@robertrea
@aedwardslevy
More than anything, I think pollsters have a duty to be public explainers of what polls can and *cannot* do, how to interpret them sensibly, how to do polling well.
We should never let ourselves just amplify the simplified, sensationist findings that some want polls to show.
Unbelievably this headline is based on this: a random twitter poll.
Beyond belief that a journalist, and an editor and whoever the hell else signs off on newspaper stories thought... just...well... I'm lost for words...
Third - No Deal is only the preferred choice of 26% of people. 11% would prefer the govt deal, 12% would prefer a customs union/single market deal, 37% would like another ref and to stay after all.
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)
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/...)
I see a lot of people have twigged and already flagged this, but just for the avoidance of doubt. The MRP in the Telegraph today is YouGov's, but the claim that it's all because of Reform and it would be a hung Parliament without them is the Telegraph's own claim.
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.
Links to all the results are here (because the other rule is you should go and look at the actual tables, not just believe stuff people stick out on twitter!)
@durrant1959
@IoWBobSeely
The Titles Deprivation Act that Bob wants to amend removed princely titles as well (it removed "any dignity or title") - the three Royal Dukes who had their titles removed by it lost their status as Princes of the UK as well.
Over the next few days there will be lots and lots of polling about potential Conservative leaders. A few things to note when you are looking at them. 🧵
The reason is that the ComRes poll was *actually* Deal 30%, No Deal 20%, Remain 42%, and the two leave options are added up.
This is far less surprising, when polls split out different options like they they do often find the Leave options sum to more than the Remain ones.
For the record, the ComRes poll doing the rounds today is presented as Leave 50%, Remain 42%.
This would be a very unusual result, as almost every poll that's asked a straight Remain/Leave question since mid-2017 has round Remain ahead. See the links here
@Samfr
Grammar schools - pro to anti - from being a parent of kids going through it (jnr passed, but it's inhumane and only serves to allow the middle class to protect their own privledge)
Ha ha. One of those candidates who was standing in two places 160 miles apart (presumably to do his bit as a paper candidate) has accidentally been elected in both. Good luck with attending council meetings...
They've changed the headline now to "Only 8% of schools in England have received air monitors, headteacher’s Twitter poll suggests"
This is still terrible, since it's a *Twitter poll*. It is not a representative sample of schools, it is a sample of arbitrary people on twitter.
...but that took 17 tweets to explain, when "Pollster A is biased and run by lizards, I only listen to Pollster B who is the most accurate" takes just one.
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
A common rebuttal to polls showing more people would now support Remain than Leave is that polls before the EU referendum also showed more people would vote Remain than Leave. A few thoughts on that below...
People are still being weird about electoral impact of VAT on school fees.
27% of adults have a dependent child at or below school age (obviously those <4 aren't at school, but reasonable to assume those who may send a child to school in the near future would be impacted)🧵
Here's that piece I wrote a few months back explaining why agree-disagree statements are a bloody awful, biased way of asking questions.
You know, just in case.
Always depressing when you refuse to run biased questions for a client, they disappear, and then a few weeks later you see another company did lower itself enough to run them. Sigh.
Again, a plea to journalists - do not take claims from press releases about what polls show as necessarily valid, ask to see the actual questions asked to check you aren't being bullshitted.
It hasn't halved. It isn't comparing like-to-like (because there are don't knows now, but you can't vote for "don't know").
The *share of the vote* has gone from 86% to 60%.
(You could look at % of Muslims supporting Lab, but then you'd need to include did not vote).
On other matters.
It's only a majority of the public who think something if it's more than 50%.
You can't just pretend it's a majority by leaving out the don't knows, those people still exist.
However, if push came to shove, if the deadline was upon us and the EU would not grant any more extensions, in that forced choice situation 44% would prefer No Deal, 42% would prefer to stay. Though of course, that's asking about a hypothetical situation we aren't in. Yet.
That means polls of the general public asking who their favoured candidate is are of very little worth - it's 90% just name recognition. Rishi will do well compared to others because people have heard of him - they won't say Tugendhat because they've never even heard of him.
@IanDunt
People didn't believe it in 97
Below are the predictions Reuters gathered just before election day. Everybody underestimated it. Even David Butler (who is the anonymous one at the top) was 26 short.
And these were all proper experts in the field, not wishful thinking partisans
@jimwaterson
I bet she's worked her socks off to build that business up, and she's going to bankrupt herself and lose it all because she believed what a crackpot on the internet said.
The latest Eurobarometer survey figures for UK - Remain 45%, Leave 37%.
Or, the Telegraph reports it "Less than half of Britons would back staying in EU, poll suggests"
"Premium"
Bottom line is that there are lots of former Tories voting Reform because they are unhappy with the Tories & feel let down by them. If Reform vanished, it wouldn't magically stop those voters being unhappy and disappointed in the government.
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)
Just saw *another* leaflet from a politician claiming one of those "Flavible" seat projections based on national polls is a YouGov poll.
It's not. We don't do individual constituency polls. Anything claiming to be a YouGov poll of a single constituency isn't one.
@d_angeletti
@AdamBienkov
Ha ha. As Adam has correctly relayed, we haven't done a constituency poll of Putney. It's a projection from a website called Flavible, based on one of our GB polls.
The projection is Flavible's, it shouldn't be represented as YouGov's.
Second - while people don't like No Deal, they do not believe some of the more extreme warnings of doom & disaster. Only 35% think warnings of food & medicine shortages are realistic, 47% think they are invented or exaggerated
As far I can tell, it's the Telegraph running the sums on what you'd get if you add the Conservative & Reform party votes together, which isn't a very good way of measuring their impact.
People keep doing polls asking how people would vote if Britain didn't leave on the 31st, if Britain leaves on the 31st with a deal, etc, etc.
Everytime I see one a little bit of me dies.
@estwebber
In fairness, Labour had a contest where the leading candidates had the same Oxford College, same secondary school, same primary school and came from the same womb. That's going to take some beating.
First - most of the people in the running are little known to the public. It's very easy to overstate how much knowledge the public have about most senior politicians - people know about the leaders, usually the Chancellor, and *very few others*.
The Daily Mail's reporting of their Survation poll is comically one-sided:
Online headline is "Most voters think the Queen was RIGHT to approve his request to suspend Parliament in his drive to deliver Brexit", implying that the public support the move.
(An alternative explanation is that it's all a huge conspiracy and Kantar are run by lizards....
...or you could actually look at their admirably detailed methodological statement and try to understand the methodological differences between polls)
This is a very good article by
@StuartJRitchie
. If BBC verify want to set themselves up as factcheckers, they need to be better than this. They need to factcheck themselves.
Is it really true--as reported yesterday--that a quarter of the UK population believes that COVID was a "hoax"? I'm sceptical.
I also criticise the way BBC Verify talked about the poll results:
Now, about Britain's appetite for rule-breaking leaders...
"Britain needs a strong leader who is willing to break the rules"
45% agree, 28% disagree.
BUT
"Britain needs a strong leader who plays by the rules"
65% agree, 14% disagree.
Results here:
If the answers are
"A pressure group did it themselves"
"Their own supporters, and activists on social media"
"They just sent it out into the world, and let it get passed about"
"A pressure group"
"Ones a pressure group wrote themselves"
Then maybe ask if it's worth reporting?
I wonder how many candidates would have realised this was potentially illegal. Certainly there have been people in the past who've openly done it. Clement Freud famously won a £1000 bet on himself to win at 33-1.
The question the claim that only 2% of cyclists think they should have to adhere to the laws of the road is based on this beauty of a question - the only way to say yes was to endorse a road tax for cyclists (12/20)
My old boss Peter Kellner used to have a good rule of thumb on neutral questionnaire design. Ideally a respondent shouldn't be able to tell from your question wording which side of the argument the client is on.
If the Reform party disappeared tomorrow, then it's likely some of their voters would go to the Tories. But some of their voters would also go to UKIP & splinter parties, some to Labour & other parties, and some would stay at home.
Here's my attempt at an explainer covering some of the rather too common responses to polling on social media.
Includes such greatest hits as "I wasn't asked", "My Twitter poll is bigger" and, of course, "That pollster is run by lizards"
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.
Oh joy, everyone is looking a poll with a one-sided agree/disagree statement with a counter-intuitive result and drawing the conclusion that public opinion is different from what all other polls show, rather than the conclusion that it's a shitty way to ask questions. Sigh.
@yuanyi_z
If they thought that long stuck convoy of Russian armour outside Kyiv was bad, wait till they try getting the Russian army down the A303 on a bank holiday weekend
@sundersays
I hadn't realised quite how insane that column is.
He's judging how accurate the polls were by comparing them not just to the actual results, but to the fantasy figures in his head.
The steady Labour leads around the 20 point mark for the last few months mean people have stopped getting foolishly over-excited about outliers. Then today we've got a Deltapoll poll with a 10 point lead, the dam has broken and stupid has started spraying everywhere. Sigh.
So what we've got here is a poll that actually showed more people *support* nationalisation of water, unless the question is pre-empted by leading information, reported as if it showed people were opposed.
As part of my ongoing mission against agree/disagree statements...
ComRes at the weekend. 62% agree with the statement "Britain needs a strong leader willing to break the rules"
Deltapoll last week. Should leaders bend the rules to get things done? 24% should, 52% should not
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)
Just as a reminder... you can't get a representative survey through a pressure group surveying its own supporters.
People who join an organisation or hang about on their website will *inevitably* be more likely to agree with that organisations aims and beliefs.
There's already some damn fool poll asking a "does this make you more or less likely to vote Tory".
Just ignore these. It's terrible survey design. People tend to use them to signify support or opposition to a policy regardless of whether it would actually change their vote.
@HugoGye
People who set up new parties fascinate me (especially people who aren't obviously deluded or dim, and have the money to hire people smart enough to tell them its doomed to fail).
What do they actually *think* they can achieve from it?
How have vox pops survived the coronavirus? If there was one thing that it should have put an end too. Nope. They're indestructible.
If the world ended tomorrow TV news would still find a convenience sampled cockroach on the street to get ad hoc reckons from.
Party members polls should be a little better. One assumes people who join political parties are more likely to know about senior politicians - and of course, these are the people who will have the vote on the final two candidates, so they should have predictive value on that.
@JamesCo82102107
@alqabadai
@tnewtondunn
@ShippersUnbound
Sometimes they even swap pens half way through their signature (or through a single letter of their signature), if they want to be able to gift pens to a lot of people for a piece of legislation. LBJ managed to use 75 pens to sign the civil rights act.
Predicting a *by-election* through an MRP is, um, brave. The point of an MRP is to make predictions at low geographic areas. You don't have enough people in that area, so you effectively model it by looking at the responses of similar people in similar areas.
@twlldun
I mean, I was doubtful there were tweets I would actually pay money to read.
Then came the "dunk on twlldun for how filthy the top of his washing machine is" thread...
Don't read TOO much into those "How would you vote if X was leader" Qs. They look good, but are not particularly good predictors of how people would actually vote. We're asking them a hypothetical Q about leaders they don't know, running on policies & platforms they don't know
In fact Survation specifically asked if people supported or opposed the prorogation, and they marginally disapproved, by 40% to 39%...
but that is only mentioned down in paragraph 21(!)
@twlldun
Tories are tanking because they're losing support towards clear pro-Brexit parties. Brexit party are cannibalising UKIP support because low-info voters have just noticed Farage isn't UKIP anymore. Labour not moving much. Anything else prob noise & over-interpretation.
Job done
Showing my age here, but can't see this Capitol protestors picture without thinking of the old Civ1 diplomacy screen.
Strong "My words are backed with NUCLEAR WEAPONS" energy.
Asking if people have positive/negative perceptions of each leader can be a bit better - but are best asked with a "I don't know about this person" option, to show if people actually are neutral or just haven't heard of that person!
What does that matter? A poll is only meaningful *if it is representative of the population it is trying to represent*. That is, if the population is 51% female and 49% male, the sample should be 51% female and 49% male, and so on (4/20)
@PickardJE
(Should add, his brother was 1st Lord of the Admiralty. So it wasn't that weird. It wasn't just randomly spaffing honours at his close relatives.)
@twlldun
My nan was alive during the Spanish Flu pandemic. According to her every family on her street got it apart from hers *because they hung an onion in their window*.
If you can't bring the same vote twice in the same session, one solution is to alter the nature of the vote. The other is to change the Parliamentary session.
Some people might want to go and have a look at the Parliamentary Session of Sept 1948. All 10 days of it...
So there is some cause to think this lead may fall back a bit. But it's 33 points. Even if it halves, that would STILL be showing a crushing Labour victory. There's no denying the Conservatives are in severe trouble.