Varför ger de brittiska opinionsmätningarna olika prognoser?

JKLs vänner på den Londonbaserade public affairs- och kommunikationsrådgivningsbyrån Lexington Communications har djupdykt i opinionsundersökningarna inför den brittiska EU-folkomröstningen. Att de skiljer sig så mycket åt – vad säger det om de brittiska väljarna och vad kan det bero på?

Are Britain and the EU about to be hit by a political earthquake?

That is what a Brexit vote on the 23 June would amount to. But it doesn’t feel like one is on the way. Most polls have pointed to a win for Remain and the betting markets point heavily in that direction. But are we looking at the right Richter scales? Phone polls consistently give Remain a big advantage which has served to skew the overall average. Yet online polls have consistently measured the contest much closer, giving Leave the lead on a number of occasions. No one has been able to satisfactorily explain the difference, and despite the enormous import of the referendum none of the pollsters is able to say for certain which forecast is the more accurate. It seems we will have to wait for 23 June to find out. In other words, we are moving towards the most important political decision in modern times with no clear idea of what will happen.

To understand why we are in this position, we present below an analysis of what the current polling suggests and the factors that are most likely to influence the outcome of the referendum.


Polls in the run up to June’s vote are generating surprisingly little interest. That perhaps owes to the experience of the 2015 general election, when not a single pollster accurately forecast the outcome. But even without the legacy of the election there would be good reason to caveat the current survey evidence. Whereas polls that track party support can be assessed over a long period, and be compared directly against the outcome of a previous election, the referendum on the EU has no such benchmark. Britain was last polled on EU membership in 1975 and the electorate has changed considerably since then. So the current measures of public opinion are snapshots taken in relative isolation. Nonetheless, with all those caveats and cautions in mind, the polls still provide some insight into the current mood of the nation.

A close contest?

Overall, they suggest that Britain is on a knife-edge. The latest ‘poll of polls’, an average of the last six polls conducted, puts the contest at 50-50 between Leave and Remain.

If we look at a longer series of polls, Remain has led in around two-thirds of the 41 polls conducted since the beginning of March:

  • Remain: 26 leads
  • Leave: 10 leads
  • Ties: 5

The betting market, which proved a more accurate – though imperfect – predictor of the general election than national polls, favours Remain very strongly. Ladbrokes betting barometer for example, which is based on live odds, gives Remain a 71% chance compared to Leave on 29% (3 May 2016). That represents a strengthening of the Remain position compared to previous odds (e.g. Monday 18th April: Remain 67%-33% Leave). Ladbrokes has consistently put Remain at short odds: currently 1/3 as against Leave on 9/4.

But Remain is favoured in the betting and has held the lead in the majority of polls, significant differences beneath the headline numbers point to a closer contest. The principal issue is the divergence between online and telephone polls.

The former have the contest neck-and-neck, with Leave increasingly registering some leads, while the latter give Remain a strong advantage. If we average out the results of the last six online polls and the last six phone polls, we can see the stark difference:

Why the poll differences?

There is as yet no definite explanation for the difference. Some psephologists attribute the variance to the tendency of online polls to produce a higher proportion of people responding “don’t know” than phone polls. There is some evidence that these ‘don’t knows’ tend to lean more towards Remain than Leave. So it is possible that the online polls are ‘hiding’ Remain voters.

But that supposes that the phone polls are a truer measure of public opinion. Moreover, this interpretation of ‘don’t knows’ is open to challenge and is not consistently supported by the data. The most recent online polls, for example, only report 10% and 13% of ‘don’t knows’ but nonetheless give Leave a slender lead.

When we look at the online and phone polls in isolation, what becomes clear is that there is no relationship between ‘don’t know’ voters and Remain. In both sets of polls, for the most part, the difference between Leave and Remain changes little regardless of the number of don’t knows.  This suggests that there is not a hidden army of Remain voters among the undecided.

