Written By:
te reo putake - Date published:
6:46 pm, July 28th, 2020 - 40 comments
Categories: act, David Farrar, Dirty Politics, election 2020, greens, labour, Politics, same old national, you couldn't make this shit up -
Tags: judith collins, national party, political polls
Radio NZ’s Jo Moir has tweeted the polling results be-grudgingly given to the National Party caucus this afternoon after significant rancour in the ranks. The results were presented by David Farrar, the Tory’s pollster, ironically meaning that the caucus room briefly contained both the Joker and the Penguin.
Holy return from oblivion, Batman!
The results are:
Labour 49
National 36
Greens 6
NZ First 3
ACT 3
These are the figures Gerry Brownlee reckoned were cause for Tory optimism!
Oh, well, Gerry, when it all turns to custard you can always put those dormant woodworking skills to use making park benches.
A huge gap to Labour, an even bigger swing required to get to a winning majority, and starting on a base figure that suggests there’ll only be 42 Tory MP’s left in Parliament is actually a portent of a coming thrashing, not a reason to be cheerful.
And, of course, the results were leaked.
Strong team, National, strong team!
The current rise of populism challenges the way we think about people’s relationship to the economy.We seem to be entering an era of populism, in which leadership in a democracy is based on preferences of the population which do not seem entirely rational nor serving their longer interests. ...
The server will be getting hardware changes this evening starting at 10pm NZDT.
The site will be off line for some hours.
I suspect the penguin overegged his counter-rogue…
Line of the week and it's only Tuesday 😆
That's Labour 62, Greens 8. I'd like the Greens to have a greater share, but overall … 70-50? Where do I sign?
In fact National are doing Labour a favour, though they don't seem to realize it.
If the swing voters who like Ardern but are not "left" believe Labour are at 60+, they may be less likely to vote Labour. They will feel they don't need to.
If Labour voters believe Labour are at 60+, they may be less likely to vote at all.
If Left voters believe Labour are at 60+, they may be more likely to vote Green.
The paradox is that Labour don't want a series of "game over" polls that make it less likely that those polls will turn into actual votes. The prospect of Collins ousting Ardern is the great motivator. I'm only worried nobody will believe it in September.
Yes, I agree, I like these numbers even better than the "rogue" poll. The GP will get to make a few more serious demands, and Labour can keep chasing that centre.
Tova O'Brien
https://www.newshub.co.nz/home/politics/2020/07/from-scary-to-tough-and-aggressive-kiwi-voters-describe-national-leader-judith-collins-in-one-word.html
RNZ reported that David Farrar presented two sets of figures to the National Caucus taken over different time periods …
… so presumably the earlier (pre-Falloon ?) poll was Lab 47% / Nat 37% … and the latest sampling (possibly covering Falloon / ILG fallout period… but we don't know for sure) is Lab 49% / Nat 36%.
Still has me wondering why Tova is implying Labour’s on 47% in latest Curia rather than 49% ? (Jo Moir had tweeted Labour just short of 50
Can the disaster that is the National Party be laid at the feet of Peter Goodfellow?
He seems to have been the one behind the scenes ushering them to this desperate place.
Polling methodology ought to get more media focus. I don't believe the `rogue' thesis satisfactorily explains the different margins between National & Labour (37/13). These polls are supposed to have an error of plus or minus 3%.
Despite passing exams in statistics at university it always struck me as hokum. Stats theory says that wacky results do happen (rarely). You know, wait long enough & a gaggle of monkeys can eventually type the Encyclopedia Brittannica. That was the imagery used back then. Who really believes that shit?? Only stats professors, I suspect.
More likely, the methodology used is spinning different illusions. Different polling companies are free to choose their own method. Artistry – not science! So we get different publics painted in each picture.
They also usually have a confidence interval of 95% i.e. 1 in 20 are "rogue" polls and will be outside the normal margin of error.
Those professors, eh – just more waffling ‘experts’, in praxis.
The election results are real, so use them to test the science. Both TV1 and TV3 were very close to the 2017 results, in their final polls.
Until then, folks will gauge accuracy in comparison with the next CB poll. If that does show up Thursday evening as some here suggested, the sampling period may be close enough to RR's for a reasonable comparison. So if CB confirms RR, the Nats ought to tell Farrar his methods aren't good enough, eh?
agree denis. this poll seems questionable for the fact that there are no- dont knows. in any poll, on any subject, they are always some people that cannot make their minds up.where are they in this poll?
It's normal for poll reporting in NZ to exclude the undecided count.
Usually the undecided bobs along around 5-7%, is only when it ticks upwards to a large degree that it is particularly noteworthy.
Your understanding of statistical models is letting you down here. The margin of error of 3% means if you took another of the same polls what is the chance it comes out the same. It does not tell you however that the poll tells you the election outcome within 3% as elections and polls are different survey methods. It also does not tell you that the corrections you made to your model have brought your answer closer to (rather than further from) predicting the election outcome. It also doesn't tell you, you received answers about the election.
In related news the pie charts MW sometimes generates from poll results are *not* showing valid election outcome likelihoods.
Yeah, Nic, I'm never confident about understanding stat models. Interpreting them is so subjective. It's the utility of the model that I try to comment on; the extent to which it seems to map the underlying social reality.
That 24% differential is the crux of the issue, eh? When I see all evading the crux, I always try to figure out why such herding is occurring. Stats theory seems not to cover the crux, or if it does no commenter has explained how. Such mental blind spots are intriguing.
The way he talks in riddles I would more put David Farrar as "The Riddler".
Gerry is more your Penguin type waddling along and squawking with a bad expression on his face.
Gerry might be a better fit for the henchvillain from The Spy Who Shagged Me.
And equally bad breath.
Isn’t it a beautiful irony that the Nats are in the shit pollwise because of the stupid decisions they made while in government. Having relentlessly hoovered up every vote on the right because of their inbuilt FPTP mentality, pretty much whatever happens they are going to be well short of potential coalition partners and the numbers to govern on Sept 19th.
The smoothness of the figures suggests that Farrar is adjusting sample results on the fly, presumably to get a desired result. Thursday could be interesting …
No matter how you spin it National is in a mess.
first and foremost, farrar is looking out for himself. whatever happens in september, he will want to have an income.
Polling is a bit more of an art than a true science, unfortunately. You have to make a lot of assumptions that can make huge differences in the final results.
An experiment was done during the 2016 presidential race, I believe by Washington Post? They took raw poll data and gave it to 4 different polling companies to process how they saw fit. From the same raw data, they got results of +4, +3 and +1 to Clinton, and +1 to Trump. It depends on the demographic weightings and especially the "likely voter" profile each polling company decides to use.
This could be a poll of National supporters 49% of them want to vote Labour .
Nice of the hierarchy to share the figures with the troops. How many times in July have you heard National MPs extolling the necessity of transparency?
It's like they have people running in a race, they want them to keep running, they want them to run harder. They won't tell them where and how they are placed though so they have some idea of what they need to achieve or change by the finish line.
Is that some gobbledegook, weird technical approach shit from some genius free-market capitalist textbook? You know the sort of practice they want all their entrepreneurs to use in the business world they know and understand better than anyone?
And getting David Farrar in to be the special guest star? That's like a proctologist and an arsehole having a discussion about who was to treat whom.
Oh, now I see! They don't tell the caucus the Farrar poll results because they don''t trust people! They think someone mighty leak what was said. Oh dear, how sad, never mind.
I fear we are ALL being played …
What a pit of despair and destruction!
I remember your post from more than three months ago: https://thestandard.org.nz/matthews-magic-numbers/
49
31
9
6
5
If I remember rightly toward the start of the year the Nats under Bridges were polling in the mid to high 30's under Bridges but he wouldn't share the results of internal polling with his caucus. The general opinion was that the Nats internal polling was worse than the formal public polls (Colmar, Reid etc) so they were ignored. Obviously rogue results.
Now suddenly the Nats internal polling is being shared and used as an authoritative source and the formal public polls are the rogue results.
Interesting how the National leaders change their tune as to which poll to prefer.
Rogue poll? "Statistician and University of Auckland professor Thomas Lumley explains whether we can see this as one": https://thespinoff.co.nz/politics/27-07-2020/brownlee-says-national-is-the-victim-of-a-rogue-poll-can-that-really-be-true/
So, Brownlee's thesis being that the rogue outlier happens 5% of the time, can the margin difference between National & Labour shown by the RR/Curia comparison, which is 37% compared to 13% be explained by "just a little more than 3%"?
Well the difference between them both is a quarter of the electorate! That isn't `just a little'. So the `rogue' thesis is crap. The stats prof has eliminated it from contention with his inexorable mathematical logic.
Huh??
Yes, I sympathise. 😉 Farbeit for me to seem an apologist for career statisticians! However the logic is there for debate. The difference between the 37% margin in RR and 13% margin in Curia is 24% – effectively a quarter of the electorate.
To claim that huge number of voters is covered by "just a little more than 3%" is intellectual gymnastics of Olympian calibre, so Brownlee's leap is audacious and ambitious, but it doesn't fool the stats prof! Nor me. But let's wait & see how many commentators here he does actually fool…
Your on pretty shaky ground by subtracting two numbers the qualified stats prof used to come to a third number and then talking about it as meaningful. I think incognito was closer to the truth, huh??
Actually its not even clear where 13% comes into it and 37% appears to be a percent shy of Lumleys numbers.
Reid Research poll had Labour 60.9% (rounds off to 61) so that 1% error was mine (National 25%, difference 36%). Curia gave the 13% difference in their latter poll (and 10% in their earlier poll).
So the comparison of those party differences gives us 23% of the electorate for the later Curia poll (and 26% for the earlier one).
So the inter-poll variance is close to a quarter of the electorate, right? Given Farrar is having to persuade National that they can rely on his method, and the rest of the country presumably persuaded by science that they can rely on RR, we are left with two vastly different pictures of the public mood.
Or, as I framed it, two different publics being measured. The stats prof evaded the variance, of course. He's an academic. Nonetheless, despite his ideological aversion, the variance exists.
So eventually I worked out what I think your saying.
First, the obscure point is that Curia and RR polls differ by around 25% while both purporting to forecast the same election outcome. Yes, at most one of these polls can be close to the right outcome given there is such a gap between them.
Further to this Prof. Lumley is not saying anything of that nature, what he is saying is that, considering the 3 most recent poll results he discusses, the one Mr Brownlee thinks is the most reliable looks the most off trend. The trend being approximately that with each new leader National sinks lower in the polls. Is that a trend? Well as Lumley also highlights nothing in statistics can tell you one way or the other especially considering there were as many National party leaders as data points involved.
The reason however that your comment is confusing is because the reasoning is invalid. At least one of the two polls can't be a good forecast of the election result, so you can't draw further conclusions from at least one of the two poll results. I could take another also unreliable poll result, say the 2014 election result where National got 47% Labour 27%. So does this mean something about 56% of the electorate? Of course not it just means the 2014 election is a terrible model for the 2020 election. (Also note this assumes RR is the best model but actually we won't know which poll is best until the election validates Curia unexpectedly and makes Farrar look like a genius).
https://www.newshub.co.nz/home/politics/2020/07/nz-election-2020-newshub-refutes-claim-highlighted-by-judith-collins-about-push-polling.html
ok National saying TV3 poll was rigged –push polling
While it appears obvious Labour shall win the election, I feel many here are placing too much stock in polls. The fact is polls are fickle, easily manipulated, often selected from key demographic groups, and as recent history has shown, they are rarely correct.
USA…Polls told us Hilary Clinton in a landslide, yet Donald Trump became President.
UK…Polls told us the UK would remain in the EU, yet the leave vote won resoundingly.
Australia…Polls told us Labour couldn't loose the election, yet the Liberal – National still won the un-winnable election.
My forecast, Labour wins on its own merits (maybe 5 to 6 seats up its sleeve), Greens and NZ First out of parliament, ACT will make a marginal increase to 3 seats.
Thanks for your concern, however, a couple of points of contradiction.
From memory, the last U.S. Polls only had Hilary up to 4%, so predicted landslide is a dubious claim, though She did still win the popular vote, polling 2.1% more than Trump.
And a 1-2% margin in the eu referendum is hardly resounding in anyone's language.
This guy from Newcastle is a carbon copy of The Chairman, don't you think?
Have thought about that as much as I usually do the Chairman's posts, but I suspect that's not the similarity you were looking for.
The National Party is made up of Gerry-atrics and inexperienced children……both tend to leak.