A big welcome to the federal polling game to Ipsos, from the Poll Bludger and all who sail in her. GhostWhoVotes relates that Fairfax’s new pollster has opened its federal account with a result well in line with the overall trend, with Labor leading 51-49 on two-party preferred. The poll targeted 1400 respondents from both landlines and mobile phones. Still awaiting primary votes, but the poll shows 51% opposed to the notion of increasing the GST and reducing income tax, with 41% in support, and 54% opposed to the government’s paid parental scheme, with 40% in support.
UPDATE: The primary votes are 42% Coalition, 37% Labor, 12% Greens and 3% Palmer United. Labor’s lead on respondent-allocated preferences is 53-47, as opposed to the 51-49 result from 2013 election preferences. Tony Abbott and Bill Shorten are tied 41-41 on preferred prime minister, the former rating 42% on approval and 49% on disapproval, and the latter rating 43% and 40%.
UPDATE 2 (Morgan): After two relatively good results for the Coalition, the latest fortnightly Morgan result has then down a point to 38.5%, Labor up two to 37.5%, the Greens up half a point to 12.5% and Palmer United down half a point to 3%. Labor’s lead on the headline respondent-allocated two-party preferred result is 54.5-45.5 while previous election preferences is at 53.5-46.5, in both cases up from 52-48 last time. The poll was conducted by face-to-face and SMS over the last two weekends, from a sample of 3117.
[247
AussieAchmed
During the winter months there is little difference in times. Summer in Broome is around 1820hrs, Perth about an hour later.]
If the citizens of the North want DS, they should have it!
Last night, way back at 117, KB said:
[On the 2PP thing, I get around the sum-to-101 thing by simply multiplying everything by 100/101; the only other difference is that I add 0.14 to ALP for leakage from three-cornered contests.]
The response came:
[That assumes the rounding down in uniform. That probably is not the case.]
This type of reasoning just belies the average human’s ability to reason about probabilities. The question isn’t whether or not uniform rounding-down is correct or true, but whether any other rounding is more likely.
Think of it this way: say you have a biased coin, but you don’t know anything of the direction bias. So you know that the probability of getting a head isn’t 1/2. Nonetheless, in the absence of any information, 1/2 is still the best guess of the probability of landing a head (until you actually toss it).
In the same way, you could put a distribution over the different rounding techniques you could use, and then pick the one that is the most likely, or the one you have the most evidence for. Chances are, in the absence of any information, you’ll go with uniform rounding.
\end(stats-geek)
AussieAchmed@247
Sounds more like Broome would benefit from a permanent shift in time by an hour. The on again and off again nature of daylight savings would cause more inconvenience than any gain in daylight hours at the end of the day. As an example, here in Melbourne, the daylight can vary by around 3 hours within the year.
Also I remembered reading here there was a small town in Kalgoorlie that had a time of 7:45 from GST or something since it was on the border of SA.
Nicholas @ 206 – there is this ACIL Allen RET Review Modelling report, or, with some time, effort and a copy of PLEXOS, you can just do it yourself 😉
What will the LNP do?
Export-related investment is falling
Export volumes have climbed, but revenues are falling
Real per capita disposable incomes continue to fall
Underlying per capita demand for labour remains in trend decline
Blame the former Labor Govt
Blame unions
Daniel Andrews live on ABC 24
[252
Libertarian Unionist
The question isn’t whether or not uniform rounding-down is correct or true, but whether any other rounding is more likely.
Think of it this way: say you have a biased coin, but you don’t know anything of the direction bias. So you know that the probability of getting a head isn’t 1/2. Nonetheless, in the absence of any information, 1/2 is still the best guess of the probability of landing a head (until you actually toss it).]
I’m not sure this is the right way to look at a biased coin. Surely, if we know it’s biased, the correct conclusion is that the probabilities of a head or a tail are not equal – that 1/2 cannot apply to either result.
We can be sure we will get either a head or a tail. The probability of this is 1.0. While we cannot know the probability of either a head or a tail, we must know that 1/2 will be wrong.
Or am I just, once again, sadly confused?
Briefly
What are your thoughts on the likely impacts of mechanization on real household incomes and the degree of income inequality in Australia?
What could we do to make mechanization lift living standards and expand freedoms for everybody?
Are we in danger of entering a Kurt Vonnegut Player Piano dystopia?
[AussieAchmed
Posted Monday, November 3, 2014 at 1:13 pm | PERMALINK
What will the LNP do?
Export-related investment is falling
Export volumes have climbed, but revenues are falling
Real per capita disposable incomes continue to fall
Underlying per capita demand for labour remains in trend decline
Blame the former Labor Govt
Blame unions]
Blame Putin?
Weasel words from the Prime Weasel of Australia:
People can come to their own conclusions.
Well… WHY THE F*CK SPEND $25 MILLION ON A ROYAL COMMISSION IF PEOPLE CAN COME TO THEIR OWN F*CKING CONCLUSIONS?
253
Raaraa
You’re thinking of the Mundrabilla….such a hot spot!
http://www.tripadvisor.com.au/Hotel_Review-g255101-d3799014-Reviews-Mundrabilla_Roadhouse-Western_Australia.html
What will the LNP do?
Export-related investment is falling
Export volumes have climbed, but revenues are falling
Real per capita disposable incomes continue to fall
Underlying per capita demand for labour remains in trend decline
Blame the former Labor Govt
Blame unions
Blame Putin?
Blame Gough/Keating/Hawke/Swan?
Blame IS/Al Queda
or flat battery in Hockey’s abacus
[258
Nicholas
Briefly
What are your thoughts on the likely impacts of mechanization on real household incomes and the degree of income inequality in Australia?
What could we do to make mechanization lift living standards and expand freedoms for everybody?
Are we in danger of entering a Kurt Vonnegut Player Piano dystopia?]
You mean like this?
http://www.youtube.com/watch?v=n_1apYo6-Ow
briefly,
Adopting a Bayesian PoV, you need to start by defining a prior over the parameters of system, which in this simplest of cases is the probability of a head p.
A decision is made to choose the form of the prior Prob(p). One standard approach is to choose the prior that assumes the least information about the system, or the lowest Shannon entropy distribution over all possible choices of Prob(p). This happens to be the uniform distribution over (0,1). However, you might make a more informed decision and concentrate more of the probability around 0.5, reflecting your (undoubtedly ample) experience with baised coins; this would still have to be a symmetric distribution to fit the assumption that you start without knowledge of the direction of the bias.
In any case, the expectation of the uniform distribution, or any symmetric distribution on a finite support, is it’s midpoint, p=0.5.
Viola!
Now, when you toss the coin, you gather some information about the system, and can update both the distribution over p and the expected value of p accordingly – which means it’s time to employ Bayes’ rule.
AA
Blame council for restricting Joe’s access to food.
[What will the LNP do?
Export-related investment is falling
Export volumes have climbed, but revenues are falling
Real per capita disposable incomes continue to fall
Underlying per capita demand for labour remains in trend decline
Blame the former Labor Govt
Blame unions]
the daily rupert will sing their praised for a low aussie dollar, low interest rates and ‘standing firm’ as australian manufacturing goes down the gurgler – as well as put pressure on for workchoices and other ‘reform’. the ABC News will copy what the Oz says.
the worst will hit in 2017 when automotive manufacturing ceases and the Vic manufacturing sector collapses almost completely. In the past I heard it said that the automotive sector, suppliers and ‘multipliers’ made up 25% of the private sector earnings and employment in Vic. not sure if it is true/still the case, but it would easily be 25% of manufacturing in Vic. That’s a lot of people and it is hard to see where the new jobs will come from at present.
Raaraa@246
A lot of work involves collaboration or interaction with others. This would break down under your suggestion unless ways were worked out to overcome the problems.
So whilst you may solve some transport problems, it would be at the cost of productivity.
258
Nicholas
Basically, the frontier economy – the place where the principles of sustainability and human creativity intersect and give rise to innovation – is our best hope.
By using design and engineering we can create the new products and enterprise forms that will increase real “income” (taken in all its senses) and lift the world’s populations into durable, equitable prosperity.
Creativity is immanent. We have no choice in this. Change cannot be suppressed. Our choice lies in how we respond to change – whether we will support and marshal the proceeds of change? Or will we try to turn away from it? We need to find and apply all the means we need – including especially the courage – to make choices rather than to run from them.
