The election thread - Two middle-late aged white men trying to be blokey and convincing..., same old shit, FFS.

[johnny]

Agreed.

And this is why pollsters get things so wrong. Their poll is only represetative of those who took the poll. If a pollster calls me I instantly hang up, so my view is unrepresented.

As you say, the ABC is not a representation of all people, just those who could be arsed doing that test.

Polls used to be a lot ore accurate though, because they balanced for issues like yours and the ones I cited. The difference, so I read now, is that communications have changed with mobile phones and the internet, but polling practices and methodologies haven't kept up.

 

[Calvin27]

And this is why pollsters get things so wrong. Their poll is only represetative of those who took the poll. If a pollster calls me I instantly hang up, so my view is unrepresented.

I've always wondered how they deal with 'no english' types and whether this is reflected in their error range. Then there is you who hangs up and me who doesn't even have a landline. So effectively most polls already fail as soon as the pick a medium, even if they have broad coverage because the engagement is variable.

Add to the mix social media and overwhelming opinions and judgements on both sides and it's become something only world cup octopus might excel at. For example, no one admitted to voting Trump but clearly half the country did. Sort of like climate change, most polls indicate there is broad support and people constantly putting it as the highest priority, but nope, franking credits are more important in the end, we'd like someone else to pay for it.

Honestly these days polls are not worth much anymore. A better approach would be to consider clarity o the message and the impact of policy on different groups.

 

[Binaural]

And this is why pollsters get things so wrong. Their poll is only represetative of those who took the poll. If a pollster calls me I instantly hang up, so my view is unrepresented.

Actually, you'd probably be surprised. While individual pollsters are not that reliable, the aggregate performance can actually be surprisingly good. The problem is that in Australia we don't have anyone doing a really good job at this, using robust statistical methods that take account of uncertainty and bias. Take a look at how the most prominent analyst in the US does this, and you will see that none of the problems listed in this thread are really insurmountable.

(Professionally, dealing with uncertainty and bias is my everyday work. Watching people overreact to one poll or the other is one of the little irritations of my daily political reading).

https://fivethirtyeight.com/features/the-polls-are-all-right/
https://projects.fivethirtyeight.com/pollster-ratings/

 

[Haakon]

Who knows anything about David Speers? I never watch anything on telly, and obviously not Sky - but he seems well regarded?

Or is my suspicion that its more takeover by stealth at play justified? The LNP has already got Murdoch goons onto the board, this seems a natural next step.

 

[Comic Book Guy]

David Speers

'Chuffed' Cassidy welcomes Speers to Insiders, but no confirmation from ABC management

Outgoing Insiders host Barrie Cassidy says he is pleased that Sky News political editor David Speers looks set to become the next host of the Sunday morning political discussion program he created more than 18 years ago.

www.abc.net.au

 

[scblack]

Actually, you'd probably be surprised. While individual pollsters are not that reliable, the aggregate performance can actually be surprisingly good. The problem is that in Australia we don't have anyone doing a really good job at this, using robust statistical methods that take account of uncertainty and bias. Take a look at how the most prominent analyst in the US does this, and you will see that none of the problems listed in this thread are really insurmountable.

(Professionally, dealing with uncertainty and bias is my everyday work. Watching people overreact to one poll or the other is one of the little irritations of my daily political reading).

https://fivethirtyeight.com/features/the-polls-are-all-right/
https://projects.fivethirtyeight.com/pollster-ratings/

I have studied statistics, it remains an educated guess. Very robust maybe, but still just an estimate.

Different sample sizes, different methodologies, many other factors can affect results.

 

[Binaural]

I have studied statistics, it remains an educated guess. Very robust maybe, but still just an estimate.

Different sample sizes, different methodologies, many other factors can affect results.

Isn't a robust estimate the point? It is a future event, after all.

The main point here is that methods exist to deal with errors and bias from all those sources you mention. All this scepticism about polling I see here is not a fundamental limitation, it stems from a lack of the use of better statistical methods.

 

[droenn]

The main point here is that methods exist to deal with errors and bias from all those sources you mention. All this scepticism about polling I see here is not a fundamental limitation, it stems from a lack of the use of better statistical methods.

You may be right, but there is the other side to it that the statistical methods are fine, just the underlying data reflects an inherent issue with the polling process (beyond obtaining a representative sample). Polls might produce answers (observations) that are different from the population, due to other factors that are part of polling (process error). And the degree of consistency in this (which may change around a lot) would determine how easy it is to capture this in the models. I think this is actually the hard bit, and as pollsters don't like being outliers from the other polls either, they already hedge with other factors that draw their predictions in line with the other results. Hard to go out on a limb and trust the inference sometimes.