Instead the difference between online and phone polls seems to be a product of ‘sampling’. That was the conclusion of ComRes, which conducted an experiment to try and understand the divergence, asking the referendum question in online and telephone surveys within days of each other, with the same demographic weights applied and with past votes weighted in the same way. Despite this, they found the same divergence that other surveys had report, with online polling showing the public split whereas the telephone polling showed a large lead for “Remain”:

According to ComRes, the variance must owe to differences in the kind of people who take part in phone as opposed to online surveys, despite their demographic similarities. Specifically, ComRes concludes that “People choosing to sign up to an online panel are by nature more online savvy. They are more likely to be engaged on social media and exposed to strongly held beliefs that we see in online encounters. Indeed, by being on an online panel, regularly receiving surveys probing your thoughts about particular details of current affairs you are becoming even more engaged and aware of the issues.” ComRes suggest that while this different exposure may not matter at general elections, where people are subject to long term cultural exposure to the parties, at a referendum “there is a much more ‘low information’ electorate, making any polling about the issue particularly sensitive to differences in the political engagement of the sample being surveyed.”

The implication here is that the more knowledgeable and engaged a group, the more likely it is to contain people who favour leaving the EU. Ignorance, on the other hand, encourages people to favour the status quo.  That is one explanation for the gap between phone and online polls, but instinctively it does not feel like the whole story.

It also begs the question about how public opinion may shift as the campaign progresses and people become more exposed to information and argument.

Key factors that will influence the outcome

There are a number of factors highlighted by the polling which may have a significant bearing on the final result, and which underline how important the political dynamics will be in the final weeks of the campaign:

Low turnout may benefit Leave: Most polls have found that Leave voters tend to be more resolute in their determination to take part than Remain. On the face of it that suggests a lower overall turnout would benefit Leave. However, there is not enough evidence to draw firm conclusions at this stage and as polling day nears the various pollsters will be devoting more attention to the issue of propensity to vote and its likely impact on the result.

Labour voters are important: Of all groups polled, Labour voters are the most likely to support Remain. Energising these voters to turn-out is critical for Cameron and perhaps underlines how important Jeremy Corbyn’s recent intervention to support the EU was. ICM recently showed that 60% of Labour voters would back Remain, whereas Conservative voters are evenly split. Labour’s latest bout of infighting could hamper the party’s pro-Eu campaigning in the run-up to polling day.

The referendum could be won or lost in the North: The majority of polling shows that the North, Midlands and Wales are the most undecided regions. The North, in particular, is likely to be a key battleground with a higher proportion of undecided voters who could be swayed either way. A recent Com Res poll showed that 43% of those polled may change their mind compared to a lower proportion in the South.

The message carrier matters: We know that individual politicians have influenced the result in previous referendums asked in other countries. Recent polling from Com Res suggests that the three most influential political voices in the UK poll are David Cameron, Boris Johnson and Jeremy Corbyn. Interestingly, Johnson is seen to be a more influential voice in the North than David Cameron, although Jeremy Corbyn is also cited by over a third of Northern voters as potentially influencing their vote, especially among Labour supporters.

There is a clear divide between younger and older voters: Older voters are more Eurosceptic than younger voters, who are more pro-European. Significantly, though, older voters have a much stronger record of turning out in elections than their younger counterparts.

Women are less committed than men: Persuading women to vote will be critical if the campaigns are focused on driving turnout amongst this group. Currently, unskilled female workers in the North are the most ‘undecided’ group, particularly those aged between 45-55. More broadly, a recent YouGov poll showed that nearly twice as many women as men were undecided. In contrast, middle class men in the South are the most resolute in their determination to vote.

The economy and immigration remain critical factors: The economy comes top of most voters’ concerns in the vast majority of polls, with immigration second; suggesting that Remain have great potential to make messages framed around risk and fear stick. It is also worth noting that concern about ‘economic risk’ is shared across most age groups and regions. Leave will need to work hard to stop this focus on the economy dominating the campaign.

Few electors are strongly committed, making the outcome vulnerable to events: While the decision to remain in or leave the EU is of immense importance, it is generally described as a ‘low salience’ issue. Only a small minority feel strongly either way about Europe. Most electors may have an instinct about how they will vote but a large number (polls suggest 20%-25%) could yet be persuaded to change their mind. Beneath the broad question of EU membership, the most salient issue is the economy, which as noted above works for Remain. The next most salient – immigration – works for Leave, but in at least four general elections it is an issue that has influenced the political mood but not really switched votes.  That suggests Remain should have the advantage, especially if they can successfully present the Leave campaign as a bunch of extremists who run against the mainstream.  But the lack of strong feeling on the bigger question means that the contest could be more influenced by a particular event or a bandwagon effect than a normal election, making the final result unpredictable and the campaign important.

from Lexington Communications