Newspoll and the Greens

The Newspoll in this morning’s Australian shows the Greens polling 6 per cent of the primary vote, down from 7.2 per cent in the 2004 election.

Newspolls this year have consistently had the Greens vote fluctuating around the 5 per cent mark, and this has resulted in commentary in the Oz and some other Murdoch papers about a supposed decline in Greens support.

Allow me to present some evidence in support of an alternative hypothesis: that Newspoll, for whatever reason, is underestimating the Greens vote.

In the past seven months, State elections have been held in Queensland, Victoria and New South Wales, and Newspoll has polled voters’ intentions in each of those elections. We can therefore compare Newspoll’s estimate of the Greens vote in the three largest Australian states with the actual Greens vote.

Queensland: Newspoll (6-7 September 2006) estimate of Greens vote in 2006 State election: 4 per cent, after fluctuating between 2 and 4 per cent in Newspolls over the previous year. Actual Greens vote in State election: 7.99 per cent.

Victoria: Newspoll (8-9 November 2006) estimate of Greens vote in 2006 State election: 8 per cent, after fluctuating between 5 and 7 per cent during the previous year. Actual Greens vote in State election: 10.04 per cent in Lower House, 10.58 per cent in Upper House.

NSW: Newspoll (21-22 March 2007) estimate of Greens vote in 2007 State election: 6 per cent, having fluctuated between 6 and 7 per cent during the previous year. Actual Greens vote in State election: 8.9 per cent in Lower House, 9.12 per cent in Upper House.

The Newspoll results at the Newspoll site carry the warning that:

Any reproduction of this material must credit both NEWSPOLL and THE AUSTRALIAN

Accordingly I hereby credit Newspoll and The Australian with the consistently and significantly inaccurate estimates of the Greens vote in the three State elections referred to. I also credit Newspoll and The Australian with persisting in a polling methodology which seriously underestimated the Greens vote in the NSW State election after it had previously underestimated the Greens vote in the Victorian and Queensland elections. It is too soon to credit Newspoll and The Australian with underestimating the Greens Federal vote, but such a possibility must be considered likely in the light of the disparity between Newspoll estimates and actual election outcomes in the most recent elections in the three largest States. And in this light one would be either a very brave or a very ideologically driven soul who would rely on the Newspoll figures as solid evidence of “declining Greens support” at Federal level, especially when the Morgan and AC Nielsen polls have the Greens currently tracking at 9 per cent and 11 per cent respectively.

Advertisements
Tagged with: , , , ,
Posted in Federal Elections, media, politics, State/Territory Elections
33 comments on “Newspoll and the Greens
  1. Chris Anderson says:

    I may be wrong but isn’t there some difference in the methodological approach newspoll uses compared to morgan and ac neilsen? If I remember correctly newspoll doesn not mention the party name, whereas the others do.

    One factor of interest is this – since the ascension of Rudd newspoll has shown the greens losing a point or two since before Rudd. I wonder is this reflected in the other polling companies and if so, why is that?

  2. QuietStorm says:

    If memory serves, the Newspoll typically tracks at least a couple of percentage points below Morgan and AC Nielsen for predictions of non-Coalition votes. As you say, Paul, perhaps the Newspollers should consider taking a closer look at their methodology.

    Oh, and nice acknowledgement. Credit where credit’s due, and all that… 😉

  3. joe2 says:

    “As you say, Paul, perhaps the Newspollers should consider taking a closer look at their methodology” says QuietStorm.

    So they should, but I wouldn’t hold my breath. Nor would I expect the now traditional election night ‘surprise’, by the usual commentators, as to how poorly the Greens have faired, to end.

    Paul Norton well done.

  4. Paul Norton says:

    One factor of interest is this – since the ascension of Rudd newspoll has shown the greens losing a point or two since before Rudd. I wonder is this reflected in the other polling companies and if so, why is that?

    Morgan Polls picked up this shift as well. I haven’t checked Nielsen yet. One possible explanation (and it is speculative) is that the change of Labor leadership caused some Labor-identifying voters who had parked with the Greens in frustration to come home.

  5. Andrew E says:

    Votes for minor parties always go up when a change of government appears likely.

    Federally, Bob Brown is even more tired than Howard. Greens appear to have momentum with them and their selection of candidates will be vital. No more tiresome campus Trots like that Senator from NSW! What about some stale old baby-boomer from the north coast who smells of bongwater and who can take the party forward after Brown gives it away, now that Garrett has shown his colours.

    How likely is it that the Greens will choose an ACT Senate candidate capable of lifting the vote to pull ahead of the Liberal? This has been a longstanding Green dream, are we actually going to see it this time or are we just going to ignore the deadline as it passes by, Kyoto-style?

    Are the Greens going to win a seat in Queensland, and thereby knock off Bartlett? Has the water crisis in SA and WA gotten to the point where they’ll elect Greens?

    The polls consistently underestimated support for Pauline Hanson. Mind you, her outfit didn’t win any seats so who cares?

  6. fred says:

    http://aussiepolitics.proboards51.com/index.cgi?board=polls&action=display&thread=1171780466

    At the site linked above some time ago I made the comment that somewhere I saw a triple graph that showed that:
    -Nielsen consistently shows the Greens around 10%
    -Morgan around 8%
    -Newspoll around 6%
    Roughly.
    I’ll try to find the original source [might have been at Mumble].
    Interesting variation.

  7. Paul Norton says:

    Some Greens results and Newspoll findings from other recent State elections:

    WA (2005): Newspoll estimate 7 per cent, actual Greens vote 7.57 for Lower House, 7.52 per cent for Upper House.

    SA (2006), Newspoll estimate 4 per cent (poll of Lower House voting intentions) fluctuating between 3 and 4 per cent for previous year, actual Greens vote 6.5 per cent for Lower House.

    Tasmania (2006), Newspoll estimate 17 per cent, actual Greens vote 16.6 per cent for Lower House.

    So outside of WA and Tasmania, Newspoll is persistently underestimating the Greens vote.

  8. Spiros says:

    “that Newspoll, for whatever reason, is underestimating the Greens vote.”

    So, what’s the reason?

    Maybe the difference between newspoll and actual polling is just sampling error, though this wouldn’t explain why Newspoll consistently underestimates the Green vote.

    Maybe voting Green is looking at internet porn. More people do it than are prepared to admit doing it to pollsters.

    The most likely explanation is that the Green vote is concentrated in a small number of inner city areas and the newspoll is spread all over the country, so they miss this concentration effect.

    “How likely is it that the Greens will choose an ACT Senate candidate capable of lifting the vote to pull ahead of the Liberal? ”

    The Liberals only need 1/3 of the vote (plus one vote), including preferences, to win their ACT Senate seat. It’s not a big ask.

  9. Fiasco da Gama says:

    Maybe voting Green is [like] looking at internet porn

    Interesting line, Spiros. Do you mean that it’s cheaply gratifying, generally demeaning to everyone involved, or that you should only do it in the privacy of the ballot box?
    Sorry Paul, for lowering the tone further.

  10. Mick Strummer says:

    My extensive knowledge of statistics and sampling leads me to conclude that the sort of minor fluctuations that are seized upon as significant by commentators are generally well within the margins of statistical error and are thus meaningless. Thus a drop from 7.2% to 6% in the greens primary vote (as measured by newspoll), in a measure that has a statistical margin of error of /- 3% is meaningless. All this shows is that if your conclusion that newspoll consistently underestimates the green vote is true, then it shows that newspoll does at least have a reliable, but not an accurate measure.
    The chance that chance alone has played a role in measuring a small phenomenon (like the intention to vote for the greens) is much greater than it is in measuring a large phenomenon (like the intention to vote for either of the major parties). And we all know that primary votes don’t mean very much anyway. IMHO, the flow of preferences (like those that saw Steve whatisname from Family First get elected with less than 2% of the primary senate vote in Victoria) is far more important in shaping the outcome of elections.
    Cheers….
    Cheers.

  11. derrida derider says:

    What mick said. This is almost certainly statistical noise.

  12. Spiros says:

    “This is almost certainly statistical noise.”

    If it was just statistical noise, Newspoll should overstimate the Green vote half the time. That it consistently underestimates it suggests the problem is more than just noise.

