Categories
Thoughts on the polling debate from Nick Sparrow – formerly of ICM – politicalbetting.com
Thoughts on the polling debate from Nick Sparrow – formerly of ICM – politicalbetting.com
Writing in the Guardian, Peter Kellner asserts that “it is common knowledge that final election polls are sometimes tweaked”. This should not come as a particular surprise. For a few days before an election, the one place neither a pollster nor the client wish to be is out of line with all the others, whether that means all the others have got it right or wrong.
0
This discussion has been closed.
Comments
Sure. It picked up the drift towards Labour during the campaign. But nowhere near the extent of it.
Plenty more were shocked to the core by '17 than '15.
He said, basically, that YouGov showed a smaller Tory lead than other pollsters, supported by the MRP, and YouGov were concerned that they would lose business if they were wrong when everyone else was right, so they found reasons to nudge it in the Tories' favour while sticking with a defensible position.
That doesn't seem hard to believe. Of course things have got a bit sidetracked by the Zahawi, which was a joke, but it was only funny because of the obvious truth of the above.
if we go back to the PB threads, I'm sure that's what we'll find - note of the changes and discussion of whether this was a sop or not.
Con: 41
Lab: 40
Keir Starmer and Wallace tied, everyone else way way down.
So it's boring vs boring then
Heatwave next weekend. Get yourself bikini ready PB and off to the beach!
https://www.yr.no/nb/værvarsel/daglig-tabell/2-6296599/Storbritannia/England/Stor-London/London City lufthavn
Majority do not expect him to lead the Conservatives into next General Election.
The public are equally sceptical about the Government’s ability to deliver on a range of issues. Around two-thirds say the Government are unlikely to reduce the cost of living (69%) or reduce crime (66%). Despite the increase in National Insurance to provide more funds for the NHS and social care, 63% say the Government are unlikely to improve the NHS.
What they cannot do is really compensate for a vote that actually matters as opposed to the answer to a hypothetical question. So, right now, huge numbers of natural Tories are disgusted by having a liar for a PM and the fall in public standards. Hypothetically, many or most will want to vote for someone else and in irrelevant bye elections they may well do so. But choosing a government is a different matter. At that point many, not all by any means, will swallow that irritation or disgust and vote for them.
It is a measure of the skills of pollsters, built on many, many past errors that they get as close as they do but I can fully understand the temptation to tweak the raw data towards the government of the day, whether that is Conservative or Labour. Their best guess at general elections is where the validity of their prognostications the rest of the time are tested, hence the "gold standard" that ICM had for many years, now probably held by Yougov. This matters to their business and could seriously undermine the confidence of clients asking more mundane questions. Given their uncertainties there is no guarantee that the raw data will give a better result.
Wallace does better too, indeed I'd support him for PM
Annoyed at myself for not backing him to top practice, but there we are.
https://www.washingtonpost.com/business/norway-ends-contract-for-nh90-helicopters-wants-full-refund/2022/06/10/aaaf0432-e89d-11ec-a422-11bbb91db30b_story.html
Norway terminated its two-decade-old contract with a France-based manufacturer for 14 maritime helicopters, citing delays, errors and time-consuming maintenance, the defense minister said Friday, calling the move “a serious decision.”
The Norwegian government will return the NH90 helicopters it has received so far and expects a full refund of the nearly 5 billion kroner ($525 million) it paid, according to Defense Minister Bjørn Arild Gram.
So yes Chris' retraction and Zahawi's explanation give more credence to his original tweet in my mind than if it had just been left.
His retraction is about as credible as a Ukranian or Russian PoW saying how everything was a dreadful mistake once captured on the other side's TV.
14 polls in the final week. 3 had a 10%+ Tory lead.
1 with Labour in front. Only 4 within MofE of the actual results.
The average of these was over 7% lead.
I call that a polling fail. Even if the clues to a possible huge fail were there, most chose to dismiss them completely.
The varying numbers around them being just MoE.
(I believe that pretty much every lead - from 3 pts for Labour to 11 pts for Labour) have had the overall scores within MoE of those numbers.
Everyone chose to ignore it. Indeed. The common reaction on here was that he'd had his account hacked.
If true it does show that polling companies may not be independent of the political process. I mean Trafalger has always seemed an outlier in the US.
The main alternative, and favourite for the moment, is Hunt - comparatively a focus on pure professionalism. I'm not sure how far that would go, or how long it would necessarily last, though.
"The PM is rumoured to be planning a walkabout in Honiton today but not (sic) media invites have been issued."
Unsure as to what the point of that is. Doubtless a gaggle of Tory activists gathered to look like a throng of excited, grateful locals.
- If they float, they are witches. Burn them.
- if they sink, they are innocent. Shame really.
You are entitled to have a negative view of our military establishment and I'm sure have your own reasons for that. But it strikes me as rather sad if you allow that to colour your perception of everything else.
If this did happen, it was an own goal.
If folk had known Corbyn had clearly won the debate, and the media had therefore picked up on his gaining traction, it may have resulted in fewer Tories sitting on their hands.
And fewer Tory incumbents on their arses.
So. It is at least arguable the non-appearance of this poll contributed to the loss of majority.
GWBWI for 9/6/22
LDm +65
Lab +43
Grn +1
Con -143
This is the worst weekly Con score since the May council elections. Sevenoaks on its own would have outscored every week so far.
Adjusted Seat Value
LDm +1.1
Lab +0.7
Grn +0.0
Con -2.4
I know this is all speculation. I've no idea how the world looks as a reluctant Tory.