 

[Dales Cannon]

This is quite misleading though as it does not accurately represent the whole nation. It only represents the segment of population that use the ABC website. Given Aunty's reputation for being left, I'd assume that there is a pretty large element of the conservative side of Australia that wouldn't have participated in the poll.

It's only a snapshot of people who access the internet and who navigate to the ABC website and have the time to take the Vote Compass poll (and those who take it seriously, unlike the way I did)

Yes, just their results and not representative of the nation as a whole. ABC didnt suggest it was anything else. Thought it worth sharing though

 

[Dales Cannon]

I have on purpose told the exit polling minion something completely different to what I voted. Other polling hasnt kept up with modern times. Just wait you bastards.

 

[johnny]

Yes, just their results and not representative of the nation as a whole. ABC didnt suggest it was anything else. Thought it worth sharing though

Definitely worth sharing and interesting to look at but I do think that ABC suggests it as something else. for example, the title and subtitle read thusly:


Federal election: This is Australia as 100 people
If the Australian electorate were made up of just 100 people, what would they think about the big issues facing the country — and how did they vote?


Then down the page we have):


Vote Compass found the environment and the economy were pretty much neck and neck as the top issue.


Then this:


Across the election campaign, Vote Compass provided a detailed snapshot of what the Australian electorate is thinking


If you look at the notes down the bottom, it leads to a page where they defend their survey against the claims of self-selection - https://www.abc.net.au/news/2016-06...the-vote-compass-methodology-holds-up/7498692

But the problem they have with that explanation (based on my limited knowledge of methodology) not only is their survey based on self-selection, as all polls are to a degree, but that they are not taking active measures, other than poststratification to manage the issue. All they are saying is that "all polls have an element of self-selection, so that means that any criticism of ours is invalid". Where as, for me, the only acceptable response would be "to manage the problem of the layered self-selection, that being, selecting to come to our website, selecting to look at the vote compass page, selecting to take the survey, etc., we have combined the online results from these surveys carried by ................ at ......................., etc. etc.".

So in closing, your honour, I'd argue that ABC have 100% suggested that their polling represents the nation as a whole.

 

[Haakon]

Cool graphics though. And if they had any money they could have had a wider survey

#nocutstotheabcetc...

 

[Dales Cannon]

True that. I didnt take that on board. I guess I was thinking it didnt come up in their coverage, that I saw, on the night.

I liked how a survey of 1 out of 1 Canberra public servant voted meth dealers as his biggest issue.

Trouble with surveys is the lack of interest in filling them out, this was clever in that it gave you were you sat in the political spectrum based on their algorithms and you won a free trip to Bali.

 

[Haakon]

Meet the Black Throated Finch https://birdlife.org.au/bird-profile/black-throated-finch

Say goodbye to the Black Throated Finch...

 

[pink poodle]

I liked how a survey of 1 out of 1 Canberra public servant voted meth dealers as his biggest issue.

It can be a huge issue when you can't find a meth dealer in Canberra.

 

[Dales Cannon]

Meet the Black Throated Finch https://birdlife.org.au/bird-profile/black-throated-finch

Say goodbye to the Black Throated Finch...

Is it becoming extinct because of your air travel?

 

[Haakon]

Is it becoming extinct because of your air travel?

Nope, that’s just your kid’s future I’m killing off there. This is the bird Adani and Clive will kill off.

https://www.theguardian.com/environ...ronmental-approvals-given-three-week-deadline

 

[SummitFever]

...If a pollster calls me I instantly hang up, so my view is unrepresented...

In this day and age of scammers and widespread telemarketing I am amazed that anyone takes a pollster call. The last thing I want to do is provide any information whatsoever to some unknown person over the telephone.

 

[Dales Cannon]

In this day and age of scammers and widespread telemarketing I am amazed that anyone takes a pollster call. The last thing I want to do is provide any information whatsoever to some unknown person over the telephone.

I start by asking what they are wearing...

 

[slowmick]

I am not sure that adding the extiniction of a particular bird species to the reasons in articles for not wanting the Adani mine is a good plan. If people don't care about the loss of water usage for food production and drinking water as good enough reasons what difference is a bird going to make. It just seems to be the thing people pick up in a conversation and use it as a bat - "so i lose my well paying jobs for a fucking bird" (Based on extended family members who work in QLD coal mines)