|
Are you a This poll is closed. |
|||
|---|---|---|---|
| homeowner |
|
39 | 22.41% |
| renter |
|
69 | 39.66% |
| stupid peace of poo poo |
|
66 | 37.93% |
| Total: | 174 votes | ||
|
What are the methodologies of the various polls?
|
#
?
Sep 20, 2017 17:43
|
|
|
|
| # ? May 25, 2024 19:02 |
|
|
Gum posted:What are the methodologies of the various polls? Landlines baby! It's especially not clear how they handle the early voting though. Lord of the Llamas fucked around with this message at 18:10 on Sep 20, 2017 |
|
#
?
Sep 20, 2017 17:52
|
|
|
Lord of the Llamas posted:Landlines baby! What about turnout models?
|
|
#
?
Sep 20, 2017 19:48
|
|
|
Gum posted:What about turnout models? http://www.colmarbrunton.co.nz/wp-content/uploads/2017/09/Prelim_1-News-Colmar-Brunton-Poll-report-9-13-Sep.pdf Looks like a simple self reporting model but you can read for yourself  The report for the very latest one doesn't appear to be on their website. Haven't seen any polling reports for the other firms. NZ polling seems to be a pretty awful. As a foreign observer who knows very little about NZ I'd conjecture from what I've seen that if there's been a recent shift to a more generational divide amongst NZers like what we've seen in the UK then you're ripe for a polling upset despite their past decent performance.
|
|
#
?
Sep 20, 2017 20:23
|
|
|
Lord of the Llamas posted:As a foreign observer who knows very little about NZ I'd conjecture from what I've seen that if there's been a recent shift to a more generational divide amongst NZers like what we've seen in the UK then you're ripe for a polling upset despite their past decent performance. The problem there is that all the polls that have been widely out have had inflated numbers on the left. So that's terrifying.
|
|
#
?
Sep 20, 2017 20:54
|
|
|
My father grew up in a communist dictatorship and he recently visited here after a few years living in Australia. I asked him about what he thought of all this and mentioned the interesting polls. He just laughed.
|
|
#
?
Sep 20, 2017 20:57
|
|
|
[quote="“The Rabbi T. White”" post="“476591963”"] The problem there is that all the polls that have been widely out have had inflated numbers on the left. So that’s terrifying. [/quote] That's normal tho. [quote="“Lord of the Llamas”" post="“476590964”"] As a foreign observer who knows very little about NZ I’d conjecture from what I’ve seen that if there’s been a recent shift to a more generational divide amongst NZers like what we’ve seen in the UK then you’re ripe for a polling upset despite their past decent performance. [/quote] It certainly feels like there's a mood for change, but young people have an even lower enrolment rate than in 2014. There definitely hasn't been anything like corbynism.
|
|
#
?
Sep 20, 2017 20:59
|
|
|
The Rabbi T. White posted:The problem there is that all the polls that have been widely out have had inflated numbers on the left. So that's terrifying. That's what they said in the UK too, you should never buy into conventional wisdom built on tiny sample sizes (by which I mean elections and their results not the samples of polls themselves).
|
|
#
?
Sep 20, 2017 21:22
|
|
|
exmarx posted:That's normal tho. The 18-24s having a lower enrollment rate than last election isn't that indicative of potential youth turnout, as the <35 age bracket is pretty left-wing in NZ and any increase in that demographic's turnout from 2014 could result in a polling upset. It's unlikely to happen, but entirely possible. e:  Confirming the NZ electorate is terrified of any kind of tax increase. El Pollo Blanco fucked around with this message at 02:52 on Sep 21, 2017 |
|
#
?
Sep 21, 2017 02:31
|
|
|
As unenrolled walk-in votes at advance places are special votes, does anyone know if they are counted in the early vote stats?
|
|
#
?
Sep 21, 2017 04:03
|
|
|
Interesting that more Nats voters think Labour was wrong to back down on tax than Labour voters. Not sure what to make of that.
|
|
#
?
Sep 21, 2017 04:31
|
|
|
Wandle Cax posted:As unenrolled walk-in votes at advance places are special votes, does anyone know if they are counted in the early vote stats? Someone asked this on Twitter, and the answer given was no, they're counted later with the rest of the special votes.
|
|
#
?
Sep 21, 2017 04:34
|
|
|
Wafflecopper posted:Interesting that more Nats voters think Labour was wrong to back down on tax than Labour voters. Not sure what to make of that. That divergence is meaningless because a % of Nat voters are always going to be opposed to anything Labour say or do. The info you get from this poll is that Labour and National voters are equally opposed to tax increases, essentially.
|
|
#
?
Sep 21, 2017 04:41
|
|
|
My baseless & probably too-soon guess is that the guy setting himself on fire is health-related 
|
|
#
?
Sep 21, 2017 05:23
|
|
|
exmarx posted:My baseless & probably too-soon guess is that the guy setting himself on fire is health-related The who, what? e: oh poo poo found it
|
|
#
?
Sep 21, 2017 05:25
|
|
|
https://twitter.com/MattCNBC/status/910715887946100736
|
|
#
?
Sep 21, 2017 05:30
|
|
|
exmarx posted:My baseless & probably too-soon guess is that the guy setting himself on fire is health-related Seems to be about protesting father's rights, and male suicide or something?
|
|
#
?
Sep 21, 2017 05:55
|
|
|
Oh drat, I laughed to myself about a poster for a fathers rights thing @ parliament when I went out for lunch today
|
|
#
?
Sep 21, 2017 06:02
|
|
|
I forgot that we even had a Communist party.
|
|
#
?
Sep 21, 2017 06:28
|
|
|
poo poo. I saw a guy standing not far from Parliament with some placards as I bussed to work this morning. Wonder if it was him. Only had time to make out something about "incomprehensible lawyers' games" before we were past him though
|
|
#
?
Sep 21, 2017 06:50
|
|
|
This was the dude's sign:
|
|
#
?
Sep 21, 2017 06:56
|
|
|
Cult and Lawyers? That's a new one.
|
|
#
?
Sep 21, 2017 07:03
|
|
|
Newshub latest poll: 
|
|
#
?
Sep 21, 2017 07:03
|
|
|
Do you think it was the same guy who posts those political cartoons around the CBD? You know, the ones with Fozzie Bear talking about how John Key was obviously murdering him with fluoride because his case wasn't heard at the supreme court? It looks like his writing style, anyway.
|
|
#
?
Sep 21, 2017 07:03
|
|
|
SurreptitiousMuffin posted:Newshub latest poll: gently caress I sure am looking forward to NZ spending $200 million on whatever referendums Winnie demands in return for a coalition deal.
|
|
#
?
Sep 21, 2017 07:12
|
|
|
TOP only needs a 4.1% swing by saturday
|
|
#
?
Sep 21, 2017 07:13
|
|
|
Good loving damnit Winston is going to get to choose a government again isn't he? Well I guess we get to see how genuine he is in his "failed neoliberal experiment" rhetoric
|
|
#
?
Sep 21, 2017 07:19
|
|
|
It'll be interesting. A few of his MPs (e.g. Martin) seem to genuinely hate the Nats. Here's hoping Jonesy misses out tho.
|
|
#
?
Sep 21, 2017 07:24
|
|
|
Can someone who understands statistics tell me whether a 3.1% margin of error means that all of those voting figures are +/- 3.1%, or whether it's 3.1% of each figure. eg If Labour's on 37.3% does that mean they could be anywhere from 34.2 to 40.4%, or does it mean they're between 36.927 and 37.673, or something else entirely I'm suspecting not the first because that would make polling almost entirely meaningless. Actually that doesn't sound so far fetched after all
|
|
#
?
Sep 21, 2017 07:54
|
|
|
It's an absolute value, but it doesn't quite mean "it could be anywhere +-3.1%" but that's pretty close. The margin of error is the 95% confidence interval, so if they did the survey 20 times the results would be within 3.1% of the values they reported 19 times.
|
|
#
?
Sep 21, 2017 08:57
|
|
|
Ahaha Winston's back baby! How far do you have to dig into the nzfirst list this time before the completely insane ones turn up.
|
|
#
?
Sep 21, 2017 10:01
|
|
|
Gonna be a Nat-NZF government imo
|
|
#
?
Sep 21, 2017 10:07
|
|
|
Wafflecopper posted:Can someone who understands statistics tell me whether a 3.1% margin of error means that all of those voting figures are +/- 3.1%, or whether it's 3.1% of each figure. eg If Labour's on 37.3% does that mean they could be anywhere from 34.2 to 40.4%, or does it mean they're between 36.927 and 37.673, or something else entirely The way I understand it is that 3.1% is for a party polling at 50%, so for any realistic NZ application it will be proportionately less than that based on the party vote.
|
|
#
?
