It seems Galaxy has settled into a three-week polling pattern, compared with Newspoll’s fortnightly, ACNielsen’s monthly and Roy Morgan’s weekly (usually – they seem to have taken last week off) UPDATE: Sorry, it’s actually been four weeks since the last Galaxy – the previous three were three weeks apart. Today’s Galaxy survey has Labor leading 55-45 – still narrower than other recent polls, but a slight correction from its quirky 53-47 of three weeks ago. Despite the flak Galaxy copped last time, respondents were again asked a question about Labor’s union connections. They were also asked if the Prime Minister was “addressing problems in Aboriginal communities because of the upcoming federal election or because he really cares about the problem”. All revealed here.
triangulum:
If it is beyond 8% difference (my spies tell me it showed 58-42) then the Howard recovery team will not publish it
Something about the public not knowing whats good for them and only the born to rule and masterclass shall decide whose poll we will see LOL
Sounds a tad suspicious doesn’t it
The history of the Howard Reich is littered with suspicious events.
Be ready for a 9/11 event, it’s possibly the only thing that ‘might’ save the Liberals now.
I’m sure Matt Price was having a laugh at the conspiracy theorists with that one.
Not that the conspiracy theorists are short of ammunition.
The other possibility is that it was so glaringly obvious that it was a rogue poll, that they’d look silly publishing it. Maybe it showed Harold Holt making a comeback. Perhaps it predicted Bob Brown being elected joint President of Oceania, and Peter Costello mistakenly getting crowned Queen of Denmark. And perhaps they haven’t explained it because they still haven’t figured out why the glitch occurred.
Looks like a strong result for Labor if they’ve hidden it!!!
Could it be that Newspoll was not conducted last weekend due to school holidays in Victoria and maybe other states?
netvegetable:”The other possibility is that it was so glaringly obvious that it was a rogue poll, that they’d look silly publishing it. Maybe it showed Harold Holt making a comeback.”
That’s the theory I’m inclined to. (That it was a rogue poll, not that Harold Holt is making a comeback.)
Gary Bruce: “Could it be that Newspoll was not conducted last weekend due to school holidays in Victoria and maybe other states?”
Matt Price in the link above implies that the poll was carried out but the Oz decided not to publish it. I think they normally poll during school holidays, don’t they?
triangulum- the only days i think they dont poll are declared public hols-eg anzac day etc
Or maybe they are putting it back a week so it won’t clash with Nielsen and Galaxy
Gary, your theory is not in line with what Matt Price said.
Triangulum yeah. I still think though, that if we don’t get any explanation beyond ‘The results didn’t conform to News Ltd policy, Jay’, that it’ll be a little suspicious, don’t you?
I suspect that Matt Price was being ironic. He is the only News Ltd journo to at least acknowledge that the Oz is getting something of a reputation for being a government propaganda rag, even if he argues that it’s not true. Consequently, I suspect he was having a bit of a lend of us, as is his style.
hugo
Newspoll has already done their latest round of polling
it is just that the result is not what they want-
spin that any way you like
Gusface, how much reliability do you place on your “spies”?
only that one or two may work for said organisation
and that others were polled by newspoll
A lot of this kind of discussion is premised on the assumption that a poll result which shows the government losing ground, is somehow bad publicity for them.
But is it so?
I thought all political parties want to claim “underdog status” at an election, don’t they? They certainly seem to. Ergo, if the Newspoll shows the ALP gaining ground, that would actually be good publicity for the government wouldn’t it?
Gusface, please answer “a hell of a lot!”.LOL
not when you are trying to keep your coaltion together and look like the strong man
also would show how the public have totally written off the rodent and his biggles adventure in NT was pure politics not compassion
not good coming into APEC so expect a slurry of trash from the Libs
“I thought all political parties want to claim “underdog status†at an election, don’t they?”
Not to the extent that it wrecks whatever credibility remains bygetting cheersquads pulling polls from publication.
“whatever credibility remains’ … hilarious.
Off topic I know, but the Bridge Corp collapse is now the fourth in the past two years, at least three of which have occurred, to my recollection, under Joe Hockey’s stewardship as the relevant minister.
