Possums Pollytics

Politics, elections and piffle plinking

My, What a Big Swing You Have.

Posted by Possum Comitatus on September 30, 2007

The quarterly Newspoll has been out for a few weeks now, so it obviously must be time to do a few seat projections with it like we did with earlier quarterlies.

This time we are going to use the marginal/safe government/safe ALP seat breakdowns in conjunction with the State based breakdowns to not only give us a list of seats and the margins they would be on should the quarterly Newspoll be right, but also some idea of the marginal/safe government/safe ALP vote in each State

So to start with, we’ll draw up a basic table consisting of all the data we need.

  NSW Vic Qld SA WA Total seats Newspoll Swing
Marginal 11 13 7 6 5 42 8.3
Safe Coalition 20 14 19 4 8 65 11.6
Safe ALP 17 10 2 1 2 32 7.1
Total Seats 48 37 28 11 15 139  
Newspoll Swing 9.2 11 9.1 9.4 4.4   8.8

What this shows is the number of marginal, safe Coalition and safe ALP seats in each state, the recorded Newspoll swings in those seat types (on the right) and the overall state swing recorded by Newspoll on the bottom. The 8.8% swing in the bottom right hand corner is the overall national Newspoll swing.

Note here, that I’m only using 139 seats – the seats in these 5 states (that Newspoll use for their breakdowns), however I have taken the 2 independents (Windsor and Katter) out of the mix.

Next up we need to make an assumption. We need to assume that the proportion of the national swing compared to the marginal and safe seat swings is uniform across Australia. Hence, nationally the ratio of the marginal seat swing to the national swing is 8.3/8.8 = 0.94, therefore we will make the assumption that the ratio of the marginal seat swing in NSW divided by the NSW State swing is likewise 0.94.Using this, we estimate that the marginal seat swing in NSW is 0.94 multiplied by the NSW state swing of 9.2, giving us a NSW marginal seat swing of 8.65%.Once we do that for all the states we end up with.

  NSW Vic Qld SA WA
Marginal Seat Swing 8.65 10.34 8.55 8.84 4.14
Safe gov Seat Swing 12.14 14.52 12.01 12.41 5.81
Safe ALP Seat Swing 7.45 8.91 7.37 7.61 3.56
State Swing 9.2 11 9.1 9.4 4.4

However, these swings are a little overblown. For instance, there are 48 seats in NSW. If we multiply those 48 seats by the NSW State vote of 9.2 (in the top table) we get 441.6 swing units. However, if we multiply the NSW marginal seat number by the NSW marginal seat vote, and do the same for the safe seats, we end up with 464.7 swing units. Hence the swing is slightly overcooked in the second table. So we’ll simply adjust the swings in the second table by the ratio of the swing units calculated in both tables e.g. Marginal seat swing in NSW = 8.65*(441.6/464.7) = 8.2

Doing this for all states gives us:

Swings NSW Vic Qld SA WA
Marginal 8.2 9.9 7.2 8.3 3.7
Safe gov 11.5 13.8 10.1 11.6 5.2
Safe ALP 7.1 8.5 6.2 7.1 3.2

What is interesting here is that if we calculate the swing units horizontally rather than vertically using this table, what we find is that the estimated swings in the marginal and safe government seats are actually slightly smaller than the Newspoll estimate, and the Safe ALP swing is a point or 2 higher less than a point higher than Newspoll estimated. But for our purpose here, which is to identify the seats the Newspoll quarterly suggests would change hands, our result will actually be a conservative estimate. We will essentially be underestimating the result.

[Update: calculating horizontally we end up with 8.03 vs. 8.3 for marginals, 10.8 vs. 11.6 for safe government seats and  7.2 vs. 7.1 for safe ALP seats]

Now we have our swings, we can simply apply them to the 139 seats and see what comes out.

There are 49 seats in total that would change hands according to Newspoll in these 5 states; 16 in NSW, 11 in QLD, 5 in SA, 2 in WA and an enormous 15 in Victoria (which gets back to my constant wondering about WTF is going on in Victoria over the last few weeks).

The results below show the Seat, the State its in, the ALP two party preferred result at the 2004 election (adjusted for the 2006 redistribution) and the projected Newspoll vote.

Division State ALP 2004 TPP Projected
Parramatta NSW 49.17 57.39
Wentworth NSW 47.49 55.71
Lindsay NSW 47.08 55.30
Paterson NSW 43.68 55.22
Eden-Monaro NSW 46.73 54.95
Cowper NSW 43.25 54.79
Robertson NSW 43.13 54.67
Bennelong NSW 45.87 54.09
Dobell NSW 45.16 53.38
Hughes NSW 41.45 52.99
Page NSW 44.54 52.76
Gilmore NSW 40.59 52.13
North Sydney NSW 39.96 51.50
Macarthur NSW 38.85 50.39
Warringah NSW 38.71 50.25
Greenway NSW 38.65 50.19
Bonner Qld 49.49 56.69
Moreton Qld 47.17 54.37
Herbert Qld 43.76 53.87
Longman Qld 43.25 53.36
Petrie Qld 42.55 52.66
Flynn Qld 42.28 52.39
Hinkler Qld 41.66 51.77
Blair Qld 44.31 51.51
Dickson Qld 41.11 51.22
Bowman Qld 41.1 51.21
Dawson Qld 40.01 50.12
Kingston SA 49.93 58.22
Wakefield SA 49.33 57.62
Makin SA 49.07 57.36
Sturt SA 43.2 54.84
Boothby SA 44.63 52.92
Deakin Vic 45.03 54.9
McMillan Vic 45.01 54.9
Corangamite Vic 44.68 54.5
La Trobe Vic 44.17 54.0
McEwen Vic 43.58 57.4
Gippsland Vic 42.3 56.1
Higgins Vic 41.24 55.1
Dunkley Vic 40.62 54.5
Kooyong Vic 40.42 54.3
Goldstein Vic 39.97 53.8
Menzies Vic 39.33 53.2
Flinders Vic 38.89 52.7
Casey Vic 38.65 52.5
Wannon Vic 37.63 51.5
Aston Vic 36.85 50.7
Hasluck WA 48.18 51.86
Stirling WA 47.96 51.64

For the national total, you can add the two Tassie seats and Solomon from the NT for a total of 52 seats changing hands.

Getting back to why Victoria is acting unusually, that big swing has to be going somewhere. If we look at the ALP primary vote measured quarterly, in the 4 states that are swinging big we get:


It’s become a case of the rise and rise of Victoria, starting from the introduction of Workchoices at the beginning of the second quarter 2006.


There’s been a bit of confusion over how to read this, which is my fault for not explaining it properly.

This is just like the national pendulum where a certain swing, if it were uniform, would deliver so many seats.

The only difference is that this is 15 pendulums instead of one, where each State has 3 pendulums – marginal seats, safe government seats and safe ALP seats.

Just like the national pendulum, there will be seats that will swing less than the average swing, but as a result there will be other seats that will swing more than the average swing to balance out the weight of what makes up the “average swing”.

So if you see a few seats in the list above that give you a “Holy Smokes Batman” moment , it may be that the seat in question wasn’t (according to the Newspoll data) actually swinging that much, but other seats in its category would have to be swinging more than the average to take the weight.




AddThis Social Bookmark Button

99 Responses to “My, What a Big Swing You Have.”

  1. Adam said

    I’m not sure what the purpose of this exercise is when it so obviously defies commonsense and the observations of everyone in the political process. Labor is not going to win Dawson, Warringah or Higgins. If your calculations say they are, there is something wrong with your calculations.

  2. Dave Solomon said

    Is that a baseball bat down your pants, or are you just happy to see me?

  3. PP said

    Perhaps the Victorian swing is coming from the Eastern suburbs – the mortgage belt seats, where there was a big swing against Latham and a great deal of worry about interest rates. The two things went together, in fact. Therefore, the numbers at the last seat would be artificially inflated.

    Remove Latham, remove the interest rates advantage for the government, and that artificially inflated swing is removed as well. So it’s a big swing back to Labor in seats like Aston (outer south-eastern suburbs and nominally a very safe Liberal seat at 13.2%), but less of a swing in a seat like Higgins (inner south-eastern, wealthy suburbs, Costello’s seat at 8.8%).

  4. James said

    An insightful piece, however I wonder whether you have taken into account the redistributions that have taken place since the ’04 election. For example,at the last election, Greenway went from marginal Labor to marginal Liberal and then after redistribution gained a 10% liberal margin due to rural areas being added.

