Polling in the weekend papers is dominated by the Conservative leadership race. The Mail on Sunday has a Survation poll, or more to the point, two Survation polls. A full one conducted on Wednesday and Thursday and then a second one conducted on Saturday after the news story of the police being called to Boris Johnson’s flat had broken.

I would always urge some caution with “Has X made you more or less likely to support Y” questions. Some people answer them in a way to register their approval or disapprove of the event or the candidate, rather than whether it has really changed their mind. Hence lots of people who really loathed Boris Johnson anyway will have said it has made their opinion worse, when actually they would probably never have supported him anyway. It also explain the rather perverse finding that 9% of people say the story makes them them more likely to support Boris Johnson – I expect those are actually just people trying to express their pro-Boris Johnson opinion, rather than it actually having improved their opinion.

The much more better way of measuring change is to compare before and after preferences. On Wed/Thurs Survation asked who would make the better Prime Minister, finding the public preferred Johnson to Hunt by 36% to 28%. They polled the same question again on Saturday and found the balance had shifted, with Johnson on 29%, Hunt 32%. Among Conservative voters Johnson continued to lead, but by a smaller margin – the break was Johnson 55%, Hunt 28% on Wed/Thurs, Johnson 45%, Hunt 34% on Saturday.

This gives an early indication that the story has shifted public opinion against Johnson a bit – though as ever, I would urge some caution. It was taken just as a story was breaking when it was all over the news. Whether it has any impact in a few weeks time is a completely different question. It is also important to remember that the views of Conservative voters are not necessarily a good guide to the views of Conservative party members.. Full tables for the Survation polls are here and here.

(A quick note for methodology geeks. On their main poll Survation are now including the Brexit party in the main prompt alongside the Conservatives, Labour and Liberal Democrats. Green party, UKIP and ChangeUK are in the secondary prompt. More interestingly, the Wed/Thurs poll was also weighted by recalled 2019 European election vote, which appears from the weighted/unweighted numbers to downweight 2019 Labour voters quite substantially and bump up the Lib Dems and Greens. I don’t know if that’s a permanent change they are adopting.)

There is also a ComRes poll in the Telegraph – it is headlined as a poll of “grass-roots Tories”, but it is in fact a poll of Conservative councillors, not of ordinary Conservative party members. The two things are really not interchangeable. For what it’s worth though, among Tory councillors Johnson leads Hunt by 61% to 39%. It was carried out on Friday and Saturday, so would have straddled the Johnson domestic row story. It not clear how much of the fieldwork was before and after the story breaking.

Finally we come to the people who actually do have a vote in this election. YouGov had a new poll of Conservative party members in yesterday’s Times. The fieldwork for this was between Wednesday and Friday, so was before the story about the police visiting Johnson’s flat had broken. However, it underlines the huge lead that Johnson had among members – he led Hunt by 68% to 23% (74% to 26% once don’t knows are excluded), with 80% of members saying they were already fairly certain who they would vote for. Johnson would really need to make a mess of things to throw away a lead that large. The other interesting pickings from that poll where that while Tory members were voting for Boris, many didn’t actually trust him – only 47% thought he could be trusted to tell the truth, 40% did not.

So, all in all, the Survation poll raises the possibility that the Johnson domestic had some impact, but it’s only one poll, done in the immediate aftermath. I’d wait to see if it lasts once the story is off the front pages. In the meantime, polling of the people who can actually vote in this contest suggest Johnson has such a large lead that it would take something major to throw it away.


There are two very different elements to polling of the conservative leadership race: polling of Conservative party members – used for predicting who is going to win, and polling of the general public, which is generally being used to argue about the electoral appeal of the different candidates.

Let’s take members polling first. The only professionally conducted polls of party members are done by YouGov, with the most recent conducted last month just before May’s resignation. It found Boris Johnson was the first choice for members on 39%. More importantly the poll asked party members to rank candidates in order of preference, allowing YouGov to work out head to head figures for each potential pair of candidates. Suffice to say, Boris Johnson won them all. The closest pairing was Johnson vs Raab on 59%-41% though given these two are both appealing to the brexiteer elements of the parliamentary Tory party that seems an unlikely run off. The more plausible contests of Johnson v Gove, Johnson v Hunt or Johnson v Javid would all be clear victories for Johnson. In the event Johnson does not get through, Dominic Raab also beats the remaining candidates, though by less convincing margins. As things stood in May, whichever of the leading “hard Brexit” candidates, Johnson or Raab, reached the final two would win.

It is worth remembering that data is from back in May, so it is possible that opinion has changed already. Certainly there is still time for opinion to change in the future. The only other data we have on party members is from ConservativeHome’s surveys of their mailing list of party members. I think they last did paired run-offs in April, but they’ve asked about members’ preferred leader more recently and found Johnson retaining a strong lead.

Perhaps more open to interpretation are the polls of the general public, especially since they are often used to make the case for various candidates in terms of their electoral appeal. Polls about leadership candidates are often very much exercises in name recognition – the fact is that many of the people being asked about are relatively obscure figures who most people who are not political anoraks know little about. If you ask the public whether Mark Harper would make a good or bad Prime Minister then the overwhelming majority of people obviously say they don’t know who Mark Harper is (most of those who do answer the question are Labour and Lib Dem supporters giving negative answers, presumably on the basis that they feel any Tory would be a bad Prime Minister!)