Surprisingly there has been little or no shooting home of responsibility to Shrieking Shrek.
Does anyone know why? Is it being saved for the campaign proper, or are there no votes for Labor in the pain of Howard’s Battlers or his beloved Mum & Dad investors and self funded retirees?
Netvegetable, I think a bounce in the polls is something that the government wants much more than to be behind. Underdog status is no good for you if the public think that you are actually a joke – they still need a reason to want to support you.
The government NEEDS to regain some of their lost support. And they have regained a small proportion of it over the last couple of months, but nothing of great significance.
By the way, if last weekend’s Newspoll seemed like a rogue, and this is why News Ltd. canned it, then ask yourself this question: If it had shown a substantial jump in support for the coalition, that is, a rogue in the OPPOSITE direction, do you think they would have canned it?
One thing about the Right is that they spend most of their time living in states of denial. Truth is their ultimate enemy.
Newspoll published their last rogue poll, you know, the 60 -40 one.
The lack of Newspoll is highly suspicious but surely not even the Oz would hide the poll if it wasn’t good news for the Govt. They published the 61-39 poll back in March and the 60-40 poll only last month. There are political junkies and scores of people in the political parties who know the polling cycle, not to mention the 1,100 people who were polled over the weekend. This would undermine the Oz’s (existing) credibility if they simply junked a poll like that. … Or am I terribly naive?
That’s true, Gary and Pseph. I’m heading into conspiracy territory! Still, it’s hard not to be suspicious of News Ltd – they seem so determined to help Howard win the next election. Their credibility, especially at The Oz, is at an all-time low.
Of course, if there really hasn’t been any change in public opinion or voting intentions, or even an increase in support of Labor, then it will probably show up in the next poll. Does anyone know if Morgan is publishing one this week?
Morgan normally polls every 1 to 2 weeks, and they haven’t released one for a while, so I’d assume that one will be up this week.
I don’t believe even the Daily Howard would simply can a poll whose result they didn’t like – it would be impossible to keep secret for one thing. But perhaps I am being naive again. And I don’t have any other explanation for the non-appearance of Newspoll.
its the synchronisation thing.
see mumble
makes sense to me
re Newspoll absence
Its probably awaiting good old Dennis Shannihan’s unique farsighted analysis of the numbers.
Remember his laughable effort last time announcing Rudd completely ‘unravelling’ when the 2PP dropped to 53-47?
c’mon dennis where are you? we all cant wait for your next rant. You make Janet A and Gilne Milne seem postiively visionary.
I love the conspiracy theories here!
I suspect that Matt was being ironic as noted above and the paper is trying to get out of sync with Nielsen.
You don’t need to can a bad poll, you just need Dennis S to “interpret” the poll according to the current doctrine. Doubleplus good result!
Further on linear regression:
It would be an interesting exercise to work out what the *typical* period over which linear extrapolation is reasonably predictive ie given the trend over the previous X weeks, the position in X weeks time will be… My guess is that it would be a few weeks, at most a month. If so, it would further highlight the nonsense of picking a trend in March and pushing it out to December.
We all know that linear regression isn’t a long-term model for voting intention. The only line that fits the long-term data is a dead flat one.
And as I suggested above, the whole exercise is based on where you start and finish.
Pick a trend starting on 16-18 March, and you get a clear trend to the government.
Start on 4 March and you get a clear trend, but about half the size.
Start in February or earlier and you get a trend to the ALP.
Pick your starting point; get the trend you want.
Anyone who points to a trend that started on 16 March but no earlier, without giving a plausible reason for that choice is on a fishing expedition.
It could be synchronisation. But it could also be that the Newspoll produced an extreme result. Given common conventions, it may seem reasonable for News Ltd to have a policy of rejecting these results on the grounds that the sampling was suspect – in a sense rejecting the outliers.
But (and it’s a big but) this does not mean that the rejected poll was wrong. It could accurately reflect, for instance, an unsuspected volatility in the electorate or a collapse in the vote for the Coalition or the ALP. So-called rogue polls should not be written off simply because the results are surprising. Doing so carries the danger of failing to see the trees for the grass.
Simple linear regression isn’t actually accurate over any given period for trends except by chance if you use point forecasts. If you use some interval forecast, like a point forecast with 95% confidence interval, then a simple linear regression will usually predict “correct†more than half of the time, but the confidence intervals are so large as to be near meaningless.
Multiple regressions get much better forecasts over the very short term, but can become very messy, very quickly the further out you go.
However, longer term trends such as the cubic time trend that has been underlying the primary vote of the two major parties since 1996 explains nearly half of the variation in the primary vote for each party. There’s a number of possible reasons for that, cubic time trends (like quadratic ones) have a lot of uses in explaining behaviour.
Black Jack, such an ‘extreme’ result would have to be incrdibly rogue to fit such a policy, if they have one. I am especially doubtful of any such policy, given that only a few months ago we saw two Newspolls just a week apart give 55-45 and then 60-40, which weren’t canned.
I was expecting Fairfax to amend THEIR schedule because of Newspoll, so it wouldn’t be shocking if News had changed theirs to avoid Fairfax. The school holidays proposition does not work because they have polled on this weekend in previous years.
And Price had his tongue firmly in cheek, almost certainly. There are simply too many people involved in taking a Newspoll for the Oz to ‘can’ one.
Seems Matt was probably taking the piss after all. Here is his response to one of the bloggers on his blog who asked him to explain what was wrong with the recent Newspoll:
“Whenever Labor’s 2pp goes above 70percent, h3llbee, we send the pollsters back into safe Liberal seats and RSL clubs for balance.”
http://blogs.theaustralian.news.com.au/mattprice/index.php/theaustralian/comments/unions_bashed_howard_horrified/P360/
I rang Newspoll and asked why no Federal Voting Intentions Poll today. They said they did not take one, and will be issuing one next week (Tuesday). No reason was forthcoming, even on close questioning.
Guys, Matt Price is a funny man at the best of times. He was just having you on.
Unlikely they would deliberately ignore a poll that showed ALP in front, considering the cost of doing the thing.
As the old saying goes, when you’ve got a choice between a conspiracy or a stuff-up, go with the stuff-up every time…
I think this point has now been established.
So are we expecting a Nielsen this week?
Adam, I think Nielsen will be coming out in 2 weeks time.
Maybe there was a problem with one or two states (ie they didn’t poll Vic or something), and so they don’t run it.
Last time Newspoll did extra polling to get out of sinc with other polls, ie do a poll last weekend and then follow up with another poll a week later as they did around budget time.
It does seem unusual for them to miss a week to get out of sinc given their previous polling habits.
If Galaxy and Nielsen had 4 weekly cycles spaced 2 weeks apart and Newspoll filled in the weeks in between we would be in the blissful position of having a poll coming out every week.
enjoyed the dialogue and the theories. here’s another: The Australian wanted to do one on the weekend after pensioners got their $500 each and the tax cuts started rolling through. Start polling on Thurs/Fri/Sat and see if you get a bounce – more to do with hip pockets than indigenous child abuse.
If Newspoll hasn’t published a poll this week so as not to clash with Nielsen, it would seem to follow that there should be a Nielsen this week. Am I missing something?
Adam, Nielsen’s are 4 weeks apart. Their last one was 18 June so their next one will be 16 July. Newspoll’s will now be 9 July and 23 July instead of today and then 16 July.
That’s more like it, Tim.
From Matt Price’s blog:
Explanation from Editor-in-Chief for absence of Newspoll:
It’s out next week. There is no conspiracy. We are trying to settle on a fortnightly cycle that doesn’t involve us coming out a day after Galaxy, as we would have had we come out this morning, or a day after Nielson. You may be aware we do not poll Thursday nights because of late night shopping. Neilson and Galaxy do, which we believe adversely affects their results. But it allows them to publish Monday while we publish Tuesday. Since we spend over $1 million a year on political polling I try to move it around so we don’t get scooped by Neilson or Galaxy. This has only become a problem recently because Fairfax is polling more regularly and the tabloids are polling monthly with Galaxy,
Cheers, Chris Mitchell
Are people prepared to put their predictions on who will win and by how many seats?