    Adam, Warringah and Higgins could well fall due to the “doctor’s wives” effect. They, along with Kooyong and North Sydney are small “l” liberal seats whereas Dawson, Greenway and Mcarthur are socially conservative, and thus less likely to swing to Labor.

  5. Possum Comitatus said

    These arent “my” calculations anymore than I own Newspoll.It is just mapping what the Newspoll quarterly breakdown stated. It’s the composition of the swing in seat types(marginal vs safe) as a function of the State swings themselves that make the results in absolute seat terms so large.But there’s nothing I can do about that – it’s simply what the data produces.

    This is how the composition of the Newspoll quarterly data breakdown plays out seat by seat under the assumption of uniform ratios in the swing between marginal, safe coalition and safe ALP seats.

    Representing the data this way is no different from tracing out a national swing on the national pendulum.It’s just here we are using 15 pendulums rather than 1.There are always seats that defy the pendulum, this is no different.

    The data is what it is.

    The problem exists with the survey respondents, not the maths.

  6. Possum Comitatus said

    James – the swings were applied to the margins as they stand under the current distribution.

  7. Adam said

    Then your assumptions are wrong. SOMETHING is wrong when a supposedly logical process produces an illogical outcome. I simply do not believe that a majority of Newspoll respondents in Dawson, Goldstein or Warringah have ever said they would vote Labor.

  8. PP said

    Adam @ 7 – They don’t need to, as long as there’s a strong enough swing in other safe Coalition seats. There could be an especially large variance in those safe seats between the wealthy inner suburban(Warringah, Goldstein) and outer suburban seats.

  9. Possum Comitatus said

    Adam, then it is just like the National pendulum issue – if a majority of respondents in Dawson, Goldstein and Warringah didnt say they were voting for the ALP, then elsewhere in safe government seats even MORE people would have to be voting for the ALP than the average swing in safe government seats, simply to balance out the swing.

    Just like the pendulum works.

  10. Adam said

    But then I am required to believe that Labor will win EVEN MORE improbable seats, like Kooyong perhaps or Berowra or Wide Bay.

  11. gusface said

    well done possum

    just confirms my february prediction even further

    Good to see a factual analysis rather than the usual tea-leaf method employed by the MSM

    ps warringah will go to the wire

  12. Possum Comitatus said

    Adam, or alternatively, safe government seats like Paterson or McPherson might swing more than the average to make up for the lesser swings elsewhere.

  13. Charlie said

    That’s an unfair criticism, Adam.

    For a start, Possum is just illustrating what Newspoll is telling us. It isn’t his fault that Newspoll is saying that seats that noone believes could fall will actually do so (and, I shouldn’t need to remind you, that if there is a record-breaking win for Labor, unprecedented results should be expected). Possum is quite right to make the basic point: these massive swings have to be coming from somewhere.

    Secondly, I do not interpret Possum as saying that these 49 seats will definitely fall – he is simply giving us a useful guide to what seats Newspoll suggests are in play. At the very least it is markedly more useful than the old ‘uniform swing would result in…’ estimates.

  14. Adam said

    Well I can’t argue statistics, but I maintain my view that an analytical process which produces results that no-one actually believes has something wrong with it. Does Possum really think Labor is going to win a swag of seats now classed as “safe” for the Coalition?

    Gusface, in a straight fight with Labor’s totally unknown candidate Abbott will hold Warringah comfortably. A good independent might threaten him but none has yet appeared.

  15. Possum Comitatus said

    What I think is that unless people are telling lies to Newspoll, the average national swing in government safe seats is around the 11.6% mark, and that unless people are telling lies to Newspoll, that swing in safe government seats is slightly higher in some states and slightly lower in others (except for WA where it’s much lower as a result of their overall lower swing).

  16. bryce said

    Adam, your comment #10 doesn’t make sense. Lesser Coalition safe seats may have much bigger swings to balance it out. The balancing doesn’t have to go to even safer seats.

  17. disenfranchised Gippslander said

    PP’s comments are more specific (and perhaps closer to the truth) than mine, but here goes.
    1. Perhaps the swing in “safe coalition” could be split into Übersafe (as Glen might put it) say those on >15%, and fairly safe .15> swing>.06. If the rusted on felt they could have a protest, then this would accentuate the swing in the Über seats, and leave a smaller swing in the fairly safe.
    2. The expected “änomolous” swings wouldn’t cancel out in seats… eg, what’s the chance of missing out in Bennelong, but winning Eden-Monaro with 60%. If the seats are more or less as you say, this IMO is more likely to occur than an even change of seats
    3. Vic’s results are too good to be true, I think. Splitting up seats like this leaves wider margins of error. What if the Vic figures are overstated by 1%, while the nsw ones are understated by .75%? etc,.. what if etc. The more dissection, the more “what if”s.
    4. Katter’s comments this morning were very interesting. In the state electorate of East Gippsland we’ve had an independent member since 1999..Át the last election there was a degree of acrimony between the LIB & the Nats workers at the two small village booths I worked at.. I discreetly sympathised with the Nats, who really should be more comfortable with a “country boy” ALP leader like Bracks or Rudd than anything the Libs come up with.
    5. I know you don’t underestimate the benefit of incumbency. For example I think your estimate of the figures in Bennelong is greater than a recent poll by about 1.5%, about what ëxperts” say is the value of being the sitting Member. And of course, specially in the bush people tend to say, “good old xxx, why I shook hands with him last Saturday in Main St”. Except for Gippslanders… they’d have to have been walking down Church St, Brighton.

    Possum, thanks for providing more framework for thought. It gives me a Sunday evening not thinking about Geelong. We Vics aren’t parochial, we hate all other Vic teams except our own! Like in the English FA a few years ago: “We’re Millwall AFC! Everybody hates us! WE DON’T CARE”

  18. Adam said

    No no, Possum, I want you to tell me whether you yourself think Labor is going to win seats of the order of Kooyong, Warringah and Dawson.

  19. fred said

    Well apparently ‘senior Liberal officials’, including the Prime minister, believe ‘safe government seats’ are in danger.
    Remember the kerfluffle a week ago about Howard and co, as a result of ‘internal Liberal polling’, ‘begging’ retiring members in 6 safe seats, including Grey which requires a swing of nearly 14%, to run again? Emergency meetings, early morning phone calls?
    [Adam,you have one of the news reports on your site.]
    And possibly also concerned about ex-safe seats that don’t have retiring members.
    I dunno how close Possum is, I can but hope, but there are more than just him who suspect his numbers are right.

  20. Adam said

    Grey is a regional seat which until fairly recently was held by Labor, although its boundaries are now less favourable. With Wakelin retiring and WorkChoices hitting the blue-collar vote in the Iron Triangle towns I am prepared to believe there is a large swing brewing, though I’d still be very surprised if Labor won. All the other seats where Howard was trying to talk members out of retiring are marginals. Warringah, Kooyong and Goldstein, by contrast, are blue-riband Liberal seats in very wealthy areas, which have never elected Labor members. Two are held by high-profile Cabinet ministers, the third by a minister. Whatever you may think of Costello nationally, as a local member he is quite popular.

  21. Possum Comitatus said

    I think the margin in Dawson will be smashed and turned into a Coalition marginal.Kooyong will swing but not be won by the ALP, and Warringah will swing and not be won by the ALP without a strong independent picking up 15% of the vote (if that occurs its game on).

    But I also dont think Turnbull will lose Wentworth (unless the Green vote gets up to the 10-12% mark).But those seats above, just like Wentworth, may defy the swing, but others will take the slack.

    As happens with all pendulums.

  22. gusface said


    we are witnessing a polarisation of the electorate -not exactly party lines more semi-moralistic lines

    I come from middle of road- my wife and her family were blue ribbon till the neocon far right hegemony took control
    now she is actively canvassing friends to vote Lab-mainly because she has been lied and had her trust abused
    (think hicks all the way thru to awb and smear campaigns by Libs)

    I personally know thru friends/work etc (i am lucky cover wide range of industries 20+ yrs across gov/priv) who would never in a pink fit vote lab now basically so disenchanted with howardism that “anybody but howard” is the prevailing ethos

    Tony abbott is viewed as a good member but and this isthe rub they are all as guilty as sin and by constantly backing howard are tarring themselves with the same brush

    this is once in a generation stuff-enjoy the ride and keep the momentum going. who nows what end result we can all achieve 🙂

  23. Adam said

    OK, that’s four seats which your calculations say Labor will win, but you agree Labor will not win. Now, what about Wannon, Menzies, North Sydney, Hughes, Gilmore, Macarthur, Aston, Flynn and Hinkler, all of which appear as Labor seats on your table? Labor hasn’t won Wannon since 1954, and the old Labor vote has disappeared. Aston is safer than Kooyong or Goldstein, and has more aspirationals and fewer docrors’ wives. Menzies is stiff with self-made second-generation Italian and Greek millionaires.

  24. Possum Comitatus said

    For any given safe government seat, an argument can be made as to why it wont fall.We dont know exactly which of the safe government seats are guaranteed to topple, what we do know is that there is an average 11.6% swing against the government in its safe seats.We also know that this swing is greater in some States than others.For that to be wrong, thousands of people would have had to be telling lies to Newspoll over a 9 month period, which I simply do not believe.

    So for every safe government seat that swings less than the average amount, others in the same category will swing more, even though an argument can be made for nearly every single one of them as to why they shouldn’t fall.But these results are simply what Newspoll is saying – it might not make sense in terms of arguments why some classes of seats shouldnt fall, but for the arguments to be right would mean that the voters have been telling fibs.

    Some seats will defy the swing, others wont, others still will swing more to take the weight.Arguments about why it should be the case or not on a seat by seat basis dont actually change the Newspoll results….. they simply are what they are.

  25. fred said

    Well the SMH identified 4 of the seats where ‘senior lib officials’ ‘begged’ retiring members to stay [and just consider the implications of that as regards loyalty/betrayal,[dis]unity, financial costs re printing/corflutes etc!] as Leichardt in Qld with a swing of 10.3% required, Forrest in WA 10.5% swing needed, Forde in Qld amssive swing of 13% required and Grey with its 13.74% swing.
    Previously thought to be very safe seats.
    Perhaps they should now be considered ‘marginals’, Nick Minchin seems to think so!
    I’ve got no idea how real their fears are but it is obvious that the Libs are indeed fearful!

  26. Simon said

    Adam, there is another possibility for why these numbers are not providing a result that matches your perceptions – that is, your perceptions are wrong. I don’t mean this in a confrontational way at all.

    This is a good piece of analysis as far as I can see. And what it points to is that Rudd is not only pulling back voters who have traditionally voted for Labor but moved over to Howard, nor is he only making inroads into the swinging voters, but that the Liberal party is bleeding it’s traditional little l liberal core constituency – indicated by the greater swing in safe liberal seats that anywhere else. Ideological liberals are moving to Rudd. If this happens then there is not really a seat in the country that isn’t in play. Why?

    The electorate can more reasonably be broken into 4 sections there days. As well as the traditional left and right, there is an ever more clearly defined ‘progressive – conservative’ axis. The sense I get from the numbers is that social progressives are moving to Rudd irrespective of there economic leanings and that the economic left is returning to Labor irrespective of their social leanings. I can only account for this if there was an issue in play that unified the economic left and an issue in play that unified the social progressives. And surprise, surprise there is. WorkChoices and Climate Change respectively.

    As I see it, these 2 issues are coming together to give Rudd a core support base in [b]3 of the 4 quarters[/b] of the electorate. Leaving Howard with only the Conservative-Right quarter.

    This is of course an almost obscene oversimplification, but you get my drift.

    [Possum, I have an image that might help my argument here but don’t know who to post it]

    [And here it is]

  27. Possum Comitatus said

    Simon, mail it to me, and I’ll upload it and add it as a link into your comment

  28. David M said


    To insist that this is a predication of what may occur when the election is called defies what this site I believe is about. It smacks of the political tricks (twisting words) that our master’s use and lazy MSM use for copy, we as the general populace are heartily sick of.

    I am not a statistician but the insightful explanations gleaned here help to counter the partisan nature of this reporting in the MSM.

    I have taken what I read and test it against my knowledge and acquire new knowledge if I do not understand. For this I am pleased that Possum has challenged me to keep learning.

    Further I also apply my gut feel to it. I cannot believe that the seats you have mentioned will fall. But I do believe that there is enough dissatisfaction in the electorate to remove this government.

    Something is happening out there. There is not the feel of the “Its Time” election, nor the atmosphere of the ‘Drovers Dog’ (maybe Malcolm should have lost his pants before he called the election), but the general population including staunch liberal voters are not happy, change is about to occur

    Long live Possum, (even if the rodent returns) and may we read his views and revel in his analysis. May he, for as long as he has energy, get up the nose of the MSM, and those whom do not wish to see.

  29. carbonsink said

    ps warringah will go to the wire

    The Mad Monk was very, very, very worried in 2001. He was popping up everywhere around the elctorate that year … until Tampa/S11 changed everything.

  30. lurker said

    There’s nothing wrong with Possum’s analysis, but if result is unrealistic it’s for two reasons:

    1. The Newspoll data is out
    2. The assumption of a uniform seat swing across each state is wrong.

    2 is most likely to be wrong, but in the absence of individual seat polling, it’s the best Possum can do.

  31. Lomandra said

    Simon #26, that’s a very plausible explanation of the shifts. Thanks.

  32. dirk provin said

    My read, forgive me if I’m wrong, is that Possum’s exercise was to measure the touted swings in each state against the 2004 results and see which seats may change hands, NOT a declaration that these seats actually would.

    Some interesting seats do show up; some of which I agree with Adam are extremely unlikely to switch yet I suspect there may be a few outside these numbers that may enter calculations.

    My call from the list:
    NSW: Warringah is the real imposter on the list. There is unilikely to be the Indie challenge that could scare Abbott. The redistribution has also taken the seat across the Roseville bridge into the Nth Shore proper. Greenway’s redistribution now takes in the old semi-rural parts of Macquarie so whilst there may be a swing in the “old Greenway, it is unilikely to be sufficient. All others could plausibly be “in play”.

    QLD:Dawson looks slightly incongruous but it may not be quite as conservative as Adam has touted. I some memories of this seat running close in a couple of elections over the past 20 years. Hinkler and Dickson look the the most unlikely of the rest yet there are a few other seats outide this range that could be in play such as McPherson, Ryan and Forde and possibly Liechhardt in a 3-corner contest.

    VIC: Agree that Kooyong is a demographic “no-way” and I would also add Flinders, Wannon and probably Casey. Higgins is a mixed seat and whilst “Dog” may well be pushed close, the Liberal fortresses in the middle of the seat will save him. Similarly with Goldstein where I suspect the bayside suburbs will save Robb. Menzies looks unlikely although it will be interesting to see whether his own electorate sends Andrews a message as to his performance. Aston and to a lesser degree Dunkley look unlikely although the swing may be worth observing.

    SA & WA: No probs

  33. Charles said

    What is the definition of a safe seat?

  34. alpal said

    Possum: Is it the case that in most recent Federal elections in Australia (the last 30 years) that the actual calling of the election date by the PM of the day swings the polls to the Government of the day – in other words, the votes solidify? The conventional wisdom is that”narrowing” will occur, because it has in the past. My own view is that the polls this time could just as easily go the other way. I also think we will still be guessing election dates for at least two more weeks.

  35. Possum Comitatus said

    Dirk – spot on, this is an exercise in measuring what the Newspoll data suggests, not a sermon from the mount on which seats will change hands…. I’m not into the latter.

    Charles, Newspoll defines marginal seats as those seats held by a margin of less than 6%.

    Alpal, I only go back to the 1987 election and its lead up (December 1985 to be exact) in my data because that’s the point where I have two independent polling series, and society changes over time (so I see no real value, statistically or sociologically going back any further than that).But since the 1987 election the government has had a marginally statistically significant movement to them in total primary vote (mainly because undecideds come off the fence and moved slightly in the governments favour), but its small (2 points max) as an average, yet the TPP vote shows no statistical significance in terms of “narrowing” or otherwise.What has happened in 6 out of 7 elections since 1987 however, is that polls during the campaign have persisted with the polling trend that had been in action for the 6 month period before the campaign.Even in 2001 if you measure Tampa/S11 as an impulse response function that is the case.The odd man out was 1998, where One Nation made the polling results run all over the place.

    So saying, the polls could well move either way depending on what happens in the campaign (although short of something monumental happening, probably not move by much), but I’m in your boat in that I wouldnt be surprised if the polls widened rather than narrowed.From every actual survey measurement of the soft vote over the last month or so, the Coalition and the ALP have a soft vote of the same level.

  36. Adam said

    Of course, as a matter of logic, if there is a contradiction between Possum’s data and my beliefs, then it is just as possible that my beliefs are wrong as it is that Possum’s data are wrong, or wrongly interpreted. And no-one would be happier than me if that were the case. But I persist in my view, based on 38 years of fairly intense observation of Australia elections and practical experience of electioneering, that at least a dozen seats on Possum’s list will not change hands, and there therefore there is unlikely to be a swing in the safe Coalition seats of anything like the scale he postulates. Since there is no logical way of resolving this impasse, I will not pursue the matter further. We will see what we will see on election night. If Labor wins any one of Warringah, Dawson, Wannon and Kooyong I will be the first to nominate Possum for the Nobel Prize in Psephology.

    Simon is correct in his observation that the Liberal vote is being eroded at both ends by a combination of WorkChoices and other issues. But I doubt that when it comes to the crunch those trends will be enough to win the more improbable of the seats on Possum’s list. People in Kooyong and Warringah may be venting about Howard at the moment, but they have objective class interests which a Labor govenment will threaten, however mildly. There will be a major focus in the campaign on tax policy, and this is always THE basic issue for high income voters, and particularly high-income/high-debt voters. It will drive quite a lot of them back to the Liberal fold by election day.

  37. Possum Comitatus said

    Adam, what I’m doing here is barely an exercise in psephology, it’s an exercise in statistics.All of those seats you mention could fall, but I would have nothing to do with it.All that’s been done here is to unravel the Newspoll quarterly on a seat by seat basis.

    All it does is show what the last quarterly Newspoll decodes as on a seat by seat basis using the state swings, and the marginal/safe government/safe ALP data.That’s it.

    The swings are as the survey respondents answered.This is just the Newspoll data speaking.

    Nothing more, nothing less.To see any more into it means you are looking to hard.

  38. Adam said

    Well maybe that’s the problem, Possum. Psephological wisdom requires that statistics be interpreted in the light of political knowledge, not presented raw.

  39. Les Patterson said

    Can’t comment too much on what seems to be an excellent piece of analysis (given the starting data), even if some of the crumbs end up falling off the cake in different ways.

    But re Kooyong (and esp Adam’s comments), it might just stumble Labor’s way this time around. Kew booths have previously returned a less than 50 percent for the Libs in 1998 or 2001 (not sure which though) on a doctor’s wives backlash I suspect, and Hawthorn has grwoing numbers of students (Swinburne Uni) and a fair number of flats/younger singles who can’t afford closer to the city anymore. Both of these suburbs (making a huge and rash generalisation) will be suspicious of what seems to be the growing social conservatism in Lib ranks.

    Balwyn/Balwyn North are the really heavy Lib dominated areas of the electorate, but are also the parts of the electorate most susceptible to mortgage fluctuations(i.e. no-one can afford to buy a family home anywhere else in the electorate), and also has a sizeable Chinese/Asian community that could be warming to the Ruddster. Which leaves Camberwell/Surrey Hills which has had mixed results in the past, and which therefore might hold the key.

    Adam, Georgiou is not a minister (and never has been) and as you no doubt know is a bit of lightning rod for Liberal anti-Howardism. Although lightning doesn’t strike twice in the same place, he could actually end up as the Captain on the Titanic this time – hitting the proverbial iceberg when it wasn’t his fault (maybe I’m wrong about responsibility on the Titanic) and going down just because Vic seems to hate Howard more than the rest of the country.

    Oh, and disenfranchised Gippslander, you may not be surprised to learn that the Vic Labor governments have long been much closer to the Nats than the Libs (Libs and Nats are deffo not in coalition in Vic), and I undertsand that the Nats are the preferred party for Labor to negotiate with in the Leg Council (19 Lab + 2 Nats = a majority in the Chamber) – esp as the psychopathically led Greens are 68 percent with the Libs these days (almost 69 you could say…).

    Sorry for the mixed metaphors, and Possum, love your work, so keep it up – it gives plenty to think about, and some great ‘what ifs….’

  40. Possum Comitatus said

    Adam, this is just a Newspoll breakdown.I havent inserted any additional information here – no Crosby Textor, no leaked polling, no qualitative data, no historical comparisons, no info that I have about what’s happening on the ground in any particular seat…. for to do so would cease to make this a Newspoll breakdown!

    And that would defeat the purpose of the article.

    If you have a problem with the way survey respondents are answering Newspoll voting intention questions, I suggest you take it up with them.

    Because there is absolutely nothing I can do about it – telling me that you dont agree with the answers the survey respondents are giving all seems a bit futile.Especially after it’s already been pointed out that this is nothing but a seat by seat breakdown of what the Newspoll data stated.

    If this article consisted of me arguing that the latest quarterly was not just a poll, but the absolute truth, you’d have a point.

    But since you wont find me arguing that about any poll, let alone this particular one – I’m left wondering why your knickers are in such a knot.

    Context mate.


  41. Greensborough Growler said


    You have made a prize prat of yourself today.

    Possum’s numbers are Possum’s numbers.

    Presenting information in the the light of political insight will lead you to a bollocking of Pearson dimension.

  42. bmwofoz said

    I’m not surprised that Victoria is looking odd, first the ALP hold the marjority of seats, and the seats the Liberals hold are in many cases very large margins compared to the varies seats therefore I’m wondering would there be seats counted as safe Liberal when they generally are marginals seats like Dunkley, La Trobe and Aston come to mind.

    I read on Pollbludger the otherday someone claimed ALP polling in La Trobe was showing them ahead 56/57 this would be a 12% swing and considering La Trobe is similar to Casey and Aston could this explain the numbers.

    Also in Victoria there are two or three ALP seats where the margin is down on what would be considered normal i.e. Holt and Issacs.

    I wonder if you were to take the margins from 2001 election and applied them to the poling if this would change anything.

    I agree with Adam, Costello is very popular in Higgins, and the odds are the Liberals will hold Goldstein and Kooyong, but if the voters in this area came to feel Rudd was indeed a good conservate man and these areas have never liked Howard they may turn on him.

    I agree with Les Peterson describtion of Kooyong the same applies to Goldstein and Higgins.

    The Liberals obtained a swing to them in Balwyn in 2004 while the Hawthorn and Kew end swung towards the ALP, to show how hard Kooyong is in the 2002 Brackslide only 6 booths in the Kooyong area were won on the TPP yet the ALP didn’t look like any on the primary, (4 Hawthorn booths, 2 Kew booths) I may be out by a booth or two.

    The prahran end of Higgins voted more strongly for Costello than they did for the Liberals in the 2002 state election.

    I feel the swing is on in seats like La Trobe, Aston, Deakin, Casey with smaller swings in other blue ribbon Liberal seats.

  43. Simon said

    Adam, you simly don’t understand what is being presented to you. Possum’s analysis doesn’t say that Labor will win Dawson, Goldstein or Warringah. It says that the average swing in safe Liberal seats in NSW will be 11.5. You can then apply that as a uniform swing across all safe Liberal seats in NSW to see what would happen if, indeed, the swing across those seats was uniform.

    No one is actually saying that the swing will be uniform though. The idea, as I understand it, is to give us an idea of which seats are in play. It may well be that the swing in a specific seat will be far lower, or higher, than the average. Sadly, there is not adequate polling to come up with acurrate swings for each specific seat.

  44. Adam said

    No, because Pearson was deliberately distorting information in pursuit of a political agenda, and from a basis of profound ignorance, as Possum very competently demonstrated a few days ago. That’s not what I’m advocating (duh). I’m saying that all data, in psephology as in any other discipline, needs to be interpreted, needs to be viewed in relation to existing knowledge derived in other ways. I wouldn’t have thought that was such a radical observation. However if we are going to get unpleasant I won’t comment further on this.

  45. bmwofoz said

    i think Anthony Green the otherday wrote that its rubbish to say that some seats defy the national swing when they mostly do, the fact remains the ALP have a TPP lead of over 10% is nearly every poll conducted this year, I can see why Adam is resising the numbers but based on the feedback I’m getting from voterland the swing is on.

  46. Martin said

    I would throw Ryan, the jewell in the crown of the Qld Liberals, into the winnable bracket for Labor on the following grounds:
    – The sitting member Michael Johnson is invisible
    – Plenty of doctor’s wives, academics and assorted small l libs in the electorate are aghast at govt inaction on climate change and Johnny’s love-in with Bush
    – The diabolically unpopular $2.3 billion Goodna bypass promised by Johnny & Vaile is upsetting the horsey set in Kenmore/Brookfield/Moggill – and Rudd has promised to scrap the road if elected

  47. aa said

    Possum, newspoll could obviously spilt out the TPP for “safe Liberal seats in NSW” but don’t because the sample size would not be meaningful. This would still be the best data though wouldn’t it? (Actually that would be a sample of 11 seats, so I guess it would have the same validity as the SA numbers (unless they over sample SA)).
    Do your numbers have a similar problem for example Safe ALP seats in Queenlsand make up 2 of 28 (QLD) and 2 0f 35 (Safe ALP) does that mean the chance that these seats are not representative more likely?

  48. Possum Comitatus said

    I agree with you Adam – data needs to be interpreted according to the purpose one is using it for.

    Here, the purpose is to demonstrate what the last quarterly Newspoll was saying.

    Not whether any individual seat would fall, but simply what the newspoll data was saying.

    Beemer of Oz,

    Victoria is unique in that it has similar levels of marginals, safe government and safe ALP seats.Any of those classes of seats could take a fair bit of swing weight to balance out the overall Victorian swing.But it also allows for vast non-uniformities in the swing within seat types in Victoria to do the same. NSW and Qld are still the big guns in terms of the bulk of the safe government seat swings nationally, but Victoria swinging so much allows for really unusual movements to occur with little problem in terms of the overall swing adding up.

    Antony is right in terms of the variance – seats defy swings, but not generally by any great magnitudes(there’s always the odd man out).So saying, the current disparities in terms of the gap between the opposition and government primary vote as measured by Newspoll have never been anything like what they are now in the history of the polling series, which does the 98, 2001 and 2004 elections.Because this election, a natural ALP push election, is coming off the back of the Latham effect which knocked points off the ALP vote – maybe this election will be one of those once in a generation jobs in terms of the variance of the swings being a little larger than normal just as an artefact of a big swing running off the basis of the Latham effect.

  49. Adam said

    I agree about Ryan. It is a big ripe Qld fig waiting to fall. Johnson’s problem is not so much that he is invisible as that he is on the nose.

    [Sorry Adam, you were hiding in the spam bin]

  50. Possum Comitatus said

    I agree Martin, Johnson is in a world of shit in Ryan.His local campaign apparatus is dysfunctional, he’s got bad PR, the demographics in Ryan are changing against the Coalition, the ALP has a good candidate and the Courier Mail doesnt seem to like him.And that’s on top of what you mention which would ordinarily be enough to create problems for him.

    aa – that is a possibility.We do have a benefit though of the national swings in the three seat types acting as a rough vector for the swings in those 3 seat types for the individual states.So while in Qld the estimate for safe alp seats is a 6.2% swing and that has a large margin of error, we know that the average Qld swing is 9.1% and the national safe ALP seat swing is 7.1%.So we can anchor what’s happening in Qld against both the Qld swing (with a low MoE), the national safe ALP seat swing (with a lower MoE) and the national swing itself (with a still lower MoE).So while the numbers for any given seat type in any given state may have a theoretically large range they could be within, the national seat type swing and the State swing (both small MoE) give us a pretty tight benchmark to work with.

  51. Possum Comitatus said

    Johnson’s problem seems to be himself.How a bloke could ruin a blue ribbon seat like that makes my mind boggle.It’s really quite an extraordinary talent.It would be the same as the ALP placing Chifley in play as a result of their candidate.

  52. HarryH said

    Adam’s 38 years of political/alp experience is stopping him from seeing what is happening.

    The old rules of Lib and Lab voters are being thrown out.

    Howard won a lot of “true” Lab voters because he gave/promised them what they wanted at the time.

    Now Rudd is doing what no Lab Leader has done in history. He has made an audacious and deliberate raid on Lib heartland voters.Howards whole frontbench, bar 1 or 2, are totally on the nose to lifelong small L liberals. And in perfect timing for Lab, Rudd is everything these Lib voters are yearning for.

    This is the perfect storm that is brewing.

    Possum has presented the data. It has been so consistent all year, and seems to be widening, that it has to be believed.

    There will be shockwaves rummaging through this extreme right wing coalition on election night, and disbelief on the faces of longtime political junkies who are not reading the data and the mood correctly.

  53. disenfranchised Gippslander said

    I’ve been playing around with AEC figures these last several hours. It occurs to me that neither 2004, nor 2001 should be used as the base for calculation of swings, but 1998. I know that there’s been a hell of a lot of demographic change, as well as redistribution since then, nevertheless, a lot of “Safe lib” seats now were “marginal lib” back then. I expect Rudd will improve Beazley’s 51% by at least 4%, and golly gee, Possums “at risk” seats start to drop!. But as Adam says, let’s wait and see…

  54. disenfranchised Gippslander said

    BTW I read reports that several Manly players refused to accept medals from JWH. It might be just as well they weren’t carrying Baseball bats!

  55. disenfranchised Gippslander said

    If anyone’s listening, Newspoll says 56/44 TPP and primary 48/39. I think

  56. Martin said

    Roll on the annihilation…

    And there is a precedent for Ryan falling to Labor, Possum…if it happens it will be a very close second – in terms of sheer humiliation for the Liberals – to seeing Johnny getting trampled by Maxine – if he has the guts to run…

  57. janice said

    Thanks for your easy to understand analysis of the polls Possum. I think the Coalition is heading for disaster not only because of the big issues of Work Choices and Climate Change, but because Howard and his Government have lead the party out of their comfort zone to extreme right ideology. I would say most Liberal Party voters are small ‘l’ conservatives and they are hell bent on getting rid of the minority extreme right that have taken over their Party and led them up the creek.

  58. Johnnie (But Not That Johnnie) said

    I know nothing about statistics, and much of the to-ing and fro-ing in this thread is above my head. But I can offer two comments. 1. I know personally that Petro Georgiou is genuinely concerned that he might lose his seat. 2. Back in 1997 Michael Portillo did not even vaguely expect that he might lose his seat.

  59. Leopold said

    I agree with Adam on those seats. And I think the answer to the debate is obvious: a significant number of people currently telling Newspoll they will vote Labor are not in fact going to vote Labor.

    56-44 is not going to happen. End of.

  60. barney said

    I think Adam you are at cross purposes with Possum. You are both arguing about 2 separate things. Simon’s graph articulates simply what we all know is going on at the moment. Possum is using the newspoll data to illustrate what an outcome might be if that data occurred uniformly. It is illustrative, not predictive. it is about a “snapshot” 6, 7, 8 weeks out from polling day. There will be swings and roundabouts. Not all these seats may fall. But what Possum’s analysis illustrates is the kinds of seats that are in play based upon newspoll data and what a hole the Rodent has dug for himself.

  61. meher baba said

    The debate between Adam and Possum on this forum is fascinating for non-pseph types such as myself.

    I agree with Adam that we know that, come election night, seats like Warringah and Higgins are not going to go Labor, nor most likely North Sydney, and I agree with Possum that Wentworth is also pretty unlikely: there is a large core of voters in all of those seats who would feel that they would be at serious risk of becoming financially worse off under a Federal Labor government and who will blink on election day. This, IMO, will always override concern among the extremely well-off about Tampa, Hicks, global warming, Aborigines or whatever. These people will never vote for a trade union party en masse: they need a Ted Mack or a Peter Andren or some such to shift them away from the Coalition.

    However, at the moment the polls are showing a national average swing of 8 per cent or more to Labor. Warringah, for example, only needs 10 per cent to fall to Labor. So, if the ALP were to get 56-57 per cent of the TPP vote on election night, Warringah will certainly go to preferences. So there is absolutely nothing wrong with a seat like Warringah or Higgins being on Possum’s list.

    I think where Adam is more on track is in suspecting that we aren’t going to see a national TPP of 56-57 per cent for the ALP on election night. I reckon that reaching 53 per cent on the night will be a remarkable achievement for Rudd. Under that scenario, you won’t expect to see Anthony and Kerry talking about Warringah or Higgins on the ABC coverage: they’ll be focusing on the real action in Bass, Braddon, Eden-Monaro, Makin, Lindsay and etc.

