The Polls before E-Day
Posted by Possum Comitatus on November 23, 2007
First up we have ACNielsen – with not one, but two polls.
First up is the telephone poll.
The headline figure has the ALP up 3 to 57, the Coalition down 3 to 43. This gives us the following two party preferred series:
On the primaries, it’s ALP up 1 to 48, the Coalition down 3 to 40, giving us the following series (thanks again to George P):
The poll had a sample of 2071 giving an MoE of a tad over 2%.
Not content with one poll, Mr Sirton in a feat of heroic polling also produced an online poll producing the same headline results of 57/43, but with primaries running 46/38 to the ALP. That one had 1421 respondents for an MoE of aound 2.6%.
Next up we have a Morgan poll of 1648 people and an MoE of around 2.5% giving up a headline TPP figure of 54.5/45.5 to the ALP, and primaries running to the ALP at 44.5/40.5.
Finally we have Galaxy suggesting a TPP of 52/48 to the ALP with primaries stuch at 42.5 for both major parties.
Someone is going to look like a dill on Sunday.
When we combine all this into our rolling 4 pollster average, we end up with the following over November.
The seats are the number of seats a uniform swing would have produced for the ALP.
91 is a very popular number, so keeping with the theme, here’s a rather popular ditty from 1991 that seems all rather appropriate:
Possums Pollytics 
kwoff.com said
The Polls before E-Day « Possums Pollytics
The poll summary for the day before E-Day
Bruce said
The Age is reporting the Galaxy Poll as having a sample size of 1200 – MOE about 3.0% and a 2PP of 52. This is not inconsistent with a population 2PP of 54.5% to 55.5%.
The combined ACN MOE on 3492 voters (2PP 57%) has an MOE of about 1.7%.
Bearing in mind that the estimates involve rounding (ACN could be 56.5%), the polls are not inconsistent.
Also-ran said
Hey Poss – Galaxy seems way out of sync. It’s done good things before in other elections. Wonder why. I guess all will be revealed by 7pm Saturday.
Gezza said
Poss,
Adel. Uni. polling in Boothby, says extremely close 2PP, though primary low for Labor independents scoring well as expected, but preferences going anti-Lib. Details reported here on Local ABC 891. Public forum on radio now, Cornes a no show, but audience & calls very anti-Lib. Will get back to you with precise info. ASAP.
Gezza
KaptainKrumpet said
Oh no, its’s a WIDENING. A new buzzwored for the punditry.
Kevin Brady said
Hey Poss:
Are you going to do an analysis of different pollsters predictions come Monday? That would be very interesting reading!!
wantok said
Since the Canberra Times, though a pretty good dead-tree product, utterly fails to comprehend the online world (“of course we couldn’t put more than three stories a day online – we wouldn’t sell any papers!”), I feel entirely justified in posting today’s Possum article for those with no access to the dead-tree edition:
(page 9, two column feature in the election spread, complete with our favourite possum pic)
“Playing possum with polls and politics”
Jenna Price
Possum Comitatus is the nom de plume of a 33-year-old Queensland economist, and he has become the cult figure of election analysis this year; the star of the political blogweb with links from every major electoral analysis site in Australia.
His prediction? Labor will win 94 seats if there’s a uniform swing but is more likely to get 89. (wantok: that’s a bit out of date now isn’t it Poss?)
Possum only started full time in September this year with his website, Possum’s Pollytics, examining each poll and every statistic. This week, on the issuing of the NewsPoll, he wrote, “The ALP two-party-preferred vote is currently at its long-term, pre-Rudd trend.”
But he resisted the temptation to mention dogs or drovers.
Possum only recently graduated in economics and specialises in econometrics by training and profession, including polling data. (wantok: ooh, what a giveaway!)
“[But] I must say analysing politics and being able to have a bit of a laugh in the process of looking a bit deeper into the empirical reality of things is a far more interesting thing to do on a day-to-day basis.”
Possum’s Pollytics came out of the blue. He’d been involved in forums and posted the occasional chart but soon people were responding.
“I started getting more comments and more traffic and within a few weeks, I found that I’d become the accidental blogger,” he says.
The editor of Crikey, Jonathan Green, spotted him and offered him a regular gig.
Green said, “He’s particularly perceptive, highly literate and can present complex information in a lucid and concise way… he’s got a dab hand with a telling graph.”
Not all are so impressed. The veteran election commentator, Malcolm Mackerras says he won’t put Possum on his favourites because he doesn’t know who he is.
“I can’t correspond with someone who is anonymous,” he says. (wantok – Er, yes you can, Malcolm.)
The anonymity may not last long. Possum, of northern bayside Brisbane, says there have been a couple of journalists sniffing around, so for the record, he went to Wingham Primary, in northern NSW, then Wingham High; will marry his girlfriend of 14 years in April; plays beach volleyball and is hoping to make a living doing what Possum does – which may require decloaking.
But the people who matter know who he is – a news junkie who curls up at night with The Financial Review. He talks to people about numbers and figures all day and usually uses his real name.
A blogger at election website Poll Bludger, William Bowe, links to Possum because he says the quality of the work is outstanding. Where’s the rivalry?
“I’m a terrible person so it is a measure of his outstandingness that I am able to put my misanthropy to one side,” Mr Bowe said.
Even the academics are now citing him. The newly appointed professor of journalism at Monash University and director of the Australian Centre for Independent Journalism, Chris Nash, says Possum has forced journalists to examine the way they use statistics.
“He has become an essential point of reference for any journalist writing about the polls,” he said.
PASOK said
Antony Green was on ABC local (Brisbane) earlier today refuting Galaxy & Newspoll’s assertions (earlier on the same station) that the swing was not consistent across the seats.
Galaxy seem to be getting in a bad habit of polling at exactly the wrong time. I remember last week’s polling being done IN BETWEEN party launches. Now their “narrowing/cliffhanger” poll is released before the current Cyclone Lindsay hit.
I tend to believe that the Nielsen results show some sort of reaction to this latest fiasco.
Possum Comitatus said
I also better add for the benefit of those that think they’ve known me for years on various other forums. You did know me, but you also knew three other people that made up the group character of Possum Comitatus. It wasnt until the group character disbanded in April that I took the character on myself and accidentally ended up a blogger. The other three involved in the group character were all economists, two in Brisbane and one in Adelaide.
So for those that knew me before the blog – I was but one of four, for those that know me just with the blog, I’m now the only person behind Possum.
nomad3 said
hey Pasok.. did you see Antony Green on Lateline a couple of nights ago .. he seemed to be jumping out of his skin … oh btw : you support papandreou ?
Kirribilli Removals said
Onya Possum, you’ve given hope and understanding to people, and we are so vwery vwery impressed, and also grateful for an antidote to the MSM and their ilk.
PASOK said
Oh, and I have a song for tomorrow:
Forgotten Years
Midnight Oils.
“…contracts torn at the edges, old signatures stained with tears”
“…the desperate and divided years…”
How did Peter Garrett know about John “Wedgie” Howard and WorkChoices in 1990?
Kate Ellis for PM said
I am disappointed- I thought you were a female Possum and you turned out to be a 33 year old guy! Oh well, still love your analyses and blog, the best there is out there. I have had a lot of fun reading your comments- so thank you again!!
mb said
Hope I’m not breaking anybody’s copyright with this, but today’s SMH has the following (fairly obvious) comment about how the parties’ internal polling compares to the out-of-this-Galaxy poll results:
“A Galaxy Poll to be published in News Limited newspapers today shows the election to be a tight contest, with the two main parties tied on 42.5 per cent of the primary vote and Labor leading by 52 per cent to 48 per cent on a two-party-preferred basis.
But sources from both main parties believed that the Herald poll more accurately reflected what their internal polling had found.”
Martin said
Labor’s Yvette D’ath has just supplanted Teresa Gambaro as Centrebet favourite in Petrie – Possum’s neighbouring electorate, judging by the Canberra Times piece
Ron Brown said
POSS
I RANG GARY MORGAN 1 hour ago
re his press release saying a 5.2% swing to Labor on his 22 marginals & said his 51.5% to 48.5% “marginal poll” result meant the swing is not 5.2% Before I could say mathematically the swing is 4.2% he jumps in and says you’re right the swing is 3.2% ! I said no , mathematically compared to Labor’s 47.26 in 2004 your marginal pill is a 4.2% swing.He says you are right -I’ll change the web site straight away !!
Now a QUESTION
1/ Has Morgan’s 22 marginal seat poll have ANY realistic value whatsoever ???
2/ is the Labor 51.5% to 48.5% in the 22 marginals as a TOTAL % have any value
3/ IF no , is your reason that the sample per seat is too small ??????
Stunkrat said
Why the anonymity, Poss?
Neilbris said
Geez Possum. We were all hoping for your usual eloquent, intelligent, insightful and fearsome destruction of Galaxy’s methodology (and a big tick for Neilsenn of course).
wantok said
Based on this rather strange story, it sounds like tomorrow’s Newspoll will show a tightened 2PP:
http://www.abc.net.au/news/stories/2007/11/23/2099148.htm
Pretty brave call, isn’t it? No result for a week? Sounds like deep-fried guano to me.
Possum Comitatus said
Neilbris – I think it’s gotten to the point where Galaxy is struggling to warrant any attention at all.
Martin said
re post #19: I suspect any side desparate enough to question an election outcome before almost all of the votes have been cast – let alone tallied – knows the game’s up
PaulC said
Funny how the reporter writes it is certain to be a cliffhanger, but the quote given is:
“I think we might find it very very close in those key marginal seats that we have to see and there may be a lot of recounts and protests around the edges,” he said.
That isn’t certainty, that’s saying ‘I think X might happen.
Cat said
Poss yesterday I said that Bogan-gate would be neutral because the dog-whistle would actually reach some of the morons to whom it appeals but also anger those more rational who would like some pride back in their country. I do dismiss Galaxy – partly because of your analysis and partly because I have heard enough of Briggs to know which side he is batting for. Now Shannessy is out saying too close to call – take a week to get a result is the outcome of the last Newspoll which takes in the post Bogangate response. Have I underestimated (or misunderestimated to use a Bushism) the number of people out there who respond to the dog-whistle. What does the Crosby Textor analysis about moronic racists and how they compare their right to hate to their right to pay their mortgages for example?
BlueSkyMining said
Hey Poss, interesting partial de-cloak. Thanks to wantok for the posting of the article. Not that it matters much, as it is the content that matters to all of us not the wrapping 🙂
Well on the basis of the Lindsay silliness yesterday morning I was inspired to lay a couple of modest bets on Sportingbet. $100 on >=56% ALP TPP @$11 covered (exactly) by $50 on >=54% & =56% ALP TPP is in to $9 from $11 yesterday. Looks like more people believe Nielson than Galaxy!
Cat said
I try that again.
What does the Crosby Textor analysis say about moronic racists section of the community and whether they value their right to hate over their right to pay their mortgages for example?
BlueSkyMining said
(oops, tryin again, forgot my HTML amps and gt/lt’s)
Hey Poss, interesting partial de-cloak. Thanks to wantok for the posting of the article. Not that it matters much, as it is the content that matters to all of us not the wrapping 🙂
Well on the basis of the Lindsay silliness yesterday morning I was inspired to lay a couple of modest bets on Sportingbet. $100 on >=56% ALP TPP @$11 covered (exactly) by $50 on >=54% and <56% TPP @$3. So if AC Nielson is close to right I am of to buy a (small) plasma TV on Johnny Rotten’s last stand (only because we decided to invest my new daughter’s baby bonus in a trust to help pay for her future education because bloody Howard has driven up the cost of education).
PS: Oh, I just checked Sportingbet again and >=56% ALP TPP is in to $9 from $11 yesterday. Looks like more people believe Nielson than Galaxy!
BlueSkyMining said
(oops, trying again, forgot my HTML amps and gt/lt’s)
Hey Poss, interesting partial de-cloak. Thanks to wantok for the posting of the article. Not that it matters much, as it is the content that matters to all of us not the wrapping 🙂
Well on the basis of the Lindsay silliness yesterday morning I was inspired to lay a couple of modest bets on Sportingbet. $100 on >=56% ALP TPP @$11 covered (exactly) by $50 on >=54% and <56% TPP @$3. So if AC Nielson is close to right I am of to buy a (small) plasma TV on Johnny Rotten’s last stand (only because we decided to invest my new daughter’s baby bonus in a trust to help pay for her future education because bloody Howard has driven up the cost of education).
PS: Oh, I just checked Sportingbet again and >=56% ALP TPP is in to $9 from $11 yesterday. Looks like more people believe Nielson than Galaxy!
Possum Comitatus said
Cat, the Oztrack Crosby Textor data doesnt actually have anything in it about dog whistling – so it’s hard to say.
Stephen D said
Damn… A 33 year old male economics graduate, huh?
Well there goes my theory that you were Hyacinth’s alter-ego.
Jess said
http://www.theage.com.au/news/federal-election-2007-news/bookies-tip-labor-win/2007/11/23/1195753272917.html
If Galaxy is right some people are going to lose a lot of money!
Ratsak said
Possum,
Don’t blow the cover. All superheros need to keep their secret identities.
Do you or anyone else have the ACN primary results for the pre election polls going back before 96. Someone on PB posted the breakdowns for 96-04 and it showed a consistent over estimation of 2-3% for the incumbent at each election (Lab – Lib – Lib – Lib). They got the opposition primary as close as you can expect ( <+/-1% ) each time, but the incumbent was always out 2-3% and always overestimated. An interesting series to my less than statistically adept brain. Interested to know if this goes back further. I can’t imagine it being right for 93 for obvious reasons, but maybe there is something in there being an ‘incumbency bias’ in ACN.
Cat said
Cat, the Oztrack Crosby Textor data doesnt actually have anything in it about dog whistling – so it’s hard to say.
Comment by Possum Comitatus — November 23, 2007 @ 12:08 pm
Coalition advice that does not cover dog-whistling. It is a brave new world folks.
One last question – are you likely to have any preview of the newspoll tonight? I only ask because I have to get up way too early just to get to the first booth I am working to check you out tomorrow.
Oh and I am for keeping your cover too because everytime I meet a 30 something stats man from Brisbane with a tail I will look at him and wonder could it possibly be Possum?
happy chap from Griffith said
folks the Oz says they’re releasing the latest newspoll results on their site at 4pm (“LATEST NEWSPOLL RELEASED HERE AT 4PM” http://www.theaustralian.news.com.au/).
Tom said
Heh, I had figures Poss for a marsupial of the female persuasion also 🙂
Jess said
Poss or anyone – do we have more detail on tomorrows newspoll figures and is Martin O’Shanessy just doing his job for his masters of getting more people to buy the GG by talking up the closeness or http://www.abc.net.au/news/stories/2007/11/23/2099148.htm
or could there actually be a “narrowing”?
Also Poss please do an analysis of which pollster has been more accurate over the past say 4 elections?
happy chap from Griffith said
I find it pretty difficult to believe that Newspoll can be saying it’s ‘too close to call’. I am up in QLD and I really do get the feel that the swing is on (for god sake the courier mail jumped on the Ruddtrain!!).
Jess said
Poss have you updated your seat prediction?
The Intellectual Bogan said
Well, I always had Possum tagged as male, but a bit older than 33. I must be getting old.
Caravel in Canberra said
Hello Possum
For this site, your comments, analyses in plain English, humour and wit, I thank you, danke schön,and tänan vaga(Estonian).
denise said
have you all read simon jackson on the abc election site.
can you explain possum is this your method as well.
jim said
So… Newspoll says “cliffhanger”… someone debunk this result for the sake of my nerves…
Possum Comitatus said
Jess – all the pollsters are roughly the same on predicting primary votes over the last 4 elections.The final election results have all generally been within the margin of error of the final polls on primary votes. Some pollsters get their preferences wrong, but each pollster that doesnt use a second preference question changes the way they distribute preferences after each election – so history is no real guide on who has the best TPP vote. At the end of the day, the pollster that gets closest usually occurs by nothing more than chance.
My seat prediction is still the same as it’s been for months, 89 seats off the back of a 55.15% TPP (which is what the model said). That’s less seats than a uniform swing, but it’s because I think the seats where big swings will occur will probably not have enough to push them over the line into the ALP camp.
But we’ll end up with a dozen or so Coalition marginals on or under a 2% margin.
Denise – The brilliant Mr Jackman does things differently to me, he uses simulation whereas I use just polling and demographics.
Jess said
Thanks Poss – you’ve done it again!
denise said
thanks poss i am sane again
paul said
alot of egg will need wiping from faces in the near future, how can this possibly be close?
Geoffrey said
The ABC Adelaide radio announced a Boothby poll as mentioned earlier AND also they did a senate poll of Boothby that gave Mr X 29% taking it manily from the 2 big parties – this will be so interesting for SA outcomes as he splits any excess between Family First and the Greens. I don’t know if this is a state wide trend or not.
wantok said
Shanahan in the Oz: Newspoll 52:48. No other details at all except a 2614 sample size.
Exasperated Calculator » Blog Archive » Election07: Penultimate Edition said
[…] It’s poll-a-palooza today. Newspoll ALP 52, Galaxy ALP 52, Morgan phone ALP 54.5, AC Nielsen phone ALP 57, AC Nielsen online ALP 57. Someone’s going to look pretty dumb in about 24 […]
David said
What I cant understand is:
The Murdoch polling is obviously skewed somewhat to Labor leading just marginally. (Compare this to Morgan and Neilsens’ far larger lead).
But Murdoch has endorsed Labor in its editorials, notably The Australian. And they rarely do that.
Whats it tryin to do with the polls?
It would seem its scaring more people to vote Labor??
Are big business scared that workers under another three years of a Howard regime will revolt? Yet Howard is a puppet in big business’ hands and would just turn down the volume if this was the case.
It has me puzzled.
David said
Oh and Rudd is a conservitive laborite so hes not really goin to challenge big business. In fact he will be supporting them.
gatewatching - » Club Bloggery 9: Not Funny said
[…] The “I was making a funny” excuse found only deaf ears and outraged faces when it was used again in the dying days of the campaign to soften the impact of bogus campaign flyers in the Western Sydney seat of Lindsay which claimed that Labor was sympathetic to Islamic extremism and supported the Bali bombers – a racist and deliberatively deceptive stunt which some commenters in the blogosphere dubbed ‘bogangate‘. […]