A few quick comments on that Hartlepool poll. You may recall that last time the past vote was very odd - despite Brexit party getting 25% in 2019, Survation could barely find any former BP voters.
That is NOT the case this time - the sample looks fine on that front.
(I'm guessing that means Damian did what he suggested he might, and prompted on the recall Q)
Secondly, any phone poll like this you *still* get comments along the lines of "well, if you only ring landlines!"
No phone polls only ring landlines. It rang mobiles too.
The poll may be wrong, but it certainly won't be wrong for either of those reasons.
I've also seen people question the 300 sample size. That's the number of people who actually gave a VI once you take away the don't knows and won't votes. The actual sample size was about 500, which is typical for a constituency poll.
Constituency polls are, as I've said before, incredibly hard to do, and have a somewhat patchy record. If you look over constituency polls from 2019 you'll find some that were bang on, and some that really did fall short.
But there's no reason to doubt this one in particular (and given the scale of the Tory lead, it would have to be very wrong for Labour to win)
• • •
Missing some Tweet in this thread? You can try to
force a refresh
I mentioned this voodoo poll last week when it was being touted about social media. It looks like the Mail on Sunday was the only paper to fall for it. Well done to all those other newspapers who did not. It's tosh of course. I'll explain why below (1/20)
I should add that the poll won't "send shockwaves through Downing Street", as people working in Downing Street will know the difference between a properly conducted poll and a self-selecting propaganda exercise. I do wish journalists did (2/20)
Let's start at the start. The poll was not conducted by a proper independent market research company that paid attention to using unbiased wording and ensuring a representative sample. It was conducted by a pressure group, sent to its own mailing list and on social media (3/20)
This story in the Times surprised me - a claim that support for nationalising water had fallen to 27% thetimes.co.uk/article/suppor…
Nationalisation tends to be fairly popular. A genuine drop from 83% to 27% would be very surprising indeed so, it's fair to say I was a tad suspicious.
Let's start with the 2017 result. That was from this Populus poll for the Legartum Institute. It may not be the greatest of questions (it really should have a don't know!), but it's straightforward and not leading populus.co.uk/poll/public-op…
A longish thread about trying to explain the differences in party support - primarily Labour - in the polls.
It is often very difficult to get to the bottom of the differences between polls. There is rarely one big, easy simple answer, it's normally a bit of lots of little, different to explain things. Hence stupid & simple explanations prosper.
One current cause seems to be different approaches to past vote weighting. Most pollsters* weight by how people voted at the last election (ensuring that the sample has 44% people who voted Conservative in 2017, 41% Labour in 2017 and so on)
Rule 1 in interpreting public opinion should be too look at the polling in the round. Public opinion can be nuanced and complicated and taking one single poll finding that appears to back up your preconceptions and ignoring the wider picture can be deeply misleading (1/...)
So, given there's been some of that sort of rubbish written about polling on No Deal today here's a quick thread summarising what YouGov's recent polls on No Deal have actually said - giving both sides of the story
First - a No Deal Brexit (or "Leaving the European Union without any deal", which is the wording we use to avoid ambiguity) is seen as a bad outcome. Only 25% think it would be good for Britain, 50% bad.
The Mirror, the i and lots of local papers are today all running stories about their "big Brexit survey" which purports to show a majority of people want to remain in the single market. (1/...)
It claims relevance due to the unusual large sample size - 200,000+. As you'll hopefully know, a big sample size alone does not make a poll accurate or meaningful. What determines whether a poll is worth paying attention to or not is whether it is *representative* of the public
The classic illustration of this is the Literary Digest poll - a magazine that used to do polls of literally millions of Americans at Presidential elections, and got it catastrophically wrong in 1936 when George Gallup's representative poll got it right.