  62. Hannibal said

    What I think Possum’s numbers are telling us is that a seat-by-seat analysis of the sitting member/candidate/demographics/primary voting history of Possum’s identified swinging seats might take us closer to the end result. I agree with much of what Adam has said, because what he is doing is looking at those identified seats and doing just that.
    Boothby for example, like many of the seats Adam has mentioned, is quite complex.
    In 2004:
    Southcott (Libs) went up 2.5% to 50.6
    Fox (ALP) went up 9.9% to 35.8
    The Dems went backwards 16.9%!
    The Greens went up 3.9% to 7.1
    FFP picked up 3%
    The end TPP result was 55.5% Lib, 44.5 ALP
    Apply Possum’s Newspoll SA marginals 8.3% to this and the result is in the high 52s.
    Then the candidates:
    Southcott has been a phantom.
    Cornes is a poor choice, but is similar to the 2004 candidate, Chloe Fox (sorry Chloe), in that she is known for being related to someone famous. Fox still picked up 9.9% in 2004.
    The Dems 2004 meltdown is key here, I suspect. Southcott’s extra 2.5% primary must have come from the Dems, and in the current climate, it seems unlikely he will be keeping any of that, and it may switch to the ALP.
    The Dems may make a small comeback this election due to the solid candidature of Craig Bossie, an ex-minister and ASIO officer, who may attract some of the old Dems back, as well as some of the religious-minded FFP voters.
    The other issue is that quite a few voters that would like to vote for the ALP will just vote Green or Dems as a protest against the ALP over Cornes, and the preferences will flow to the ALP anyway. This will probably affect the swing in Boothby.
    Demographic changes caused by the Blackwood Park development will have increased mortgage/interest rate sensitivity (if that plays as an issue during the campaign) and may also increase the FFP vote.
    I would expect preference flows to take it down to the wire. Can’t see a nearly 3% win by the ALP based on the above. I think Cornes as ALP candidate, combined with some demographic changes will cost the ALP between 1.5 and 3%.

  63. Spiros said

    The problem with arguments that say “I don’t believe the analysis because seat X has never been won by Labor” is that, if Newspoll is to be believed, Labor will get a higher vote (55%) than it has ever received (the current record being 53%). This is uncharted territory. If Labor gets what the polls say it is going to get, then it will necessarily win seats that previously were unimaginable.

    These things happen in landslides. One of the seats that will fall on Possum’s analysis is Goldstein, centred in Melbourne’s very wealthy bayside suburbs (but also containing some less wealthy suburbs).

    In the 2003 Victorian election, which was a Labor landslide, Labor won, for the first time, the upper house seat whose boundaries were (more or less) the same as Goldstein.

    It was unimaginable. But it was a landslide election, so all the old assumptions about what seats were winnable, did not apply.

    (The Labor candidate was a not high profile person at all, by the way, but he was the Labor candidate in a Labor year, and that is what mattered).

  64. Spiros said

    Sorry, that should be the 2002 Victorian election.

  65. Charles said

    Sorry I posted my initial results in the wrong place.

    I attacked the problem from a different angle. I took the ABC swing data as I don’t know what newspoll uses, took the 6% safe seat value ( thanks for the value ) and divided each state up into Lib_NP_CP, ALP and marginal sets. Worked out the average swing within each set and assumed as you did that the differences between the different types of swings across the country is constant. I used the points to fit swing lines for each state. My murray swing is 16.49 my batman swing is 6.83 ( my interest is Victoria).

    My results leave the Libs_nat with 3 seats in Victoria, Murray,Mallee and Indi.

    Noting that one of these was a national seats and noting that national seats are strongly held encouraged me to do the whole exercise. I get:

    114 Labor
    26 LIB
    8 NP/CP
    3 Ind

    The trouble with all this, you can’t extract the state by state swing differential from the available data. It may be the NSW differential is greater than Victoria etc and so on.

    Doing this for yourself is a very interesting exercise, it forces you to look at the data, however you cut the cake, if the Liberals continue to mess up they are in deep trouble.

  66. Charles said

    Sorry the result is:

    114 Labor
    24 LIB
    8 NP/CP
    3 Ind

    The CP guy got added in twice

  67. Andrew Cockburn said

    Hi Possum

    This is an interesting analysis but I think some of the speculation that has emerged on the difference between safe Government seats, marginals etc is based on a misunderstanding of probability theory, or the analysis of odds.

    The current Coalition 2PP is 52.7%. To get a 10% swing to the ALP we actually need 10/52.7, or 19% of Coalition voters to change their behaviour. Let us say 20% to make the calculations easy. Now assume that the changing voters are spread randomly among seats. We expect in very safe ALP seats (held with 70% of the vote), that the swing will just be 20% of the 30% of Coalition voters, which is a 6% swing. By contrast, in very safe Coalition seats (70% two party preferred), the swing will be 20% of 70% or a 14% swing. Hence we expect what Newspoll is revealing without invoking any special explanations. All things being equal, a large swing against the Government should reveal greater swings in safe Government seats, and lowest swings in safe Opposition seats. It is departures from this pattern that require special psephological analysis, not the pattern itself. Notably, it is only when the swing is quite large that these effects become so pronounced, but the Government does have reasons to be concerned about seats beyond the current TPP level. Now if only it were Abbott & Costello that were affected

    Andrew C

  68. Possum Comitatus said

    Hi Andrew, you are exactly spot on.

    What you have described is the very backbone of extrapolating polling results into a seat by seat basis.
    The quarterly newspolls are behaving exactly as they are expected to, with larger swings in the areas with a larger coalition vote, and smaller swings in the safe ALP seats and the marginals stuck in the middle.

    The variation in the swing within seat types is where the demographic and issue based analysis comes into play.But the basic ratios between the three seat types are pretty much what one would expect with an 8.3% national average swing.

    The ratios of the three seat types havent always been this way in the past (so people are confusing what is actually normal), and are often not anything like this pattern at this stage in the electoral cycle, but that’s purely as a result of deliberate and successful targeting of certain demographics in certain geographies by the major parties.

    In many respects, the current pattern tells us that the Coalition and ALP campaigning techniques have thus far not greatly distorted the distribution of the swing.With the swing being so big, it is the purity of it, the normality of the distribution of it, that is the greatest danger to the Coalition in terms of the magnitude of the seats it is facing annihilation in.

    I’m glad you enjoyed your play with the data – watch it though, it becomes addictive.If it gets too bad let me know, we might be able to start up a datapig anonymous support group or something 😉

    If you then frame the results you ended up with in terms of both the State swings, and the national seat type swings, you get the feeling that there probably cannot be a large discrepancy in terms of the state differentials for the national and state swings to add up, especially when you look at the number of seats in each category, and the numbers in each state.The whole thing is behaving very normally as Andrew pointed out.


    1998 would be a courageous year to use for the swing basis, as that has the One Nation effect enmeshed in the results.Now you’ve really given some food for thought!

  69. Charles said

    Yes I see what your both getting at. The vote is not coming from the minor parties and if the percentage of small l liberals is constant across the country, this is what you expect.

    The poll data supports the notion the right wing nutters have destroyed the Liberal party. On election night the result for murray will be very interesting.

  70. Eratothanes said

    My take is that the Libs are loosing theie core support base.

    Adam, I think you and many, many others will be very surprised on election night by some seats that we imagine would never come to Labor, but the numbers are saying are in play right now. Unless something reasonably extreme happens, they will be in play on election day and inevitably some of them will fall. They might not exactly match Possum’s list, but some of these seats and/or seats like them, with safe margins and demographics that – if your relying on conventional wisdom – would never go to Labor are up for grabs.

    There is a qualitative change going on in the electorate. Little l liberals are moving to Rudd and this means that NO previous election is a good indicator.

  71. disenfranchised Gippslander said

    “1998 would be a courageous year to use for the swing basis”
    Possum, I’ve always been prepared to use heroic methods to justify an Hypothesis.
    I think that the underlying problem that many, including myself, have with the polls is that we can’t believe in any cause for such a huge swing. Gough, yes; Hawke, yes; but Kevin?
    Perhaps the Hugh Mackay thesis, that the nation is waking from the bad dream of core promises, Tampa, AWB etc, explains it. In which case the small l Liberals are the ones who are really going to swing. This fits in with your calc, and maybe my heroics.

  72. Peter Fuller said

    The most plausible explanation which fits the unprecedented polling is that Howard has been able to postpone his day of reckoning by the extraordinary events of 2001 and the Latham factor of 2004. Keating had a comparable escape in 1993 thanks to the Hewson/Fightback element.
    I think Adam is understandably exhibiting something of the same disbelief that has plagued the MSM commentators for much of the year, although many now seem to have accepted the likelihood of a change of government. Perhaps it can be likened to the statement by the Geelong-supporting commentator Anthony Hudson who during the massacre on Saturday said something to the effect “I see this but I do not believe it”.
    Adam, if (and perhaps that remains a big “if”) there is a swing of upwards of 8%, and it is concentrated in Liberal safe seats, then nothing other than the roll of the electoral dice will protect Higgins, Goldstein, Menzies, Kooyong etc. Unless the actual vote is markedly different to current and recent past polls – and they’ve been unprecedentedly stable – then some of these unthinkable conservative seats will fall.
    There are plenty of precedents for such unbelievable “losses”. Cowper, Mitchell, Capricornia in 1961: Adelaide, Kingston and Kennedy in 1966, Sturt and Riverina in 1969. All Labor’s losses in Tasmania, as well as Macquarie, Perth and Tangney in 1975. Ballarat, Isaacs and Lilley in 1980;Page, North Sydney and Richmond in 1990; Hughes and Lindsay (to say nothing of Oxley) in 1996; Dobell among others in 2001.

  73. canberra boy said

    This is a pretty interesting thread. Interesting because people find it hard to believe what the data is saying. We had a similar dicussion here a couple of months ago over Possum’s analysis of the last Newspoll quarterly data, although there were fewer and less vociferous ‘objectors’.

    The objections which have been offered are most decidedly unscientific – “I don’t believe” or “I know that” seat x will not change hands. For several of the seats mentioned – for example on Sydney’s North Shore and for Wannon, I can point to personal experiences which provide anecdotal evidence which might support there being a swing. And aside from ‘knowing’ that certain seats won’t fall, we’ve also been treated to the old ‘narrowing’ argument (hey, folks, have a look at Possum’s last post but one!) and the canard that no-one ever gets 56 or 57% 2pp.

    In fact, a result of 57% 2pp would not be unprecedented for either the size of the vote or the swing (9.7%).

    Peter Brent’s list of 2pp winning margins since 1949 shows that there have been 7 results with >53%, and two with >55% – the highest being Holt in 1966 with 56.9%.

    Peter also uses Adam Carr’s pre-1949 estimates of 2pp in a table here which shows that in 1931 Lyons got 60.9% and in 1943 Chifley got 59.1%. Finally, if you scroll down to Peter’s 23 Sept post Sol’s electoral history here he points to large swings in the past: over 7% to Labor in 1929, 15% to the UAP (Lyons) in 1931, and about 9% to Labor (Chifley) in 1943.

    [Sorry CB, you were antispaminated for a bit]

  74. Goldstein does include Labor-inclined areas across the Nepean highway from Brighton and Labor did win the upper house seat that includes Brighton. Labor almost won Dawson in 1990 with a state 2PP of 50.2% and on the state level figures it has a Labor majority.But there is no Labor base or history in Kooyong and it is 23 years since Labor polled a state level majority in Warringah.

  75. Mark said

    Just catching up on the best thread seen in pollie blogland for a while (after a long weekend sans-politics). Coupla points.

    1. No-one really seems to be able to explain, as Possum so elegantly puts it, WTF is going on in Victoria. From my occasional visits the mood does seems stronger than in Sydney. Is it because your State Labor Govt is in better shape than ours? If that was much of a factor then the ALP should be struggling in NSW and it clearly ain’t.

    2. My Qld spies tell me to watch the Sunshine and Gold Coasts (specifically Fisher and McPherson) as well as Ryan.

    3. If the Howard legacy turns our to be, as it currently appears, a serious trashing of the Liberal brand then where does this leave the Nats? If they perform relatively well(and I dont think they’ll lose Hinkler or Dawson or Gippsland)the new Coalition talks will be interesting. It could be Qld writ large…..

    Thoughts anyone?

  76. Further, I think the way to bridge the gap between Adam and Possum is to look at recent state elections, here the ALP has done as well as national polls predict it might. Ryan and Dawson would be possibilities, but in Wannon state Labor could not even win South-West Coast which includes the areas of relative Labor strength in Wannon, same for Kooyong.

  77. Martin B said

    Well my comment over at Pollbludger seems to have been ignored, but I’m stubborn so I’ll post here (where I probably should have in the first place.) 🙂

    the assumption that the ratio between the three seat types measured is consistent across the country.

    Surely this is the assumption that is causing some grief?

    For every Wannon, Kooyong and and Warringah that swings less than 11%, some other safe seat in the country would have to have been swinging more than that at the time the polls were taken, unless you are suggesting that literally thousands of people have been telling lies to pollsters since February.

    I suspect that the safe goernment seats that are swinging by more than 11% are dominated by those that achieved big swings to the coalition at the last election, unlike Warringah and Kooyong (and to a lesser extent Wannon) ie seats like Aston, Bowman, Longman, Paterson, Petrie etc.

  78. Maverick said

    Canberra Boy of 74 surely 1943 was Curtin

    Geoff Robinson of 76, you are absolutely right but if you look at the resuslts of Lowen the ALP did well in Hamilton even winning 2 booths on primaries for the first time since 1954.

    One question that could be asked is in the country are people waiting to hit the federal government for inaction on water and will do the same thing to the next state governemnt

    IF Fisher retires in Mallee will the libs pick it up?

    I also predict a massive swing back to the ALP in Calwell

  79. canberra boy said

    Maverick – absolutely right, silly me: it was John Curtin in 1943.

  80. stevet said

    Martin B, I am in agreeance with you and I think you can put it down to the ‘Latham factor’. An interesting exercise is to look at a pendulum prior to the 2004 election. This makes things look pretty scary for the Coalition.

  81. stevet said

    Meher Baba, I partly disagree with your analysis. In Warringah there are no areas that tend toward Labor. It is a firmly entrenched Northern Beaches electorate.

    Whereas, Higgins is a bit schizophrenic. Sure, it has suburbs such as Toorak, but it also contains suburbs such as Prahran. For this reason, and I qualify it by saying it is a big outside chance, Higgins could fall if a tsunami is on the way.

  82. EconoMan said

    Leopold at #59… (and Meher baba at #61 too)
    Thank God someone finally stated the obvious. One of the reasons, perhaps the main reason for the discord between Possum’s list and Adam’s assessment of seat falls, is (that Adam expects) the aggregate TPP at the election won’t be 56/44.

    If you scale back the numbers to 54/46, with a concomitant reduction in the state and safe/marginal swings, the number of seats falling would drop a fair bit! Including many of the seats Adam is referring too.

  83. KC said

    Nice analysis again Poss.

    It does seem to upset some, the labor supporters who do not want to believe or promote the predicted results as it may scare some back to the libs if it is going to be a landslide and the lib supporters who do not want the view pushed that Howard and co are badly on the nose and a great majority may vote them out, this view may swing more to labor by going with the majority.

    Of course there are those who simply refuse to believe it because a result of around 57% or better has only happened 3 or 4 times in the past.

    Regardless of the views Poss is just presenting a summary of the polling results as they are presented. If you want to disbelieve the polling results that’s OK, but you are relying on a gut feel of one, yourself, instead of the polled opinions of what must be close to 100,000 people.

    Personally I have always been surprised that the Australian electorate is so polarised and that we haven’t had more landslides in the past.

    I think Gippslander is onto something in taking Beazley’s loss in 1998 when he achieved 51% as a base, then a swing of 6-8 % on that will produce the results that the polls are saying and Poss has presented.

  84. Martin B said

    Regardless of the views Poss is just presenting a summary of the polling results as they are presented. If you want to disbelieve the polling results that’s OK, but you are relying on a gut feel of one, yourself, instead of the polled opinions of what must be close to 100,000 people.

    No, there’s more to it than that.

    It’s easy to say “swings are never uniform”.
    We all know that swings are never uniform.
    We all admit that seats won’t go in pendulum order because swings are never uniform.

    It’s not unreasonable to point out that there’s good reason for thinking that seats that had smaller than average swings to the Libs last time (like Kooyong and Waringah) are lijely to have smaller than average swings to the ALP this time.

    Yes that means acknowledging that there will be seats that have *greater* than average swing to the ALP, which I did.

  85. The Doctor said

    I have always been surprised that the Australian electorate is so polarised and that we haven’t had more landslides in the past.

    I am not, mainly because of both preferential voting and reasonably fair electorates. Preferential voting gives the pollies clues as to which way the electorate is leaning and fair electorates means they have act reasonably quickly – this is the antithesis to the USA mostly runs things, with FPTP and gerrymanders of varying degree with consequent regular landslide election results.

  86. KC said


    Yeah, Canada and UK have also seen big landslides.

    But we see swings up to 17% in instances and quite common of around 14%, as you say the individual elctorates come into play, but we have also seen a electorate swing 14% one way and then a year later swing 14% back again. This seems to bely the bits about swinging voters and the general interest in politics.

    But my point about individuals opinions versus the tens of thousands polled is not on indivual seats but the size of the swing and the resultant 2PP. IE why don’t we get the big swings 10% plus on a national basis, we see swings this big over 2 elections, yet something seems to stop the big nationwide swings.

  87. The Doctor said

    two other factors in Australia may also be in play.
    1. The Senate’s usual government minority restricting radical changes coming through quickly; and
    2. It is very hard to change the Constitution, thus putting a bit of a brake on Executive power.

  88. Possum Comitatus said

    I’ve just gone over the 2001 and 2004 swings, and there are two important things to note.

    Swings have been 1.8 (2001) and 1.9(2004) times larger than the national swing in the opposition safe seats, calculated from a common sample of the 134 of the seats we’ve measured above. This compares to the current coalition safe seats swinging 1.3 times the average swing in this common sample based on the Newspoll data. So a larger swing in safe seats of the party that is having a swing against it is normal, as basic probability theory states it should be.

    Secondly, on the assumption of the ratio of the three seat types being consistent across the country – over the last two elections, the variance of that ratio for any seat type in any state has been less than 1.4% except where the number of seats is under 3 in the particular category (such as SA safe ALP seats in 2001 and again in 2004 as well as Qld safe ALP seats in 2004 – in these cases, the variance was larger than 1.4%).So the consistency of the ratio holds 27 times out of 30 within 1.4%.

    It’s an interesting debate folks, thanks.

  89. Martin B said

    Secondly, on the assumption of the ratio of the three seat types being consistent across the country – over the last two elections, the variance of that ratio for any seat type in any state has been less than 1.4%

    But surely the question that is being asked here is whether the variance for individual seats is consistent (or more precisely whether errors in mean variance are randomly ddistributed) or whether there are particular seats with demographics that mean that they consistently swing less than other seats (although obviously these demographics can change over time).

  90. Possum Comitatus said

    Martin, on the question of which individual seats will defy the category average and by how much is a question for the entrail readers and open to infinite debate.For the purposes here, all we can conclude is that the assumption of the consistency of the ratios has supporting historical evidence and isn’t an assumption without foundation.That at least provides us some evidence within which to frame the debate about the individual seats.

    So now we know that there are categories of seats which are in the shit, and roughly how much shit they’re in.So which seats need floaties and which seats need scuba gear?

  91. Martin B said

    For the purposes here, all we can conclude is that the assumption of the consistency of the ratios has supporting historical evidence and isn’t an assumption without foundation.That at least provides us some evidence within which to frame the debate about the individual seats.

    Certainly, but it’s this last step that is the tricky bit and I would be interested to see how it can be done semi-objectively. Are there seats that can be shown to be more volatile swingers than others in recent elections? Or is it just too noisy to say?

    “So now we know that there are categories of seats which are in the shit, and roughly how much shit they’re in.So which seats need floaties and which seats need scuba gear?”

    Sure. But my point above is that making arguments for individual seats is not necessarily just personal-opinion-ignoring-the-data.

  92. Philip said

    “Which seats need floaties?” Has someone been watching GetUp?

    It seems to me that the thing some people (mostly media and conservatives) are having trouble digesting is what 57% 2pp really means.

    If (AND IMPORTANTLY IF) the poll is true it means that Labor is going to win more seats than they ever have done before. Ergo they’re going to win seats that are usually conservative.

    I grew up in Higgins and it’s not all as conservative as you might think. There’s a lot of expensive schools in the area, but schools don’t vote and the attendees of the rich schools don’t all live in the area. Jeff Kennet’s old seat (Burwood) is right next door and STILL held by the ALP in the state parliament (They memorably won the by-election in 1999). It’s a Bracks-Like victory (2002 and 2006, not 1999) that’s currently being predicted by the Federal polls.

    Now if you’re saying that you think the ALP vote will soften and so the 2pp won’t be 57% and so Higgins (and North Sydney, etc) won’t fall then that’s logical. But no argument based around historical precedent is worth anything – there is no historical precedent for a 57% 2pp vote for the ALP. Historical precedent would also suggest that the Coalition could never have won control of the Senate in 2004.

    If the swing is 10% and the definition of a marginal is <6% then there are going to be some 8%, 9% and 10%
    “safe” seats which are lost. Possum’s analysis clearely shows which seats they might be. Which seats are “in play”.

  93. RobertBe said

    I suspect that the Coalition’s “Don’t risk Rudd/Economic management” argument will bite during the campaign in conservative leaning areas where there are concentrations of highly greared new money types. This is likely to mean that the national average safe Lib seat swing is a soft number overall, with wide variations. Some old money small-l “doctors wives” seats will likely swing harder than the current poll averages suggest (but not fall) whilst some mortage belt safe Lib seats swing a lot less.

    This may well give the Libs a firewall of sorts (with a few surprising holes) somewhere north of defeat but south of annialation. Meaning the overall seats won by labour could come up 10-20 short of Possum’s number of 52.

    What’s fascinating about all this is the collision between maths and emotion. As Possum has said repeatedly, the polls are what they are and 57% is what it is. But, we all, regardless or allegiance, really struggle to emotionally accept what those numbers will mean if they become real. With some in the conservative press in full on, clinically treatable, denial.

  94. Doug said

    The assumption that the election campaign will suddenly have a big impact seems to assume that there is no context against which it will work.

    We have had a context of continuous campaigning for all of 2007 and Coalition attempts to get a message across. they have continuously hit the wrong note – simply turning up the volume during a campaign is not likely to be a productive strategy if it simply results in a continousnes shrillness of tone.

  95. lurker said

    Just an an exercise, does anyone know which safe govt seats swung big time against Keating in 1996? Just to show that it is possible?

  96. Peter Fuller said

    I have spent some time trawling through Adam’s indispensable record of all(?) past elections.
    There weren’t a lot of the sort of seats you’re thinking about. Lindsay is the obvious case swing required 10.3, swing occurred 11.9. Hughes was 6.5/11.4, Bowman 8.0/8.9, Lilley 6.2/6.9.
    The other two are quite special cases Oxley (Pauline H.) 12.6/17.9, and Calare where Labor had a double figure margin, and the seat was won by the estimable independent Peter Andren (I have no idea whether that result was apparent prior to polling day).
    What was significant about 1996 was that Labor had “lost” three seats (net) through redistribution, but more significantly was defending a large number of seats on tight margins (I counted 14 under 2%, and a further 4 on margins 2-4%), which inevitably rolled over in the context of a big swing.

  97. Doug said

    Calare Had Nationals, ALP and Liberals competing with Andren on that occasion. there were rumours around that the ALP ran a low profile campaign as the best chance of keeping the seat out of the hands of the coalition.

  98. Martin B said

    Again my comment got lost at Pollbludger, so again I’ll pop it here.

    Possum, I’d like to see you apply the “same proportion of voters swing” (SPoVS) model quantifiably ie assume that in every seat eg 1 in 6 coalition voters changes their vote to the ALP.

    As discussed above, this produces a bigger swing in safe government seats than in marginal seats. In fact my BOTE calculations indicate that this model would account for well over half the difference in swing between marginal and safe seats.

    The thing to note about this model is that the size of the swing is proportional to the current coalition vote, so you introduce an error by attributing the average swing in safe seats to all seats in that category. The swing will be much less in the just-safe seats than in the ones with whopping majorities.

  99. […] by Possum Comitatus on December 12, 2007 A long time ago in a poll far, far away – the September quarterly Newspoll breakdown to be precise, some people got a bee in their bonnet about such outrageous overanalysing of the […]

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 )

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: