Nov 3, 2009

Outliers Outliers – get’em while they’re hot!

Today’s Newspoll via The Oz comes in with the primaries tied at 41

Possum Comitatus — Editor of Pollytics

Possum Comitatus

Editor of Pollytics

Today’s Newspoll via The Oz comes in with the primaries tied at 41 a piece with the ALP down 7 and the Coalition up 7. This washes out into a two party preferred of 52/48 – a 7 point change from last Newspoll. The Greens are steady on 10 while the broad “Others” are unchanged on 8. This comes from a sample of 1149, giving us an MoE that maxes out around the 2.9% mark.

Is this an outlier?

Probably, and it comes from something Newspoll head honcho Martin O’Shannessy said to the Australian.

The majority of the change in the Labor primary vote is attributable to a fall in Labor’s primary among those aged over 50 under 50

The only issue over the last fortnight in media has been asylum seekers, and all the polling we’ve ever had on that issue suggests that it is the over 50’s that are the strongest supporters of tough border protection – and as yesterday’s Essential Report suggested, the over 50’s are also more likely to believe that the Coalition would do a better job at managing this issue..

If this wasn’t an outlier – we would expect the over 50’s to move strongest, so the composition of the poll is inconsistent with what we would expect to occur if the poll was, in fact, an accurate representation of the true state of public opinion.

There is also other evidence for it being an outlier. Firstly, we expect that, on average, 1 in 20 polls are out by an amount larger than the margin of error – which in Newspoll’s case is 3%. Further, we expect that, on average, 1 in 100 of all Newspoll’s will be about by a margin larger than 4%.

This is simply the probability statistics of random sampling in action.

Secondly, this issue has been in the field as a dominant news item for well over a month now. If people were changing their vote so dramatically over the asylum seeker issue, we would have expected to see the ALP vote start leeching away with the last Newspoll, the last 2 Essential Reports and the last Morgan – if not earlier.

Yet that didn’t occur.

Thirdly, all political polls in Australia over any given period of time “move together”. The various pollsters might all have their relative leans – some have a relatively higher vote estimate for Labor, some relatively lower – but they are generally consistent relative leans. Essential Report was in the field over the same period and didn’t pick up any movement at all.

Fourthly, as William Bowe at Pollbludger has pointed out, big moves in the primary vote like this are extremely rare – usually the preserve of polls taken directly after an election, or directly after a leadership change. There are a few exceptions like horror budgets and the occasional, inexplicable random poll (what we would expect), but they are rare.

Finally, this is just completely out of whack:


Only time will tell if public opinion has shifted, and by how much – but on the balance of probability, this is a dreaded outlier.

So kick back, grab some popcorn and enjoy the political show – it should be a doozy. The usual charts come in like this.

pmsatsnov opsatsnov

netsatsnov ppmnov


Leave a comment

85 thoughts on “Outliers Outliers – get’em while they’re hot!

  1. So You Think You Can Interpret Polls? – Pure Poison

    […] conclusion from recent polling is that last week’s Newspoll is an outlier – as Possum noted here and […]

  2. Possum Comitatus


    Morgan F2F and Essential should have lower variance over any given period because Essential uses a two week rolling average for their vote estimates, while Morgan generally polls over two weekends and both use large samples. Newspoll on the other hand does it over one weekend, so on the sample side alone we would expect a slightly larger variance for Newspoll. On the human behaviour side, some of the day-to-day political circus issues that people get an instant bee in their bonnet about – like some Pollie saying “X” and causing a bit of mild consternation, making a few people change their voting intention for a week – would probably get washed out over a two week period. So the the multi-week pollsters probably get a lower variance form that as well compared to the phone pollsters.

    Cud chewer went:

    [Is there a bit more volatility in that data near the larger points of inflexion.. or am I just imagining things?]

    There is – it generally happens with the other pollsters as well. It’s like public opinion takes a little while to stabilise after it suddenly moves, which makes a bit of sense.

    The net satisfaction ratings are out of whack with the vote, so if Rudd loses a few more netsat points next Newspol, on recent history a poll around the 54-55 mark would seem a fairly decent probability.


    On the protest vote – I suppose that mostly depends on what the government actually does.There’s no real consistency in history – some governments State and Federal get a backlash at their second election, some go on to get absolutely enormous thumping majorities.


    I’m a bit two bob each way on the whole conservative vote splitting thing at the Federal level with compulsory preferential voting. At the state levels where optional preferential is at play it certainly has a bit of an effect, but Federally… hmmmm

    In a three cornered contest you get a bit of leakage where primary votes go to either the Libs or the Nats, and the second preference gets given to the ALP above the other conservative party. If people deliberately allocate their preferences that way, why would they give their primary vote to the other conservative party which they didnt even preference if their favoured conservative party didnt run a candidate?

    Some Liberal voters won’t vote for a Nat in a pink fit. Similarly, some older working class Nats voters that think the Liberals are city spivs still remember Labor as a working class party and will park their vote with the ALP in a Lib vs ALP contest in the regions.

    A three cornered contest would, on average, be expected to slightly increase the size of the informal vote since there is a solid relationship between the size of the informal vote and the number of candidates on the ballot. But we’re talking fractions of a percent here – and we don’t know which party those extra informal votes would have ordinarily gone to if the additional candidate wasnt on the ballot paper.

    I think the three cornered contest allegations at the Fed level are mostly about coming up with some excuse for a poor local result.

    Queensland throws another problem into the mix in it being a volatile electorate anyway – out of the cities even more so. With the LNP merger there’s also a bit of local political animosity going around by those that disagreed with the merger. If the Nats and the Liberals go into the next election with even slightly different policies in some areas, expect to see Labor drive a wedge into it with the LNP candidates in Qld, forcing the locals to reveal whether they’re really Nat LNPs or Liberal LNPs – and causing chaos as a result!

  3. DaveM

    Poss, (or anyone else so inclined to answer)

    A question out of left field related to nothing in particular. If – hypothetically – the election turns out to be roughly as close as this, I’ve been wondering about the effect of the LNP in Qld.

    Not in terms of policy or state issues, but in a few specific seats where the last election was a 3 way contest between ALP, Libs and Nats (I’m thinking Leichhardt, Forde, and Flynn…). It was often noted that the conservative vote was “split” in these cases, allowing the ALP to win all 3 (other factors at play of course).

    Was this vote “splitting” actually relevant to the results? How strongly did preferences flow from one to the other? Secondly, would a single conservative party this time around somehow increase the Coalition’s chances of winning these seats (all other things being equal)?

  4. Venise Alstergren

    VP: Ahem, it’s Venise actually 🙂

    Really? That surprises me. Or is it seats with certainty-Higgins (Lib), or rusted on Labor seats where people who register a protest vote which is absorbed by the favourite candidate spring to mind.

    Anyhow, you have a hell of a lot more experience than me, so I accept what you say.

    My own protest vote will be to vote against the Greens because of that foul candidate they’ve just fielded.

    Anyway, time to go to bed.

  5. vp

    If I may be permitted to anticipate Poss: they rate the new government up or down a little: no protest vote.

  6. Venise Alstergren

    Poss: From your experience with polls of the past, does a government-at the end of its first term-and fronting up to an election-cop much of a protest vote?

  7. vp


    Digging is what we are here for. More power to your shovel.

  8. David Richards

    yeah vp – was just having a dig at Newspoll 😉

  9. vp

    The only thing it shows, David, is how far it is away from past recent polls, given that the other past recent polls have a much lower standard deviation.

  10. cud chewer

    Urgh, @74 above should have been posted on the next topic, it refers to the pollytrend data.

  11. cud chewer

    The other interesting point is that the trend down in net satisfaction started a month ago… entirely to do with the news media and the boat people beat up? If you’re right then there will be a correction but it’ll be back up to 55 or so.

  12. cud chewer

    Poss, just a wild speculation here.

    Is there a bit more volatility in that data near the larger points of inflexion.. or am I just imagining things?

  13. David Richards

    so Newspoll are a bunch of deviates?

  14. vp


    For the last 9 observations (ALP 2PP), including Newspoll 52, standard deviations are:
    Newspoll 2.17, Morgan F2F 1.54, Essential 1.00.

    For the last 8 observations, excluding Newspoll 52, they are:
    Newspoll 1.58, Morgan F2F 1.51, Essential 0.71

    using Excel – Data Analysis – Descriptive statistics

  15. It's Time

    It’d be less scary if you put that cane down.

  16. It's Time

    [I do more economics, statistics and social demography while William does a hell of a lot of on the ground psephology and election news.]

    Possum, I prefer to think of Pollbludger as the playground and your site as the classroom.

  17. Luckydave

    Raw Prawn, perhaps those lefites that were going to move to the Greens over this issue have already gone there previously.

  18. Cat

    I’m from Qld and I’m here to help 😛

    And thank raspberries you are Mr Possum. My sanity in these periods of national mental derangement would head the way of the mob if it were not for a tiny possum chucking buckets of water on those rolling past on the media bandwagon.

  19. autocrat

    Thanks for the link to Jones/Turnbull.

    Must go and wash.

  20. Defamed Raw Prawn

    TY Poss!

    Do you think it’s unusual that the ALP vote has bled directly to the Coalition? i.e, There were no lefties deserting the ALP for the Greens.

  21. Bogdanovist

    Next weeks Essential poll will be very interesting to see. And of the course the next Newspoll the week after….

    I wonder if Neilsen will come out of hibernation and do a poll at a time when it would get a lot of exposure?

  22. vp

    Confidence levels, precision levels and margins of error are not exactly arcane concepts. It is sad that the (il)literati of the political commentators don’t give them due respect. It would be interesting to know which of them don’t know and which choose to ignore them.

    It is good to come here for some sanity, especially in the “I’m wondering what you people are smoking in here today?” moments.

  23. Possum Comitatus


    I do feel sorry for Martin O’Shannessy. Trying to get to the bottom of this without having the next poll’s worth of data must be a huge pain in his arse at the moment.

  24. Greensborough Growler


    Maybe it is the Newspoll equivalent of a few bundles of votes on the wrong pile.

  25. Possum Comitatus

    Cuppa – they really are out to get you!

    Thankfully, tin foil hats now come in a number of fashionable designs!


    Seriously through, over the last month there’s been plenty of things in the media, but nothing like the saturation boat people receive.

    The story is cheap, easy and presses buttons – you can see why they run it.


    The over 50’s have their rusted ons and their swingers like everyone else. They also have another group that’s telling pollsters that they’d vote Labor at the moment, even though you could probably count on one hand the number of times they’ve ever actually voted for Labor in their life at any level.


    The thing I don’t get about the protest vote if it’s true is how the Greens and “Others” basically managed to escape any vote leaking in their direction at all. That’s quite a feat!. To me, that’s the most unusual thing about the entire poll – a seven point change from one major to the other while the minors are left sitting in the naughty corner wondering what all the fuss is about.

  26. Luckydave

    Hi Possum

    Agree that the poll is an outlier.

    Here’s a contrarian thought – what if the under 50s were registering a protest poll? The logic runs like this, quite a few of the under 50s (formerly pro-Labor respondents) are appalled when they hear Kevin Rudd sounding all Policy Wonk and heart hearted to that 9 year old Sri Lankan refugee pleading for his assistance. They think, gee Kev’s sounding like John Howard on this one and we just got rid of old Johnny. As a protest they say they will vote for someone else.

    Most mainstream people don’t view the Greens as credible so these respondents simply tell the pollster they would vote Liberal, the other mainstream default “alternative”, even though they won’t bring themselves to vote this way because upon reflection the Libs are worse to refugees and asylum seekers who are driving this response.

    I think Paul Howe, AWU National Secretaries comments resonated exceptionally well with a large section of the Labor leaning/voting public when he criticed Kevin Rudd on this issue. However in realpolitik, Kev has avoided the over 50s backlash, calmed things down and can now set about producing some reasonable governing to bring back the fat margins that the under 50s deliver these polls.

  27. Venise Alstergren

    Re the over fifties: Wouldn’t most of these peoples’ vote be rusted on, or set in concrete anyway? Meaning they may pad out a poll, but have little relevance because, no matter what they say they are still going to vote the way they’ve always voted. If you sort of know what I mean…?

  28. Cuppa

    [The alternative explanation for the poll not being an outlier is that asylum seekers as an issue had very little to do with it. … If it wasnt asylum seekers, then … ]

    Possum, maybe my memory’s getting faulty, but I’m scratching my head to recall any issue in recent times that’s received as much non-stop coverage, for so long, as this asylum seeker thing.

    It’s now been four weeks of news bulletins, day in, day out, breathlessly covering every little twist and development in this “saga”. WTF! The media has worked itself into a lather about this, particularly Ltd News and their ABC.

    The question is: why?? Why the disproportionate coverage given to this relatively inconsequential “issue”? I suspect the explanation involves one or more of the following:

    * to even up the polling scores, so the media will have a “contest” to report for interest value

    * the media love the Liberals and want them back in government for the BILLIONS they get from the Liberals for government advertising

    * political appointments and interference with the ABC predispose editorial content to a Liberal bias

    … Am I paranoid, or are they out to get me?

  29. Possum Comitatus

    No probs!

    I’m from Qld and I’m here to help 😛

  30. rabitoh

    I see – I hadn’t realised there was a significant ‘parked’ ALP vote in the older cohort. Cheers Possum – a very clear explanation.

  31. Possum Comitatus

    Also worth mentioning is that it’s muddy to look back at the Tampa polling because the data is quarterly, and Tampa gets thrown in with the big mover – September 11.

  32. Possum Comitatus

    Defamed Raw Prawn (what a handle!),

    It is weird. Really weird – previously when a big change in vote estimates occur and it sticks (like, say, when there’s a leadership change), the satisfaction and PPM metrics change with it. When the poll gets freaky and throws an outlier up in terms of big vote movement that washes out of the system by next poll – often those secondary metrics don’t actually change as much.

  33. Possum Comitatus

    On the Alan Jones interview, I didnt know who to feel sorry for – Turnbull for having to try and treat seriously Jones as he sounded more and more like some fringe dwelling lunatic as the interview went on, or Jones himself for engaging in such a display of public self-humiliation!

    Either way it was hilarious.When The Parrot reached that crescendo of squawking over the treaty, I think Turnbull just really wanted to job him one.

    Rabitoh and LO,

    On this bit:

    [If this wasn’t an outlier – we would expect the over 50’s to move strongest, so the composition of the poll is inconsistent with what we would expect to occur if the poll was, in fact, an accurate representation of the true state of public opinion.]

    Under Rudd, it’s the first time that Labor has held a lead in the Over 50’s and it’s also where he has his lowest satisfaction ratings and his highest dissatisfaction ratings.A good chunk of that support in the Over 50’s appears to be very tentative… almost grudging.

    In the Essential Research polling it’s usually the over 50’s that are most likely to disagree with any given government policy as well the cohort that rate the government most poorly on various issues. With that group most in favour of strong border protection and being the least tolerant cohort when it comes to boat people, if the asylum seeker issue was the driver of these Newspoll figures – then we’d expect the group most concerned with the issue to move on it.

    There isnt as big a gap between the primary vote of the three measured age cohorts as people think with the ALP. Last quarter for the Labor primary, the 18-34’s were on 45, while the 35-49’s and the over 50’s were both on 44.

    So if the poll is true – then it’s inconsistent with what we know about how asylum seeker issues play out with the public in terms of age dynamics (and assuming that it really was asylum seekers that caused the vote change).

    The alternative explanation for the poll not being an outlier is that asylum seekers as an issue had very little to do with it. The problem then becomes one of explaining “How come the minor party vote didnt change?”. If it wasnt asylum seekers, then it was something else that caused public opinion to move straight from Labor to the Coalition without any of it moving left from the ALP to the Greens.

  34. David Richards

    The Stooges are all in the Lib camp, GP.

    If you really think the tide has turned, you might be a bit disappointed when the next Newspoll comes out.

    Remember that earlier outlier that had the ALP on over 60% 2PP? That’s one way overcooked, one way undercooked, with the majority occupying somewhere around where they’ve been for 18 months or more.

    I would say it’s a fair bet that the real state of play is 56/44 to Labour. Mind you, if this gets Rudd to stop playing footsies with the racists, climate change denialists, and other assorted rightist ratbags, it might be a good thing. This imitation Howard act is past its use by date.

    If, on the other hand, he goes the other way, the Greens vote will soar.

Leave a comment

Share this article with a friend

Just fill out the fields below and we'll send your friend a link to this article along with a message from you.

Your details

Your friend's details