Possums Pollytics

Politics, elections and piffle plinking

Newspoll Tuesday…. Again

Posted by Possum Comitatus on October 29, 2007

How time flies, another Tuesday, another Newspoll.
The headline figures give primaries on 48/42 to the ALP, with the two party preferred coming in at 54/46 to the ALP.

First up; every Newspoll since October 2006 showing both Coalition and ALP primary vote estimates:


This again is just business as usual, but there is one thing that should keep even the most hysterical media pundits hyperbole in their pants about “The Narrowing”.

This Newspoll puts the minor party vote at its lowest level since 17-19 December 1999, at 10 points. Now the minor party vote has been declining, but 10% is probably a few points too low.

It has actually been the fluctuations in the minor party vote estimates that are behind most of the volatility in the TPP figures over the last 10 months.

Meanwhile, back at the much more stable monthly average figures we can now complete the October numbers and add them to the graphs.


Everything looks pretty stable there. But the same can’t be said for the minor party and others vote as it continues to decline.




Another poll, the same result. ALP primary of around 48, Coalition primary around 40 and a TPP around 56/57 to the ALP.


George has sent in the current primary votes with MoE bands:




AddThis Social Bookmark Button


36 Responses to “Newspoll Tuesday…. Again”

  1. El Nino said

    It looks like that the only ‘narrowing’ happening here is in minor party votes. I expect they are mostly going to the ALP.

  2. George said

    Hey Poss, what’s the sample size and EoM on this Newspoll result?

  3. CL de Footscray said

    Possum, thanks again for the steadying effect of statistics. It’s like they say in pilot training: ‘trust your instruments’.

  4. Possum Comitatus said

    Dunno George – we’ll have to wait for tomorrows Oz I’d imagine.

  5. George said

    No probs – just wondering if they’ll keep to the higher sample size from now till the election.

  6. […] the always informative Possum remarks: It has actually been the fluctuations in the minor party vote estimates that are behind […]

  7. Newspoll: Yawn

    According to the latest Newspoll, the narrowing is on. Or something. Because with Labor still on a primary of 48%, it’s evident that some of any genuine movement that’s being picked up is coming from elsewhere – perhaps independents or Fami…

  8. Harmless Cud Chewer said

    Possum, is it possible to see a graph of your 4 poll average? Thanks 🙂

  9. Mark said

    Sample size: 1130. MOE: +/- 3%

  10. Mark said

    Ooh, and there’s a lovely picture of $weetie boy too!


  11. Mark said

    Sorry, for primary and 2pp it’s a sample of 1707. And MOE is 2.5%

  12. kwoff.com said

    Possums Pollytics – Newspoll Tuesday….

    How time flies, another Tuesday, another Newspoll.

  13. The News Ltd press are certainly giving it a good try with the usual hysterical Shanahan overeaction in the Oz and the Tele and the Herald Sun chorusing as well. As ever, they will grasp a chance to appear infuential like the Hoff going for a beer.

  14. George said

    Sol Lebovic actually wrote some reasonable analysis for a change, not reading too much into the “movement”:

    “We should not read too much into one poll, but look at the real underlying trend over several weeks. That trend at the moment is that not a lot is changing and Labor remains in a very comfortable position”


  15. PaulC said

    Now that Newspoll is on weekly, and doing larger samples, they would actually have the potential to break it up by state, age etc a lot more often than quarterly – they’ve surveyed nearly 3500 in the last two weeks, that would probably at least give a big enough sample for say NSW and Vic. I wonder if they will.

  16. codger said

    It’s just the ‘Kellying’ Poss as announced on insiders and picked up by cactus tip yesterday. Much ado about MOE.

  17. Sol’s piece is very measured. Almost as if he had a background in polling. Who knew?

  18. GS said

    never mind the quality guv’nor, feel the trend

  19. […] week, so it’s even Stephvens right now!). Sol Lebovic is giving rationality a try this week. Possum Comitatus notes that a lot of the movement is attributable to changes in the minor party performance across […]

  20. BlueSkyMining said

    It makes last week’s Newspoll look like a statistical outlier, which as pointed out above makes it look like “more of the same”. Boy it was fun at the time though! Is this one the same on the other side of the equation? Looks like it.