YouGov had to take legal action, since their reputation was being directly questioned.
I'm merely pointing out that it may not have been such a cunning plan.
In general therefore it cannot be verified either then or later, except by other polls which suffer from the same problem.
They suffer from the same problem as empirical verification: this can only happen through further empirical verification, which can't tell you if the process itself contains a flaw.
Herd instinct is bound to creep in as elections approach, as it is more likely that the outlier is wrong than that the herd is wrong, and if you are all wrong that's, of course, OK.
The big exception is John Curtice's big poll on election day itself, which is, I think based on claimed actual voting behaviour, and has proved remarkably robust, even when surprising.
For this alone he deserves his knighthood. A seat in the HoL would be even better.
Maybe they should do both, be open about both like how weather forecasters give us alternate forcasts these days that are very different - our latest unadulterated sub sample said this, but a chat over coffee pooling our psephological minds reckons it nearer this: so there’s the raw data and separately here’s our thoughts.
I’m serious, it works with weather forcasts “if it stops about here we will get this, if it pushes up to Iceland we will get that instead” you can have more confidence when it’s put like that?
Did I mention it’s a heatwave end of next week?
Nb. I don’t know how seriously to take the Zehawri phone call thing. As soon as PM didn’t turn up to the big leadership debate she promised to, the whole point of the debate poll no longer existed, it was never going to measure what it was set up to measure. Anything released saying Corbyn won would have been ignored as mostly irrelevant anyway as it wasn’t a win over the prime minister. The fact she hid in a fridge eclipsed any damage an honest post debate poll would have inflicted.
Hunt/Mordaunt/Wallace need to calm the Blue wall without frightening the RedWall. It's a tough balance to achieve, but at present Johnson has pissed off both sets of voters.
If he could defend what he said, he'd not have backed down, the fact that he has so rapidly and unequivocally suggests he knows full well that he can't.
He'd held those jobs for very little time, relatively, and a lot of people's response was : "Who ?"
And polls try to make up for bad sampling with heavy weighting.
I'd like to see the raw result of a simple one question poll with a huge response, perhaps by tacking it on to a popular web site such as the BBC or Google. And even that would automatically exclude voters with no web access.
Damned if you do, damned if you don't it seems.
“A con win on Redwall council! Starmer in MELTDOWN” but it’s the seat on Redwall council that’s been blue for most 100 years, the one if lost points to a 97 meltdown.
We have a poster here called HY who set Labour a Bar of winning Swindon Council if they hope to win the general election. Win six from nine seats that bar sounds reasonable - but last year Tory’s picked up seats they have never held before, that’s party of their position this time, it made the true picture of what Labour needed to do this time that much harder set in past history of seats and council, in this way the bar was actually higher for Labour in this years locals, hence the poor headlines they got didn’t truly represent what was going on for them this year, particularly in this places it’s only a third of the council up.
Those headlines of Labour underperforming probably saved big dog and allowed him to stay until the next election, but it was based on spin and misconception really because it was not based on the index widgets you have.
Shitty business all round.
It's ironic that the rail strike is supposed to damage the government if it does in fact prevent LD activists from travelling to Tiverton and therefore increases the chances of a Tory hold.
I will aim to post the scores each week from now on, and give a summary with graphs and everything every now and then.
You could argue we are already in a new cold war with China. But with Russia it very much depends on the outcome of the war. A quick humiliation for Putin means things could be nipped in the bud. The 90s after all saw a peace dividend.
And it is not just voodoo polls that suffer from self-selection. It is not as if people can be compelled to answer polls. Phone polls (and face to face polls although I think these are no longer used) suffer from response bias, and panels are made up of volunteers, often political activists and even PBers!
We need to check coming front pages for little negative stories about Mourdant appearing, as that’s how it started when Number Ten murdered Sunak’s career.
I've thought about this a bit and, besides from making responding to opinion polls mandatory, I can't think of any way round it. You don't know how the small minority who choose to respond to opinion polls differ from the large majority who are too busy, pissed off, paranoid or otherwise not disposed to respond.
Like DavidL said earlier. I'm surprised they do as well as they do.
We are spoilt on this site, all different people bringing different dishes to the party table, making this the must be haunt for psephology.
Being on PB is what God’s Beard must feel like when it’s getting stroked.
I actually think the MoD will take the opposite lesson from the Ukraine-Russian festivities. They won't gear up for a return to Cold War levels of defence spending but conclude that the Russians are crap so let's get rid of a load of heavy armour and artillery.
https://en.wikipedia.org/wiki/List_of_prime_ministers_of_the_United_Kingdom_by_length_of_tenure
Depp's team had to prove malice against Heard in court, it's a much much tougher job than in the UK courts. That's the legal fact - that the US jury came to conclude he'd gone over that very very high bar and the UK judge reckoned he'd failed to clear the much lower one is simply evidence that the US jury was wildly more pro-Depp than the UK judge (Or the UK judge much more pro Heard). The US bar for libel is way higher than the UK.
https://publications.parliament.uk/pa/cm/cmregmem/contents2122.htm
By 1995 however Redwood ran against Major as the true Thatcher heir as would Portillo had he not bottled it
How quaint. The Thatcherites are long gone now. No Thatcherite style Conservative within a sniff of a chance of becoming PM.
https://twitter.com/sajidjavid/status/1535248253222109184?s=20&t=ONYSpGQ6MEH8LbdYHozHSQ
*which, on reflection, was perhaps not something Mrs T would do, at least for herself personally as opposed to No 10 as a whole, let alone a cabinet post