Any attempt to gauge public attitudes towards the candidates needs to be viewed through this prism. Here, for example, are the most recent YouGov figures on if people will make good or bad Prime Minister. Boris Johnson is one of the best known politicians in the country, so has the highest proportion thinking he will make a good Prime Minister (26%). However, he also has the highest proportion thinking he would be a bad Prime Minister (53%). Michael Gove and Jeremy Hunt are also familiar to most people, though both have proportionately more negative ratings than Johnson (Gove 15% good, 51% bad; Hunt 15% good, 46% bad). After that recognition falls away – 55% of people gave an opinion about Sajid Javid (18% good, 37% bad), 42% Dominic Raab (14% good, 28% bad) and so on.

Some have used this to argue the lesser known candidates are more popular candidates on the basis of their net figures, or the proportion of those who know who they are who give them a positive rating. For example, in the YouGov poll of those who expressed an opinion about Rory Stewart over 40% were positive… but that’s because over 70% of people didn’t have an opinion. There is no guarantee that the opinions of the 29% who did are a reflection of what the rest of the country might think were they to form an opinion of him (though in their defence, it may be easier to start with a blank slate and convert people who have no opinion than change the views of those who have already formed negative perceptions of the better known candidates). On the subject of Rory Stewart, it should be noted that he is the only candidate who has really improved his ratings substantially during the campaign, albeit from an almost non-existent base. In May 5% thought he would make a good PM, by last week that had risen to 12%. On the other hand, the criticism that he is the candidate popular amongst people who aren’t Conservative does seem to have some truth to it – his best ratings are amongst Labour and Liberal Democrat voters.

That brings us more directly to the issue of the electoral impact; which candidate would do better at winning over voters at a general election? I should begin by adding a caveat here – people are not necessarily good judges of these matters. They may have an idea of whether they like Boris Johnson or Michael Gove, but they don’t know the policies they are campaigning on, how the media are reporting them, whether the party has united behind them and so on. We are asking people to imagine a hypothetical situation when they really don’t have much to go on. In cases where people don’t even know much about the candidates themselves, like Matt Hancock or Rory Stewart, I don’t think there’s any real point in even asking the polling question. Respondents simply don’t know enough to judge.

For the better known candidates, it can at least tell us something and, when it is asked, there is a clear pattern. This YouGov poll for Lord Bell found the Conservative party on 29% under Boris Johnson, 24% under Dominic Raab, 21% under Hunt, 20% under Gove and 22% under Javid. Johnson clearly does better – but it appears to be a straight forward transfer of support directly from the Brexit party, who drop to 13%. There is a similar but smaller effect from Dominic Raab becoming leader. The fairly obvious interpretation is that the impact we’re seeing here is not Johnson or Raab’s magnetic personalities, but Brexit party voters returning to the Conservative party if the the new leader is someone they trust to deliver a genuine Brexit. That’s certainly something I would expect to happen… but it does also mean that such support would likely be conditional upon the new leader actually delivering Brexit in a timely fashion (and, one assumes, since it would be happening on their watch, delivering a Brexit in a way that isn’t a total disaster).

My advice for people looking for polling clues to future Tory performance under different leaders is that the impact of the candidates’ personalities may in reality be dwarfed by the impact of whether or not they actually deliver something that their potential voters perceive as a successful Brexit.


-->

A quick post about the YouGov poll in Friday’s Times. Topline Westminster voting intention figures are CON 19%, LAB 19%, LDEM 24%, BREXIT 22%.

These are obviously startling figures, unprecedented even. There are historical examples of third parties taking the lead (Cleggmania, for example, or the early successes of the SDP-Liberal Alliance), but I don’t think there are any when the Conservatives and Labour were both pushed out of the top two.

However, even leaving aside the traditional warning that this is “just one poll”, this is one poll conducted in the immediate aftermath of the European elections. Part of what we are seeing is a boost for the Liberal Democrats and Brexit party from doing well in the Euros, getting lots of media coverage and looking like winners. Under normal circumstances we would expect that boost to fade in time (though a success for either of them at the Peterborough by-election could potentially keep it going).

Realistically though, we’ve got several weeks of coverage of the Conservative leadership election ahead of us, followed by the media circus around the elevation of a new Prime Minister. The media agenda will move back towards Labour and the Conservatives, and I’d be surprised if we didn’t seem one or other of them move back into the lead.

Nevertheless, it’s a remarkable poll, and like the election results last week, again brings home the extent to which Brexit is tearing apart the party identities, loyalties and assumptions that have traditionally underpinned our electoral politics. Our party system really does seem to be straining under the pressure. I don’t expect it to break just yet, but looking ahead we still have Brexit itself to deliver (or not, as the case may be). There is almost certainly plenty more political instability to come.


To start with, here’s an update of all the pre-election polls (Ipsos MORI, Survation and NCPolitics all published theirs on the morning of election day, after my last post).