  13. KoopaTroopa says:

    My extensive knowledge of statistics and sampling was forgotten years ago, but shouldn’t the errors themselves be normally distributed? If there were an assortment of polls overestimating as well as underestimating the Green vote the ‘statistical noise’ theory would make more sense, although I guess you’d want a bigger sample of polls than we have covered so far.

  14. KoopaTroopa says:

    Or what Spiros said…

  15. Paul Norton says:

    What Spiros and KoopaTroopa said.

  16. observa says:

    If it’s cosistently low then it’s methodological and the answer probably lies in the way the question is worded or if that’s similar to the other pollsters then it has to be sampling error. Now it may be that Newspoll is more accurate on the majors and they sacrifice accuracy in the minors deliberately for that, remembering that increasing the sample size is costly and who’s paying for all these polls anyway?

  17. MrLefty says:

    I think most Greens expect this to be a fairly disappointing year, because there’ll be a bleed to Rudd whilst he’s new and hasn’t yet demonstrated how lefty and environmentally-concerned he’s not. 2010, though – that’ll be a great one for us.

  18. Bridie says:

    It is funny you say that Mr Lefty. Being pathetically, I mean pragmatically, realistic and all that, I think you are right. But a lot of us have been hearing, and swallowing, this forever postponing apologia for a hell of a long time now.

    Just how bad does it have to get, how craven must we be, exactly how many times and for how friggin’ long do we have to go through the same old, ever worsening cycle, before enough people wake up and say, beforehand, so far and no further will we allow this to go?

  19. Phil says:

    The Greens? Ultimately they will be a real force when they produce a slate of candidates who present to the electorate in a way that is satisfactory to them………

    Anyhoo, yes blame a bleed to Rudd, that puts off actually having policies that resonate with the public. And in 2010? I’m sure they’ll find someone to blame for another poor performance.

  20. philip travers says:

    As soon as I can afford it I may Join the Fed.Democrats,and ,frankly even if they miss out badly at the next Federal election it will not be out of being dishonest and snarky,just seen for a time as irrelevant.Thats when people like myself,excel,because under the darkness of eternal political doom,I remain alive as I am and,well Senator Bartlett doesnt lose ever.He is himself I am sure,and he is spreading himself well,and it would seem being somehow trusted,is a much better outcome,in the long run,than any political date.Personally,I think, the Greens are way to showy,a product of the rear gaurd action.It is disillusioning to not being able, financially, to fight Gunns down in Tasmania.I wouldnt wish that on anyone.And if you can afford it help Bob Brown on that one.Because I dont think that is irrevelant,nor do, I suspect, The Democrats.

  21. joe2 says:

    “Anyhoo, yes blame a bleed to Rudd, that puts off actually having policies that resonate with the public.”

    Better and far cleverer, politically, not to have clearly identifiable policies that the public do not know that they are going to have to resonate with, until after the election.

    When either of the major parties impose them.

  22. Better and far cleverer, politically, not to have clearly identifiable policies that the public do not know that they are going to have to resonate with, until after the election.

    Only if you are a major party likely to form a government, not a minor party struggling for notice.

    Since the Greens are unlikely to form a government, or even a governing coalition, presumably the best policy strategy to follow is to popularise policies that the major parties are forced to steal steal.

  23. joe2 says:

    Yep, David which is what , sometimes, happens. Even if the Greens had better numbers in the Senate the two major parties would most always see their interests as closer.

    So it goes and sad for all, in MHO.

  24. aidan says:

    Re: ACT senate seat

    At the last Federal election Kerrie Tucker (well liked former ACT Green MLA) ran, and the Greens only got 14% of the vote. As noted above, two senate candidates means it is a fait accompli that they’ll be split, one liberal one labor, unless either a candidate or a party is REALLY on the nose.

  25. Barney Maroon says:

    Slightly off topic, minor trivia on Kerrie Tucker. I was on the campaign team in her first run for senate (March 1993). We had secured all the minor parties preferences, as well as Labor’s. This meant that if the sitting Liberal got a few thousand votes under her quota (33% 1),
    Kerrie would probably get elected. Malcolm McKerras learned of this and nominated the ACT Senate seat in his Oz column as one the likliest seats to “do a Bass”, I forget the meaning, but basically: have a completely unexpected result (i.e. a Green senator). We made sure MacKerras’s (apparently the s’s is good grammar) column was the first page in the press pack we used thereafter, and it generated subsequent press. I think the sitting Liberal has cleared 33% by about 10pm, but it was damn exciting ;-). Back on topic, you don’t need to get 33% to get elected in the ACT, you just need the Liberal to get less than 33% and do some preference deals.

  26. Rebekka says:

    I think most Greens expect this to be a fairly disappointing year

    If in 2007, when people are finally waking up to climate change, when over 80% of Australians think it’s the biggest foreign policy issue facing Australia, when water is increasingly a buzz-word, we’re in the middle of a prolonged (and probably caused by climate change) drought and there has NEVER been more concern about the environment, the Greens are expecting this to be a “fairly disappointing” year, then maybe the Greens need to wake up and realise that whatever they’re doing is not cutting through to any more than 10% (approx) of the population, even though over 80% of the population is very, very concerned about climate change. Something’s not right.

    And if they can’t cut through in 2007, sorry, but they’re never going to.

  27. Paul Norton says:

    I’m not expecting this to be a “fairly disappointing year” for the Greens, although of course disappointment is very much a matter of expectations.

    A Greens vote of 10 per cent in this year’s Federal election would be very high by historic standards for a minor party in Australia, would be high by the standards of Green parties internationally (especially in the English-speaking world), and would mark the third successive Federal election in which the Greens have hived off an extra 2-3 per cent of the electorate.

    That the 80 per cent of Australians concerned about global warming doesn’t currently translate into a higher vote for the Greens is due to the fact that a majority of those people currently either disagree with Greens positions on important non-environmental issues, or think that some other party might be better able in practice to do something about issues on which they agree with Greens policies in principle. The winning of these people’s hearts and minds is clearly going to take a while yet, if it can be done at all.

  28. aidan says:

    Ah bugga! I had a nicely formatted table with the ACT Senate votes from 2004 and like a dill I closed the damn tab.

    Can’t be arsed doing it again, but the upshot was above I was referring to Canberra only, but I should have added the two electorates of Canberra and Fraser. If you do that the Libs had 35% of the first preference vote, ALP 38% and Greens 15%.

    The ACT Labor Govt is on the nose, but I’m not sure it will affect Federal voting intentions. The Greens may have a chance if the Howard Senate majority is perceived as a very bad thing. Kerrie Tucker would have to stand again for the Greens to have any hope.

  29. Rebecca says:

    aidan: Kerrie Tucker is planning to stand again this year. I don’t think it’s impossible, but it is a long shot; she didn’t come nearly as close in 2004 as the Democrats did in 1998 and 2001.

  30. Angharad says:

    I may be wrong but isn’t there some difference in the methodological approach newspoll uses compared to morgan and ac neilsen? If I remember correctly newspoll doesn not mention the party name, whereas the others do.

    I think you are right. I recall that Newspoll ask “if a federal election was held tomorrow who would you vote for: a) Labor b) Coalition c) Someone-else”. If the polled person answers “c” THEN I believe they read out a list. I’m not sure what Morgan do.

    And, as more and more people shift away from having landlines, particularly in younger demographics, you would have to think this would skew polling which is traditionally done by ringing up landlines. The pollsters must be aware of this but I haven’t seen it discussed anywhere.

  31. Paul Norton says:

    Angharad, good point about landlines. We can also consider the extent to which phone polling techniques allow for the possible distortions created by the use of silent numbers (which would be more prevalent in some demographics than others) and the greater incidence of large shared households with just one landline number for four or five people amongst younger, relatively low income demographics.

  32. Spiros says:

    “The pollsters must be aware of [the landlines issue]”

    They might be aware of it, but what they can do about it? There is no phone book of mobile numbers for them to use. Even if there was, it’s a lot more expensive to ring a mobile.

Comments are closed.

  • An error has occurred; the feed is probably down. Try again later.
%d bloggers like this: