Another week, another Brexit poll for partisan twitter to get overexcited about. In this case the fuss was caused by a YouGov poll that appeared to show people backing Brexit by 48% to 39%. This survey was actually the GB answers to question asked to several EU countries – the intention of it wasn’t to measure UK support for Brexit, but to see whether or not the public elsewhere in Europe still wanted Britain to stay, or whether we’ve got to point that they’d really just like us to hurry up and go away (for the record, most of the German, Danish, Swedish and Finnish public would still like Britain to stay. The French are evenly divided). There was also a question earlier in the survey about Martin Schulz’s vision of a federal Europe which may or may not have influenced answers – however, this post isn’t about the specific question, but about all the Brexit surveys we tend to see.

As ever, when a poll comes out that appears to show public support for Brexit it is excitely retweeted and shared by lots of pro-Brexit voices. When a poll comes out that appears to show public opposition to Brexit it is excited retweeted and shared by lots of anti-Brexit voices. Both of these create a deeply misleading picture. To start with, there are three different questions about current attitudes to Brexit that people often treat as being measures of public support for Brexit which don’t always show the same answers…

1) Questions asking how people would vote in a Brexit referendum tomorrow
2) Questions asking whether people think Brexit was the right or wrong decision
3) Questions asking whether people think we should now go ahead with Brexit or not

Starting with the first type of question, BMG and Survation both ask EU referendum voting intention regularly, and ICM, Opinium and YouGov have asked it on occassion. BMG’s most recent poll showed a ten point lead for Remain and got a lot of publicity, but this was something of an outlier. Typically these polls have shown a small lead for Remain of between one and four points.

Any question asking about voting intention in a referendum or election is really two questions – it’s working out who would vote, and then how they would vote. When polls ask how the public would vote in an EU referendum tomorrow they tend to find not much net movement among remain and leave voters, the Remain leads are down to those who didn’t vote in 2016. This raises all sorts of questions about whether those past non-voters would actually vote and whether they are actually representative of 2016 non-voters, or are too politically engaged and likely to vote.

There’s also a question of how useful a referendum voting intention question is when there isn’t actually a second referendum due. The most likely route to a second referendum is a referendum on the terms of the deal…which obviously aren’t known yet. In my experience, most people who contact polling companies asking whether we’ve asked a Brexit referendum question aren’t primarily interested in how people would vote in a second referendum, but really want to see if the public have changed their mind about how they voted in the first one…

YouGov regularly ask a direct “Bregret question” to get at that question, asking whether people think voting for Brexit was the right or wrong decision. The results here are quite similar to referendum questions, but because it is a question about public attitudes as a whole rather than voting intentions concerns about likelihood to vote don’t arise. Looking at the regular YouGov tracker, there has again been a slow movement towards Regret, meaning that for the last three or four months the poll has consistently shown slightly more people thinking Brexit was the wrong decision than the right decision.

The final group of questions is “what do we do now” questions. No company asks a regular tracker along these lines, but there are several questions asked on this sort of basis. By stating with “at this point” the question in the YouGov poll this week tilts toward this sort of question, but there are other more explicit examples asking what people think should happen next – for example, YouGov have a semi-regular tracker that asks how the government should proceed with Brexit, which this month found 52% thought the government should go ahead with Brexit, 16% that they should call a second referendum, 15% that they should stop Brexit and remain in the EU. The reason for the difference in these questions is that a substantial minority of people who voted Remain in 2016 consistently say that the government should go ahead and implement Brexit (presumably because they see them as having a democratic duty to implement the referendum result).

It is true to say that more of the public now tend to think Brexit was the wrong decision than the right decision, and say they would vote against it in a referendum. It is also true to say that most of the public think that Brexit should go ahead. Neither measure is intrinsically better or worse, right or wrong… they are just asking slightly different things. If you want to understand public attitudes towards Brexit, you need to look at both, rather than cherry pick the one that tells you what you want to hear.

YouGov’s latest poll for the Times has topline figures of CON 42%(+2), LAB 41%(nc), LDEM 7%(nc). Fieldwork was Sunday to Monday, and changes are from last week. While the movement from the last poll is well within the normal margin of error, it’s worth noting that this is the first YouGov poll since the election to show a Conservative lead.

Looking at some of the other results it does suggest a small boost for Theresa May from the progress on Brexit last week, but one that still shows the public judging the government’s negotiating efforts very negatively. 26% of respondents now think the government are doing well at negotiating Brexit (up five points), but 57% still think they are doing badly (down seven points). Asked who has the upper hand in the negotiations so far 50% think the EU are doing better and Britain are accepting their demands, 26% think there has been give and take on both sides and just 4% think Britain has the upper hand.

Of course, this is just one poll done just after some good news for the government. It remains to be seen whether it is replicated in other polls and, if so, whether it lasts or rapidly fades away.

The Times story is here, tabs should be up on the YouGov site tomorrow.

Survation have a poll in today’s Mail on Sunday. Topline figures are CON 37%(-1), LAB 45%(+1), LDEM 6%(-1). Fieldwork was Thursday and Friday and changes are since early October.

The eight point Labour lead is the largest any poll has shown since the election, so has obviously attracted some attention. As regular readers will know, Survation carry out both telephone and online polls. Their telephone method is unique to them, so could easily explain getting different results (Ipsos MORI still use phone polling, but they phone randomly generated numbers (random digit dialling), as opposed to Survation who phone actual numbers randomly selected from telephone databases). However, this was an online poll, and online there is nothing particularly unusual about Survation’s online method that might explain the difference. Survation use an online panel like all the other online polls, weight by very similar factors like age, gender, past vote, referendum vote and education, use self-reported likelihood to vote and exclude don’t knows. There are good reasons why their results are better for Labour than those from pollsters showing the most Tory results like Kantar and ICM (Kantar still use demographics in their turnout model, ICM reallocate don’t knows) but the gap compared to results from MORI and YouGov don’t have such an easy explanation.

Looking at the nuts and bolts of the survey, there’s nothing unusual about the turnout or age distribution. The most striking thing that explains the strong Labour position of the poll is that Survation found very few people who voted Labour in 2017 saying they don’t know how they would vote now. Normally even parties who are doing well see a chunk of their vote from the last election now saying they aren’t sure what they would do, but only 3% of Labour’s 2017 vote told Survation they weren’t sure how they would vote in an election, compared to about 10% in other polls. Essentially, Survation are finding a more robust Labour vote.

Two other interesting findings worth highlighting. One is a question on a second referendum – 50% said they would support holding a referendum asking if people supported the terms of a Brexit deal, 34% said they would be opposed. This is one of those questions that get very different answers depending on how you ask it – there are plenty of other questions that find opposition, and I’m conscious this question does not make it clear whether it would be a referendum on “accept deal or stay in EU”, “accept deal or continue negotiations” or “accept deal or no deal Brexit”. Some of these would be less popular than others. Nevertheless, the direction of travel is clear – Survation asked the same question back in April when there was only a five point lead for supporting a referendum on the deal, now that has grown to sixteen points (50% support, 34% opposed).

Finally there was a question on whether Donald Trump’s visit to the UK should go ahead. 37% think it should, 50% think it should not. This echoes a YouGov poll yesterday which found 31% think it should go ahead, 55% think it should not. I mention this largely as an antidote to people being mislead by twitter polls suggesting people want the visit to go ahead – all recent polls with representative samples suggest the public are opposed to a visit.

Tabs for the Survation poll are here.