mechanisms they developed to cope with this problem led to a dramatic increase in accuracy over the next four elections. Clearly those methods did not detect the scale of Conservative support at this election.

All these factors may have played a part, and the answer probably lies in a combination of all of them. There is a lot of hard thinking to be done in the polling world, and it has already started with the inquiry launched by the British Polling Council. Pollsters have shown themselves to be willing to learn and change in the past, and I am sure that will be the case again. That is my snapshot – I won’t be making any prediction about what the inquiry will conclude and I won’t be commenting any further on the matter before its work it complete.

But this conference is about the future of the Conservative Party, not the future of polling. Everybody here will be delighted that the party has confounded the pollsters, and I don’t blame you. But just as I have always been at pains to help ensure the Tories don’t learn the wrong lessons from defeat and disappointment, it is just as important not to learn the wrong lessons from success.

One such mistaken lesson would be that Conservative support is always bigger than it looks in the polls, or that it will remain so throughout this parliament and into the next campaign.

If, as many believe, this election was won by shy or reluctant Tories, the party’s response for the longer term should be to ask what makes them so shy or reluctant in the first place.

On election day itself I conducted a survey of more than 12,000 people who had already voted. The purpose was not to replicate the exit poll but to find out more about why people had chosen the party they had.

For those who voted Conservative, these were the most important factors: the party had the best available Prime Minister, they liked the Tories’ motives and values, they preferred the Tories’ specific promises, and they thought the Tory team would make a more competent government.

For Labour voters, the list is different. By far the most important factor was that they trusted Labour’s motives and values more than those of the other parties. Next, that they preferred Labour’s policy promises. Rather fewer thought the party had the best available PM, and the fourth most important reason for Labour voters was that they had always voted Labour.

To look at it slightly differently, voters for whom choosing the best Prime Minister was the single most important factor chose the Conservatives by a landslide.

The same is true among those for whom the biggest priority was choosing the most competent government.

But among voters for whom the single biggest driver was the parties’ motives and values, Labour pipped it.

In other words, “competence voters” went with the Tories and “values voters”, albeit by a smaller margin, went with Labour. For me, this prompts the thought: why can’t we have both?

People may have been reluctant to admit to pollsters (or perhaps even to themselves) that they were going to end up voting Tory, but we know the reservations people have about the Conservative Party. Last week’s result shows that those reservations were outweighed by the fact that the Tories were doing a good enough job in government to be allowed to continue, or that their concerns about Ed Miliband and Labour were even greater, especially the thought of them having to team up with the SNP.

But as long as people feel they have to choose between competence and values, the Conservatives will always be vulnerable to a Labour Party that starts to look as though it knows what it is doing and rediscovers an understanding of aspirational middle Britain. That might seem a distant prospect, but five years is a long time and we all remember what happened between 1992 and 1997. Labour may be 99 seats behind the Conservatives, and unless the party’s recovery in Scotland happens almost as fast as its demise, the same questions about SNP influence could hinder its next campaign. But Labour’s strategy will not always be as narrow, or its economic rating so poor, or its leader so unconvincing as in 2015.

The key here lies in the priorities the new Conservative government chooses to project. Over the next five years many vital issues will present themselves: an EU referendum, reform of the Human Rights Act, boundary changes, and of course the Scotland question – but in dealing with the urgent the party should not lose sight of the important.

As I found in my post-vote poll, those who voted Conservative were as likely to name the NHS as one of the most important issues for themselves and their families as they were to mention the economy.

And overall, those who put the NHS near the top of their own agenda voted Labour by a nine-point margin. Amid the other pressing priorities, we need to remember that commitment to public services is one of the reservations many people still have to overcome in order to vote Conservative.

My post-vote poll also uncovered some differences in attitude between Labour and Conservative voters that I think are revealing, and encouraging.

By a huge margin, those who voted Conservative on Thursday think that “if you work hard, it is possible to be very successful in Britain no matter what your background”. Labour voters disagree by nearly two to one.

Most Tories expect that “for most children growing up in Britain today, life will be better than it was for their parents”. Again, Labour voters disagree by two to one.

Two thirds of Conservative voters think changes in society are bringing more opportunities to improve their standard of living than threats to it; three quarters of Labour voters think the opposite.

Three quarters of those who voted Labour think people have a right to things like housing and enough to live on, and that it is the government’s responsibility to provide them. Three quarters of Tories are more ready to say people have a responsibility to provide for themselves.

And while Labour voters are divided over whether life in Britain is better or worse overall than it was 30 years ago, Tories are not so ambivalent – they say it is better.

In other words, the Conservatives need no longer be described as the party for the few. The Conservative Party can be – indeed, already is – the party for people who are optimistic, open-minded and self-reliant. And now it has an unexpected opportunity to prove that this is the case – and to connect with those who still think the Tories are not for people like them.

Not only that, it seems to have the intention of doing so. David Cameron promised on the steps of Number Ten to govern “as a party of one nation, one United Kingdom”, and to ensure the recovery “reaches all parts of our country, from north to south, to east to west”.

If he can deliver on that we will have a Conservative Party that not only will people vote for, they won’t be shy of saying so. At that point, the future of the Conservative Party will be assured.

Thank you very much.

Read this speech on

Related Stories