The Times’s first YouGov poll since the election was called has topline figures of CON 48%(+4), LAB 24%(+1), LDEM 12%(nc), UKIP 7%(-3). The Conservative lead of twenty-four points is the highest they’ve recorded from YouGov since way back in 2008. In terms of a starting position for an election campaign this is a huge gap – to put it in context, when the 1997 election was called, polls in the first week put Labour between 21 and 29 points ahead of the Tories. The Tory lead now isn’t as large as Blair’s huge Labour lead then… but you can see we’re in the same sort of territory.

More interesting to me is that UKIP score – the lowest YouGov have shown since 2013. This echoes the ICM flash poll yesterday, which also also had UKIP dropping sharply to a record low. While I’d still like to see it repeated in other polls before assuming too much, it looks distinctly as if an actual election being called has led to some people who were saying they would have voted UKIP switching to the Tories. Perhaps it’s the sudden difference between a theoretical election that could be three years away, and thinking about what they might do in an election just seven weeks away.


I think I can assume everyone reading this is already aware there will likely be an early general election on the 8th June. There will be lots of polling ahead, but here are a few initial thoughts:

The overall polling position is a strong lead for the Conservative party. There is some variation between different polling companies, but all the polls are showing robust leads for the Conservative party, most are showing extremely strong leads up in the high teens, a few breaking twenty. As the polls currently stand (and, obviously, there are seven weeks to go) a Conservative majority looks very, very likely. The size of it is a different matter – the twenty-one point lead in the recent YouGov, ICM and ComRes polls would produce a majority in excess of a hundred, a nine point lead like in the Opinium poll at the weekend would only see a small increase in the Tory majority.

It’s harder to tell from the polls how well the Liberal Democrats will do. The swing between Labour and the Conservatives will normally give us a relatively good guide to the outcome between those two parties. The Lib Dems are a trickier question – the polls generally show them increasing their support, and this has been more than backed up by local by-elections. How it translates into seats is a more difficult question, my guess is that their support will be concentrated more in areas that voted Remain and the Lib Dems have a history of very effective constituency campaigning. I would expect them to do better in terms of seats than raw swing calculations would suggest.

The elections will be the test of to what degree pollsters have corrected the problems of 2015. The BPC inquiry into what went wrong at the general election concluded that the main problem was with sampling. Polling companies have reacted to that in different ways – some have adopted new quotas or weighting mechanisms to try and ensure their polls have the correct proportions of non-graduates and people who are have little interest in politics; others have instead concentrated on turnout models, moving their turnout models to ones based upon respondents’ age and social class, rather than just how likely they say there are to vote; some have switched from telephone to online (and some have done all of these!). The election will be a chance to see whether these changes have been enough to stop the historical overestimation of Labour support, or indeed whether they’ve gone too far and resulted in a pro-Tory skew. I’ll look in more detail at the different methodological approaches during the campaign.

Elections that look set to produce a landslide results may bring their own problems – in 1983 and 1997 (both elections that mostly relied on face to face polling, so not necessarily relevant to today’s polling methodologies) we saw polls that largely overstated the victorious party’s lead.

The local elections will still happen part way through the campaign. The local elections will still go ahead at the start of May. It’s been a long time since that happened – in recent decades general elections have normally been held on the same day as the local elections – but it’s not unprecedented. In 1983 and 1987 the local elections were in May and the general elections followed in June. Notably they were really NOT a good predictor of the general election a month later. Comparing the Rallings & Thrasher estimates for the local elections those years with the subsequent general elections, in the 1983 local elections the Conservatives were ahead by 3 points… they won the general election the next month by 14 points. In the 1987 local elections the Conservatives were ahead by 6 points, in the general election a month later they were ahead by 11 points. In both cases the SDP-Liberal Alliance did much better in the locals than the general a month later. In short, when the local elections happen in May and Labour aren’t 20 points behind don’t get all excited/distraught about the polls being wrong… people just vote differently in local elections. It may well give the Lib Dems a nice boost during the campaign though.

The Fixed Term Parliament Act probably ended up a bit of a damp squib. There is still a vote to be had tomorrow and the government still need two-thirds of all MPs so it’s not quite all tied up, but as things stand it appears to have been much less of an obstacle than many expected. There was no need for a constructive vote of no confidence, the opposition just agreed to the election. The problem with the two-thirds provision was always the question of whether it would be politically possible for an opposition to say no to a general election.

The boundary changes obviously won’t go ahead in time for the general election. However, it does not mean they won’t happen. The legislation governing the boundary reviews doesn’t say they happen each Parliament, but that they happen each five years. Hence unless the government change the rules to bring them back into line with the election cycle the review will continue to happen, will still report in 2018, but will now first be used in the next general election in 2022. If the election results in an increased Tory majority it probably makes it more likely that the boundary changes will go ahead – getting changes through Parliament always looked slightly dodgy with a small majority.


-->

Two new ICM polls

I’ll post some thoughts about the general election later, but first a brief note on the two ICM polls today (yes, two seperate polls). Earlier today we had the usual regular ICM poll – Martin Boon had said he was disclined to publish because the fieldwork was done over the Easter weekend, and we pollsters tend to be a little wary of fieldwork over bank holiday weekends. Calling a general election made the result rather more relevant though, so out it came – topline figures were CON 44%, LAB 26%, LDEM 10%, UKIP 11%.

Later on in the day we got an ICM “flash poll” conducted after the election announcement, with fieldwork in space of four hours. That had voting intentions of CON 46%, LAB 25%, LDEM 11%, UKIP 8%, GRN 4%. The poll also asked about the decision to call an election, finding 55% supported an early election, 15% were against. Full tabs for that are here.

A note about the methodology, the flash poll appears to be a proper ICM poll using all their normal methods and normal weights… just conducted very quickly. Whether that makes any difference is unclear (people who reply quickly or reply during the daytime may or may not be different in ways that may or may not be adequated corrected by weighting) – notably the UKIP score is the lowest we’ve seen from ICM. It remains to be seen if that’s reflected in future polls, is a result of the methodology, or is just normal random variation.


Just to follow up on the voting intention polls yesterday, there was also a new YouGov poll in this morning’s Times. Topline figures were CON 44%, LAB 23%, LDEM 12%, UKIP 10%. The twenty-one point lead is the same as the weekend’s ComRes poll and the largest YouGov have given the Tories in government (it also equals the highest the Lib Dems have hit since the election).

Full tabs are here.


Sunday Polls

There are several polls in today’s papers. ComRes in the Indy and Sunday Mirror have topline figures of CON 46%(+4), LAB 25%(nc), LDEM 11%(-1), UKIP 9%(-1), GRN 4%(nc). The twenty-one point Conservative lead is the largest anyone has shown for them so far this Parliament (and, hence, the largest since when they were in opposition).

ComRes also did a split sample experiment, asking about some Labour policies. Half the respondents had the policies described as “Jeremy Corbyn policies”, half had them described as “Labour Party policies”. This turned out to make no difference whatsoever, suggesting that association with Jeremy Corbyn is no worse than association with the Labour party… though that could easily be just because the two are now so closely linked. On a broader point, the policies that ComRes asked about all remain popular – 71% support increasing the minimum wage, 62% support increasing the top rate of income tax, 53% support free school meals paid for through VAT on private schools. The importance of shallow approve/disapprove ratings of individual policies on party support are often grossly overstated… but it is worth noting that Labour’s evident problems do not appear to be caused by proposing unpopular policies. Full tabs are here

There was also a voting intention poll from Opinium in the Observer. Topline figures there are CON 38%(-3), LAB 29%(+1), LDEM 7%(-1), UKIP 14%(+1). Tabs for that are here.

The big gap here between ComRes and Opinium will be largely down to methodology. Following the 2015 polling error ComRes switched to a turnout model based upon demographics rather than how likely people say they are to vote. Essentially this downweights younger and poorer respondents on the basis they have historically been less likely to vote. Typically this has produced larger Conservative leads compared to other companies.

In contrast Opinium produce topline figures that consistently show some of the smallest Conservative leads. Crucially they are one of the only companies that don’t weight by past vote (instead weighting by a version of party ID). Looking at the recalled vote in today’s poll as many people claim to have voted Labour in the 2015 election as claim to have voted Conservative, suggesting the poll may well have a sample that’s a bit too Labour.

As ever, if you are trying to work out what the actual state of party support is you should avoid cherry picking the polls you’d like to be true. It’s all too easy to find reasons to convince yourself that the poll showing the results you’d like is the poll that must be the most accurate one. A sensible rule of thumb – especially at this stage of the Parliament – is probably just to follow the broad average of the polls, which suggest a Conservative lead somewhere in the mid-teens.

There were two other polls in the Sunday papers. An ORB poll in the Sunday Telegraph asked about Brexit – 55% thought Brexit should go ahead, 45% did not. 55% also approved of the way Brexit negotiations were going head, 45% disapproved. UPDATE: The reason the two figures were the same is that there was only one question – the Sunday Telegraph just reported it incredibly badly. 55% approve of the way the government are handling negotiations, ORB didn’t ask if people supported Brexit.

Finally the Western Mail had a Welsh poll by Beaufort. From their report there only appears to have been one question: 39% thought the Jerfemy Corbyn should resign, 32% think he should stay.