Sep 21, 2017 10:12
|
|
|
El Pollo Blanco posted:That divergence is meaningless because a % of Nat voters are always going to be opposed to anything Labour say or do. The info you get from this poll is that Labour and National voters are equally opposed to tax increases, essentially. Or it could mean Labour voters are convinced that the tax policy would have been unpopular with other voters and therefore it was correct to back down in the interests of winning. Wafflecopper posted:Can someone who understands statistics tell me whether a 3.1% margin of error means that all of those voting figures are +/- 3.1%, or whether it's 3.1% of each figure. eg If Labour's on 37.3% does that mean they could be anywhere from 34.2 to 40.4%, or does it mean they're between 36.927 and 37.673, or something else entirely bike tory posted:It's an absolute value, but it doesn't quite mean "it could be anywhere +-3.1%" but that's pretty close. The margin of error is the 95% confidence interval, so if they did the survey 20 times the results would be within 3.1% of the values they reported 19 times. It's important to remember this only refers to the sampling error which is the level of variation introduced by taking a random sample of a given size from a larger population. Of course telephone polls are not simple random samples for reasons including, but not limited to, (i) people who don't have landlines, (ii) people who don't want to answer your poll, (iii) people who lie, (iv) people who tend not to be at home when you try and contact them, (v) households where e.g. the man is always the person to answer the phone, etc. Internet polls also face similar issues, especially stuff like response bias. This is why we use weightings to try and turn a raw sample into a representative sample, and how accurate those weightings are (you can't know for sure) introduces more uncertainty. Finally in terms of producing polls that are indicative of election results pollsters will tend to apply some kind of a turnout model, e.g. only those who say they are certain to vote, which is also an imperfect instrument that has a degree of uncertainty attached to it. And that's not to mention the effects of stuff like tactical voting, although that's probably not much of an issue in the NZ electoral system. Or the fact that people can change their minds between a poll and the election, and undecideds might break one way stronger. A good example of that problem was in the US election where there were twice as many undecideds than normal but they skewed heavily as soft Republican voters who didn't like Trump much - but they still tended to break Trump on the day. Overall the takeaway is that you should not be surprised when election result is further than the naive sampling error bounds that are often touted around with great certainty by people who don't really understand how polling and statistics works. The assumptions that a pollster makes beyond demographic weightings can have a huge effect on the results. In the UK election this year mainstream pollsters had the Conservative lead at anything between 1% and 13% (!), but this variation was mainly due to modelling differences and not regular statistical variation. To take one example this IPSOS/MORI poll which had Labour and Conservatives tied at 41% each after demographic weights suddenly turned into 44% vs 36% lead for the Conservatives when they applied their turnout and refused reallocation models. It doesn't look like Colmar Brunton are doing anything overly complicated from the basic information they publish. But they don't actually publish their weightings or the demographic crosstabs so all we can do is speculate. One of the sources of "extra" uncertainty you get with weighting is when you have to inflate a small raw sample for the final result, e.g. you know 10% of the electorate is 18-24 but only 3% of your 1000 respondents were in that demographic. This means that your sampling error for that sub demographic is actually for 30 people not 100 (of course you can have an over-sample which would have smaller variance too, e.g. for very old people who always answer the phone), so hopefully now you can see that a poll is really the weighted sum of lots of small polls and therefore taking a 1000 size sample and computing a ~3% sized 95% confidence interval is really just junk statistics.
|
|
#
?
Sep 21, 2017 10:14
|
|
|
Lord of the Llamas posted:hopefully now you can see that a poll is ... really just junk statistics.
|
|
#
?
Sep 21, 2017 11:53
|
|
|
should i party vote labour or greens? I would prefer to vote labour but not if greens are gonna miss 5%
|
|
#
?
Sep 22, 2017 02:11
|
|
|
Should be a nice day tomorrow which favors leftie turnout
|
|
#
?
Sep 22, 2017 02:39
|
|
|
A vote for Greens is a de facto vote for Labour as that's their only coalition option. That said, i think their chances of not making 5% have been heavily overstated by a media desperate to have something to talk about in the face if National's strategy of having no policies to promote.
|
|
#
?
Sep 22, 2017 02:42
|
|
|
Vote Greens to get a higher chance of getting the amazing Golriz Ghahraman into parliament. Vote Labour if you want a higher chance of Trevor Mallard getting bac-ahahahah ahahahah ha.
|
|
#
?
Sep 22, 2017 02:45
|
|
|
|
| # ? May 25, 2024 19:02 |
|
|
When it was looking like Labour and National were neck and neck there was a case to be made for voting Labour to ensure they were the biggest party. But the chances of that are basically non-existant at this point, so if you have no strong preference then vote Greens because they are great~
|
|
#
?
Sep 22, 2017 02:45
|
|