264
Libertarian Unionist
I get all that.
But a priori we know the coin is biased. Isn’t it therefore consistent with Bayes to predict the probability will not be 0.5. It’s just about the only thing we could predict with high certainty. Depending on the bias, just about every other probability is possible, but 0.5 is not.
Jokowi shows he’s willing to upset the establishment (and it won’t do his popularity any harm either) Jakarta Post
[President Joko “Jokowi” Widodo has forgiven Muhammad Arsyad, a satay vendor who was detained by the police for allegedly defaming him online.
Jokowi said he forgave Arsyad during a meeting with the satay vendor’s parents, Mursidah and Syafrudin, at the State Palace on Saturday.
“I forgive him 100 percent. I forgave him through his parents,” Jokowi told reporters. First Lady Iriana was also present during the 30-minute meeting.
Jokowi added he had asked the police to release Arsyad from detention. “Regarding the legal process, I will check on it again later. But most importantly, I already asked {for the detention to be postponed},” he said.
Jokowi also called on the public to learn from Arsyad’s case and promote decency and respect for others online.
]
Jokowi is taking a big risk in responding to public opinion in a society where the powerful elites demand “respect” from the plebs. However he seems determined to cultivate popular support in a manner that will hopefully see a reduction in the power of the rich and corrupt.
258
Nicholas
In our economy and society, and given the technologies that exist and are being created, opportunity for change is a resource. It is one that is often under-used and misapplied.
By definition, if we do not change our current practices and forms, we will derive the same results. If we want new results, we must find ways to impel, select from and manage change. This is a practical as well as a political necessity – one that none of us is permitted to ignore.
Do we know if the two party preferred figure is respondent-allocated or preferences distributed in line with the 2013 election?
pseph,
Yes , we do.
[But a priori we know the coin is biased. Isn’t it therefore consistent with Bayes to predict the probability will not be 0.5. It’s just about the only thing we could predict with high certainty. Depending on the bias, just about every other probability is possible, but 0.5 is not.]
I didn’t ever say that 0.5 has have positive probability under the prior, just that for any symmetric prior, its expectation will be always 0.5, and to choose a non-symmetric prior would be to pick a bias… so your best point estimate/guess is still 0.5 😉
And taking that back to the uniform rounding example – just like the 0.5 guess for the coin’s bias, you can be almost certain this is not the correct way to do the rounding, yet still be happy that it is the least-wrong thing to do.
[yet still be happy that it is the least-wrong thing to do]
If only the better half was satisfied when I did the least wrong thing!
I have a sneaking suspicion that George Brandis isn’t without political/professional sin so I dare say that if he insists on living by the sword, he’ll end up….
It’s really intriguing how a strong public political campaign, no matter how ludicrous the policy issue is, can capture the imagination and support of the public.
[If only the better half was satisfied when I did the least wrong thing!]
Amen, bro 😉
But taking it to the full geek – it may be that your BH’s prior belief over what actions you should take is radically different from your own… but I couldn’t possibly comment on that!
http://kevinbonham.blogspot.com.au/2014/11/victorian-state-election-late-october.html
Victorian State Election: Late October Polls And Seat Model
This is a first run of my Victorian state election seat modelling methods, which will become more complex when we see some seat polls. I’m also still fiddling with some of the parameters, which may change.
I am currently aggregating the 2PP at 52.6 to ALP.
I am currently projecting that that 2PP, if it happened on election day, would result in a seat breakdown of about 48-40-0.
The article comments on the ABC calculator’s latest nonsense in which no matter what poll you plug in, three seats are projected to the Greens. They do have chances in some seats but at present there is no polling evidence to back that they are all that likely to win any given one. Of course a Liberal decision not to preference against them would change that.
Anyone else just have a prob connecting to PB?
Yes.
281
Yes.
Seems better now though
The crikey gerbils were whining that they didn’t like daylight saving. Went on strike.
Raraa
Broken Hill runs on South Australia time, as well as having a lot of other stuff from SA rather than NSW. It feels more connected with SA than NSW as Sydney is so far away. If the border was redrawn to include Broken Hill in SA, Broken Hill would not be dismayed. it will never happen though.
yes…jamming…
LU, you’ve made me terribly unhappy. If you’re proposition is correct, then we are obliged to assign values that we know in advance will be wrong. This is a lousy rule. It means the one thing we know for certain will be false is the selection we will nevertheless make. I’m mortified. It’s the kind of rule that ordered hundreds of thousands of soldiers to march towards annihilation on The Somme.
Greensborough etc,
Thanks!
[285
lizzie
The crikey gerbils were whining that they didn’t like daylight saving. Went on strike.]
I must be a gerbil. I usually blame DS when anything annoying happens.
[The crikey gerbils were whining that they didn’t like daylight saving. Went on strike.]
I thought they had gone out in sympathy for garden gnomes.
briefly, you’re not obliged to do anything. The question is how fast you will converge on an answer.
For example, given an adversary and several coins, the best strategy may be to start randomly. Otherwise your adversary could use your known starting point/strategy to set things up in such a fashion that you always end up taking the longest time possible to come to an answer.
Briefly
[LU, you’ve made me terribly unhappy. If you’re proposition is correct, then we are obliged to assign values that we know in advance will be wrong. This is a lousy rule. It means the one thing we know for certain will be false.]
I think you are confusing two distinct things:
1) the probability of getting heads on a single toss of a biased coin.
2) the probabilty of guessing the bias of a coin correctly with a single estimate.
Without any additional evidence the best guess of the probability for the first thing is 0.5 (as LU has outlined).
Without any additional evidence there is no sensible way to guess the bias of the coin. You are correct that the probability that it is unbiassed is zero.
It’s a most beautiful day here. Calm, sunny, 19 deg. The garden smells sweet with flowers. Baby ducks on the lake and tortoisess sunning themselves. Blue and yellow irses just in bloom at water’s edge. Young magpie calling for food. Robin in the shadows searching the ground for insects. Golden whistler. Butterflies.
And, of course, for a continuous variable, no answer is going to be “the answer”, just answers with varying degrees of wrongness.
[It’s a most beautiful day here. Calm, sunny, 19 deg. The garden smells sweet with flowers. Baby ducks on the lake and tortoisess sunning themselves. Blue and yellow irses just in bloom at water’s edge. Young magpie calling for food. Robin in the shadows searching the ground for insects. Golden whistler. Butterflies.]
Sounds like heaven don’t think I’d ever make it onto PB!
WWP
A girl has to sit down for food sometime!
Big call from Clive.
[Clive Palmer @CliveFPalmer 46s ago
A QLD-elected @PalmerUtdParty Govt will resume public assets leased by @theqldpremier’s Govt without compensation to private sector #qldpol]
[292
Jolyon Wagg
I think you are confusing two distinct things:
1) the probability of getting heads on a single toss of a biased coin.
2) the probabilty of guessing the bias of a coin correctly with a single estimate.
Without any additional evidence the best guess of the probability for the first thing is 0.5 (as LU has outlined).]
I’m pretty sure I’m confused. You’re right about that.
What we have is a conundrum. Even though we’ve been given a coin we know to be biased, we will assign a probability of 0.5 to throwing heads on the first throw, even though we know the probability is not 0.5. That is, we will choose a probability that is one of the least likely to be right.
Are we not conflating two universes: one in which coins are not biased and one in which they are. Why should we suppose that the probabilities that applied in an unbiased universe should be the starting point in the biased universe? Would we apply this rule in reverse?
Isn’t it really better to say that all coins have a possible bias including a nil bias? It would follow that until we know more about what the bias is, we can’t make predictions about the probabilities for that coin/universe. We can make guesses, but this is not the same as forecasting probabilities.
[291
DisplayName
briefly, you’re not obliged to do anything. The question is how fast you will converge on an answer.
For example, given an adversary and several coins, the best strategy may be to start randomly.]
This is the Donald Rumsfeld dictum, which accepts there are things we do not know that we do not know…. 🙂
Morgan: ALP 37.5 (+2), L-NP 38.5 (-1), GRN 12.5 (+0.5), PUP 3 (-0.5). Respondent-allocated 2PP: 54.5-45.5, up from 52-48. Previous election 2PP: 53.5-46.5, up from 52-48.
http://www.roymorgan.com/~/media/Files/Findings%20PDF/2014/October/5905-Fed-Vote-November-3-2014.pdf