Note that ComRes and Hanbury also produced polls during the campaign, but not with fieldwork conducted on or after the final weekend of the campaign. For what it’s worth, they tended to show high Labour support, though we’ll never know what their polls would have shown in the final week.

Needless to say, the pre-election polls varied wildly from one another for all the main parties. Labour had a twelve point spread (13% to 25%), the Conservatives eight points (7% to 15%), the Liberal Democrats (12% to 20%), the Brexit party eleven points (27% to 38%). In the event, the polls that had low Labour scores and high Liberal Democrat scores were closest to reality. Compared to the final results, Ipsos MORI took the laurels, getting close to the correct result for all parties. YouGov were next, getting within a point or two of most parties but overstating the Brextit party. Other companies recorded significant errors, with a few double-digit overstatements of Labour support.

It is difficult to point at a single obvious cause for the wide variation. When there were huge differences between polls at the 2017 election the reasons the were clear: pollsters had adopted demographic turnout models and other post-fieldwork adjustments which backfired and overstated Tory support. There is no such easy explanation for the 2019 polls – pollsters have mainly reversed the missteps of 2017 and, while there are some variations in approaches to turnout, the elaborate turnout models that made such a difference in 2017 have disappeared. Different approaches to turnout perhaps explain differences of a point or two, they don’t explain differences of 10 points. The differences here look as if they are more likely to be down to pollsters’ different approaches to sampling or weighting, and the representativeness of their samples.

From the beginning these European elections were going to be a challenge. They are a low turnout election, when at recent elections polls have struggled to correctly reflect the demographic pattern of turnout. In recent decades most British pollsters have also relied upon past-vote weighting to ensure their polls are politically representative, and this was an election when past vote was a particularly poor predictor of current voting intention.

In terms of what this means for wider polling, errors here don’t necessarily transfer directly across to Westminster polls. The challenges posed by high-turnout elections can be very different to those posed by low-turnout elections and just because some polls overstated Labour support in the European elections does not necessarily mean they are overstating Labour support for general elections. On the other hand, given the recent history of errors, it probably isn’t something we in the polling industry should be complacent about.


There are five polls with fieldwork conducted at least partially since the weekend – I don’t know if there are more to come overnight (I think there may be at least one more. ComRes and Survation have both polled during the campaign, but I don’t know if either are doing a final call):

Panelbase (14th-21st May) – BREX 30%, LAB 25%, LDEM 15%, CON 12%, GRN 7%, ChUK 3%, UKIP 3% (tabs
Kantar (14th-21st May) – BREX 27%, LAB 24%, LDEM 15%, CON 13%, GRN 8%, ChUK 5%, UKIP 4% (tabs)
Opinium (17th-20th May) – BREX 38%, LAB 17%, LDEM 15%, CON 12%, GRN 7%, ChUK 3%, UKIP 2%
YouGov (19th-21st May) – BREX 37%, LAB 13%, LDEM 19%, CON 7%, GRN 12%, ChUK 4%, UKIP 3% (tabs)
BMG (20th-22nd May) – BREX 35%, LAB 18%, LDEM 17%, CON 12%, GRN 8%, ChUK 4%, UKIP 2% (tabs

The broad story across the polls is the same – the Brexit party are ahead, Conservative support has utterly collapsed, the Lib Dems are doing well in the mid-to-high teens, and both Change UK and UKIP have failed to shine. There is more variation in the detail, and particularly in how well or badly Labour are doing. Kantar and Panelbase have them not far behind the Brexit party; Opinium and BMG have them down in the teens, YouGov have them below the Liberal Democrats in third place.

This isn’t an election like 2017 when pollsters took very different approaches and the differences are easy to explain. The polling companies aren’t taking radically different approaches – there are some differences in turnout modelling (BMG and Opinium, for example, are taking only those most certain to vote, which will be boosting the Brexit party and Lib Dems), Kantar are estimating the likely vote who say don’t know based on their demographics and answers to other questions, which explains their comparative low figure for the Brexit party (they’d be on 31% otherwise). And don’t overlook simple things like when the fieldwork was conducted – all the polls have been showing a downwards trend in Labour support, so it may not be co-incidence that the polls from Panelbase & Kantar whose earliest fieldwork is over a week old have higher support for Labour.

The bottom line however is that this is a tricky election. Firstly, turnout for European elections is normally low (and one of the problems with polls in recent years is getting too many of the sort of people who vote, and not enough of those who don’t bother). Secondly, most polling companies rely on some degree to weighting by past general election vote to make sure their samples are representative, as how people voted at previous elections normally correlates pretty well with their current vote. An election like this, when an awful lot of people are not voting for the party that they voted for at the last election, will make those techniques less effective. We shall see on Sunday.

In the meantime, several people have asked me about exit polls tomorrow. There won’t be any. The big, offical BBC/ITV/Sky exit poll is only conducted at general elections anyway, but even if they wanted to, they couldn’t do one tomorrow. For the European elections the rules that ban the publication of exit polls until after polls close apply across Europe, so it wouldn’t be legal to public any exit poll until the polls have closed everywhere in the European Union… and some countries won’t finish voting until Sunday night.