    However both last week’s and this week’s Newspoll had relatively large samples, each with an MoE of +-2.5%. So if our theory (apologies Possum, by appropriation of course) of a flat-earth…whoops…flat-line poll “trend” is true then doesn’t the latest 2 Newspolls indicate the real figure is (Labor) 56+-0.5% TPP?

    Or is this stretching the statistical significance of the numbers too far, given the variables (e.g. preference allocation scheme)?

  21. PaulC said

    Actually on the subject of statistics – if the margin of error is 2.5%, but the percentages are rounded to the nearest whole number, then isn’t the margin of error effectively 3%? After all the poll could have had the ALP vote at anywhere between 53.50 and 54.49%.

  22. David Gould said

    I wonder why the Green vote has consistenly averaged two to three points lower than at the election in 2001?

    I would have thought that this election they could do better than at the last election.

  23. Peregrine said

    I wonder if Newspoll can tell us whether the interest rate questions attached to the poll were asked before the voting intention.

    The Greens vote is always going to show up as lower nationally with such a high Labor primary vote as many of the 2004 Green voters were disenchanted Labor voters who have gone back to a competitive Rudd. It may end up being still pretty high in some seats at the election so Labor’s not likely to end up with 48%.

  24. George said

    Agree Peregrine on the line of questions. I find these variations from poll to poll (within the same company) totally distracting, and the consipracy theorist in me says they’re doing it on purpose to have something to “analyze”. As a previous research scientist, experiment methodology stretegies were set in concrete. Do it many times, reduce your margin of error, have postive/negative controls where possible and minimize any variations from experiment to experiment. Pitty no one runs polls this way 😉

  25. happy chap from Griffith said

    anyone know where I can live steam the costello-swan debate?

  26. happy chap from Griffith said

    The location where I can live ‘STREAM’ the debate would also be useful haha

  27. Paul said

    Wonder if Newspoll toggles between “safe seats” and “marginal seats” for consecutive polls – might explain the changes (ie, the current poll was taken predominantly in safe Coalition seats).

  28. Possum Comitatus said

    Cud Chewer at 8 (love that name 🙂 ),

    I…..er…., arm. I was going to run an ongoing graph of that but I sort of wrote over the file by accident, so I’ve had to start again.

    BlueSky at 20 – probably something about that. There’s always going to be a 1 to 2% of newsvoters at any given time changing their mind like they change their duds on the basis of the latest fashionable headline, but over time they sort of cancel each other out. So with those blokes taken into account, the ALP primary vote looks like its probably wobbling around that 47/48 mark and the TPP doing a waltz around the 54-57.

    But it’s the upside spikes every couple of weeks with the ALP that get me, and without any downside spikes. It’s like every so often, Newspoll hits a pocket of true believers that then head back into the underbrush, not to be seen for another 6 weeks. And the fact that the spikes are always on the upside hints at the possibility that there’s something more too it than noise.

    Paul C, if you start rounding, the MoE changes. There’s quite a lot of rounding issues in the polls as they are published, and it sometimes gives us some funny TPP figures that look to be out of whack by couple of points considering the primaries they are running off. The last ACN telephone poll was like that.

    Paul, I would imagine that Newspoll tries to keep as random a sample as possible when it comes to these things. If they are polling weekly now, they will naturally pick up a good mix of safe government, marginal and safe ALP seats anyway.

  29. Bruce said

    I was rung by Newspoll this afternoon. After checking whether I was a male eligible voter, they asked if we live in the electorate of Parramatta or Greenway. We live in Greenway – now safe Liberal on about 10.5% swing since a recent redistribution (formerly marginal). The lady said they were seeking voters in the Parramatta Electorate and that was the end of that!

    Damn! I wanted to hear the questions!

  30. Bruce said

    Re 28 Possum.

    One way to check whether there is evidence that the volatility there is more than just sampling noise is to do a statistical test on the variance of the polls since March. If there is more volatility than would be expected (using on a p=0.05 test) based on the calculated means of the Labor primary and 2PP preferred vote you could conclude that this is evidence for genuine underlying volatility, although there is clearly no evidence of a trend.

    I don’t have the raw figures, so I can’t do it myself. Also the rounding to the nearest percentage may introduce some bias in this test.

  31. canberra boy said

    An additional issue which affects margins of error, and for which the pollsters don’t ever report details, is the weighting they use when they discover that their random sample does not reflect the demographics of the overall population they are sampling. Simon Jackman has covered this in two interesting posts, most recently on Sunday and in a more detailed explanation of how weighting works and the effect on error back in June.

  32. Nicko said

    Hi Possum

    I posted the following at pollbludger but on reflection is probably suited here. it’s a bit long but I do have a special interest in Aston even though I am no longer enrolled there. By the way your stats do my head in (it goes to show how much different a finance/eco/politics degree is to a maths/stats based one) but enjoy trying to decipher your hypotheses….anyway this is what I posted at wiliams site

    “never would have thought about Goldstein in a pink fit until those weird polls of the last 3 months.I still dont think Goldstein will fall, but there does seem to be an awful lot of polling going on there, and Aston by at east two companies.”

    possum I have been banging on about Aston to my (non political) wife for ages as we lived there until May 2001 till just before Peter Nugent died (we voted in Aston at the 1998 election). I voted ALP but would have voted Liberal as a mark of respect of a popular local member at the by-election given the general election was only a “few” months away when I would have reverted my vote to the ALP.

    Anyway enough about history, do the bookies run a book on a seat that will swing the most as this seat is prime to in Victoria. I just read the ABC site that said the Lysterfield Primary booth had a 62.3% vote. We lived just the other side of Napolean Rd in 1994 – early 2001 and this the archetypal MORTGAGE belt territory (if you didn’t guess by the capitals) of the Melbourne outer east. As a further history lesson (sorry) when we built out house there in 94 our block backed onto the Scoresby drainway and we looked out at cows in the back paddock. Those paddocks became houses in 1999 – 2001 just before we left. The interest rate fear campaign was made for this electorate.

    If there is a swing on and the RBA is not biased and raises rates on Wednesday morning, then this seat is going to swing by plenty, the only question is will there be enough to unseat Pearce?

    If you look at the seat geographically then the top end of it (Vermont, Wantirna, Scoresby is balanced out by Bayswater, Knoxfield, Studield and Mountain Gate). How PO’d is Rowville and Lysterfield over the “record low interest rate” lie marketing of the 2001 election?……this seat is one to watch!

    for enjoyment sake, after watching downer go down this will be my next “drinking” seat watch on the 24th

    added for here…are there any poll numbers for Aston that says something big might happen?

  33. Rod Hagen said

    Possum writes:

    “But it’s the upside spikes every couple of weeks with the ALP that get me, and without any downside spikes. It’s like every so often, Newspoll hits a pocket of true believers that then head back into the underbrush, not to be seen for another 6 weeks. And the fact that the spikes are always on the upside hints at the possibility that there’s something more too it than noise.”

    Looking at George’s graph it is interesting that the poll following a positive “spike” for the ALP is always lower for them than the result in the second poll after the spike. In other words , they fly up high, drop back down and then rise a little again. This, of course, tends to maximise the “swing back to the coalition” media headline rate change in the week following the spike.

    This could, of course, be simply due to chance. It could also be due to some sort of cyclical variation in voter behaviour. The cynic in me says it could also be due to some form of cyclical variation in the actual polling process or the subsequent number crunching / weighting etc.

  34. Rod Hagen said

    Intriguing to see the “re-analysis” of the latest two newspolls in the Australian today – http://www.theaustralian.news.com.au/files/newspoll-1nov.pdf – and the commentary from Shanahan in the Oz at http://www.theaustralian.news.com.au/story/0,25197,22682792-601,00.html.

    Note that the side bar says “The data has been weighted to reflect the population distribution. This latest
    survey is based on a reanalysis of surveys conducted on October 19-21 and 26-28, 2007”.

    Does this imply perhaps that , if the results for the two latest polls were shown in dis-aggregated form that the results may look a little different to those shown in the headlines yesterday and the week before?


  35. Rod Hagen said

    (Sorry for the series of posts)

    Another thing that strikes me is that given the increased differential in anticipated voting behaviour based on age in this election, comparatively small changes in the age base of the sample on any given poll could significantly change the result. I wonder if they use a ‘weighting” system which in part depends on the age breakdown too.

  36. marrickville mauler said

    Further MSM recognition Possum in the Telegraph today: “Possum trashes priceless relics” – http://www.news.com.au/dailytelegraph/story/0,22049,22690473-5001028,00.html . While most of the metaphors were a bit obscure for me, this line was right on the money: “For a really small possum it did a lot of damage.”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: