With three weeks to go until Victoria’s November 29 state election, I have finally got my act together to put a poll aggregation up, which you can find on the sidebar. Whereas the federal BludgerTrack purports to improve upon aggregated polling through bias adjustments based on historical performance, this really is just a poll aggregate, except to the extent that historical performance is used to determine the weight the various pollsters carry in the model. Had historically based bias adjustment been used, the Greens vote would be coming in quite a bit lower. The seat projection, which presently has Labor on 50 and the Coalition on 38, is determined as it is in the federal BludgerTrack – by calculating two-party win probabilities for every seat based on the size of the overall swing and specific seat-level factors such as the effect of retiring members, and determining aggregate seat totals by adding together the probabilities. The model assumes that all 88 seats will again be won by the major parties.
I will try, but probably fail, to update this each time a new poll comes in, which will probably be quite often. Ipsos at least is currently in the field, and we should be getting a second result from it over the coming days. Here’s some of the rest of what’s been happening:
• John Ferguson of The Australian reports that internal polling has “hardheads from both major parties” expecting Labor to win the Frankston line trio of Bentleigh, Mordialloc and Carrum. Liberal polling is nonetheless said to be “better than it expected” in Frankston – although earlier in the week the same reporter related that the party “expects to fail” in the seat, and is hoping for a compensating gain in Cranbourne. Labor’s polling is also said to show it on track to retain its own marginals of Eltham and Yan Yean in Melbourne’s north-east, where Labor is respectively handicapped by the retirement of a sitting member and an unfavourable redistribution.
• The Age reports that a Greens-commissioned Lonergan Research automated phone poll, targeting 400 voters per seat, shows them leading in Melbourne (40% to Greens candidate Ellen Sandell and 30% to Labor member Jennifer Kanis, a 53-47 lead to the Greens based on 2010 preference flows) and Richmond (39% to Greens candidate Kathleen Maltzahn and 29% to Labor member Dick Wynne, for a 54-46 two-party lead).
• Clive Palmer will today unveil his candidates for the eight Legislative Council regions, having hitherto kept their identity close to his chest. Palmer United will not be fielding candidates in the lower house.
• Tex Perkins, veteran frontman of The Cruel Sea, The Beasts of Bourbon and Thug, will run as an independent in Albert Park as part of a campaign to restore St Kilda’s Palais Theatre.
• Daniel Andrews has rejected the notion that Labor will enter a preference deal with the Greens, scoring helpful headlines ahead of the Liberals’ likely confirmation that they will again place the Greens last, which proved the tactical masterstroke of their 2010 campaign. This presumably means that Labor will behave much as it did in 2010, when it placed the Country Alliance ahead of the Greens on two of its three regional upper house tickets and the Sex Party ahead in Northern Metropolitan region.
Rocket Rocket@37
This made be chuckle, I did wonder how they’d kick it up a gear.
Clearly myki is coming up – any guesses on the headline? How about
“Myki CFMEU cover up”
William
Do you think the Morgan Victorian polls suffer from the same Labor bias that their national ones do?
The LegCo could be an interesting contest. A few Labor people I talk to seem confident that Dimity Paul, the #3 for the Eastern Metropolitan region (behind Opposition Whip Sean Lean and Shadow Planning Minister Brian Tee) could be elected. I don’t share that confidence, the Eastern Metropolitan region being what it always has been, conservative. I also note the two ALP members for Eastern Victoria are both retiring…
I think the LegCo is interesting in that it’s probably the only case of substantial electoral reform in recent times (granted I’m not as familiar with what happened in other states). It’s also the most practical area for the Greens to make gains and yet people only seem interested in the possibility of winning Prahran (while seemingly writing off – on paper – easier wins in Richmond, Melbourne and Brunswick).
Oh, and thank-you, William. Your efforts are always appreciated. Love the guide so far (spent a fair while on it when I was meant to be writing a newsletter) 🙂
Bugler@53
Even the subtle detail of an odd number of seats in each district suits the greens.
I think the focus on Prahran comes from it being the rare chance of trying to poach a seat from the right. I think the rank and file would much rather be taking the fight up to the conservatives and therefore Prahran energises the base. It says “we are mainstream”.
I don’t, Darn. The problem seems to be with the face-to-face method. Morgan has never had a bias issue with its phone polling, and while the SMS polling obviously has problems (see those Greens results), it doesn’t seem to include a bias against Labor – who have in fact been coming in implausibly low on the primary vote, making room for the inflated Greens numbers.
Thanks William
http://baysidenews.com.au/2014/11/10/faine-visit-puts-frankston-in-the-spotlight/
http://baysidenews.com.au/2014/11/10/shaws-nasty-station-stoush/
http://baysidenews.com.au/2014/11/10/pollies-ignoring-methadone-plan/
I think the original plan for the Legislative Council change was 6 x 7 member regions, (or 7 x 6 member regions?) It was said that they went for 8 x 5 member regions so that there would be 3 country regions. Though of course it made the quota for a smaller party to get one seat a higher 16.7%.
A bit like when they changed the Tasmanian Lower House to 5 x 5 instead of 5 x 7 – meaning the Greens neede to get 16.7% rather than 12.5%. Of course in the end that change locked in a fair few 2-2-1 scenarios as the Greens vote increased.
Seat-based polling (Venom) updated to include the latest Morgan poll
http://www.dailykos.com/story/2014/10/27/1339458/-Victorian-State-Election-Modeling-With-Venom
53
In the redistribution coming into effect this election Eastern Metropolitan, which has a declining population because it lacks any real growth areas, gains ALP held Ivanhoe and is thus a more favourable seat for the left.
However the idea that the ALP, which did not even get two quotas in Eastern Metro at the last election, would get enough surplus and minor and micro party preferences after the election of its second MP ahead of the Greens, who got 0.74 quotas (redistribution adjusted), is not a particularly reasonable idea.
The Greens chances for gains in the Legislative Council are all dependent on other parties because of the high quota and so some focus on the Assembly, where the Greens have chances without so much help, is reasonable. It also adds credibility to a party in they have lower house seats and party status in Victoria is 10 seats and the Greens are not going to get that in the Legislative Council alone.
59
I think it was a mistake to shrink the Legislative Council as part of the reforms. They should have increased it to 49 (7×7) or 45 (9×5) and also increased the Assembly. 7 member regions probably would have been better. Maybe if the Greens had won Melbourne Province, they could have achieved a better constitutional change (unless the ALP decided to wait until the next term once the 1999 MLCs had been replaced by a more ALP friendly cohort).
A good analysis of the financiql madness that is the $17 billion EW road link. Is this the last spiteful act of a doomed government?
http://www.theage.com.au/victoria/victoria-state-election-2014/universities-analysis-shows-east-west-link-bill-to-reach-178-billion-20141110-11jxox.html
An annual repayment cost of $400 million (2014 $)means that, on average, every Working Victorian adult adult will pay around an extra $2 every workday in taxes to cover it, whether they use the road or not. Those who use it will also pay the tolls. This is madness.
Morgan poll:
http://www.roymorgan.com/findings/5919-morgan-poll-victorian-voting-intention-november-10-2014-201411100547
ALP 53.5 (+1%)
LNP 46.5 (-1%)
Primaries: ALP 36 (+2), LNP 38 (+0.5), GRN 18.5 (-), FF 2 (-0.5%), CA 1 (+0.5%), OTH 4.5 (-2.5%)
http://www.theaustralian.com.au/national-affairs/state-politics/jacqui-lambie-too-busy-to-campaign-for-palmer-united-party/story-e6frgczx-1227118742949
Not sure if making Lambie campaign in Victoria appears like a good move from the view of Tasmanian voters.
I do hope that Guy does not become PM.
Read more: http://www.theage.com.au/victoria/victoria-state-election-2014/melbourne-city-council-to-ask-matthew-guy-to-reverse-decision-on-collins-street-site-20141110-11jxb9.html#ixzz3Ijusu1eT
Still 200,000 missing voters.
http://www.theage.com.au/victoria/victoria-state-election-2014/victorian-state-election-200000-eligible-voters-missing-in-action-20141111-11kh66.html
Tom @ 62
Australians already seem to be horrifically overgoverned, so I’m not sure many would get behind the idea of yet more politicians.
Betfair betting market seem convinced … ALP firmed into 1:15, LNP friendless and out to 5.6:1 from 4.6:1 a couple of days ago…
http://blogs.crikey.com.au/theurbanist/2014/11/11/is-it-time-melbourne-got-a-rail-line-to-the-airport/
http://www.theage.com.au/victoria/victoria-state-election-2014/victorian-state-election-kim-wells-and-nick-wakeling-lose-liberal-campaign-tag-20141110-11jv7j.html
http://dailyreview.crikey.com.au/razer-tex-perkins-belongs-on-your-itunes-not-your-ballot/15408
Razer seems not to have heard of preferential voting. She’s trying to argue (among other things that actually might have merit) that Perkins running could cause the Liberals to win the seat.
I might add this is about the fifth or sixth Razer article in the past month or so I’ve considered to be at least partially rubbish, against about one in that time that was actually useful. Not that I was ever a fan in the first place, but there seems to be a running theme of her saying silly stuff in an attempt to pretentiously troll any passing thing that gets more attention than she does.
Watching the ABC News last night, I spilled my glass at the interpretation of the “Vote Compass” survey. For example, it was claimed it shows that half of Victorians support the Tunnel.
As far as the attitudes of Victorians is concerned, Vote Compass tells you nothing, niente, zilch – since participation is self-selected and not random. In present company, I need say no more.
Shouldn’t the ABC apologise? Who knows what Vote Compass will purport to demonstrate over the next fortnight?
72
She does mention how to vote cards at the previous election but not how to vote cards for Tex. There is also no mention of the Greens running in the seat, who have a significant vote but are still likely to come third again.
I keep hearing Liberal ads saying how bad it is to be associated with the CFMEU.
Obviously far better to be associated with Neo-Nazis!!
http://www.theage.com.au/victoria/victoria-state-election-2014/liberal-party-campaigner-scott-harrison-a-former-neonazi-20141112-11l1dx.html
ps – heard Dan Andrews doing his own ad on the radio. Very good media strategy and a postive sign.
SB 73 – Of course they could have announced this story, using the same dat, by headlining “More than half of all Victorians do not support building the Tunnel” !!
As in 46% support!
RR 76 — the point is that they cannot say anything about their data, except that it reflects solely the views of those who have dialled in to the ABC website — and those people only.
It cannot be claimed in any respect to represent or reflect the views of Victorians or the electorate.
Yet the ABC News made such a claim.
77
They claim this because they have adjusted the results by weighting them demographically. This is still dubious.
Yes they may claim that demographic weighting has resulted in correct voting intention figures before, but it does not follow that it results in correct figures on specific issues. Though it’s really most likely to fail on such issues on cases where the selection method introduces an ineradicable bias, like attitudes to the ABC, computing technology issues and so on. For most things it might work OK, but I’d be cautious about betting on it.
Also the problem is that they include people doing it as a joke or experiment. For instance my attempt to faithfully do the questions as an extreme Tea Party “Republican” which resulted in a result far to the right and below the LNP “dot” would be included in their results!
I just had a thought – imagine if Mr.Ex-Neo-Nazi had been a Labor staffer. What would be the Herald-Sun front page tomorrow?
LABOR NAZI LINKS
DANIEL, MEIN FUEHRER!
LABOR’S DIRTY SECRET
I’m sure they could come up with something inappropriate!
Rocket, that’s exactly the same as people lying when they are interviewed by Newspoll. Every poll assumes the respondent is telling the truth. Error margins are a measures of statistical error, not sample bias error or respondent bias error.
Respondents are also only included in the Vote Compass sample for analysis if they fill in all the demographic questions. If a survey has no demographic data it is excluded from analysis because it can’t be weighted.
The data is weighted by eight census variables and by electorate.
Polling companies like Newspoll also weight their sample before publishing their results. In addition, they use stratified sampling by age, so they always reach a point where they have enough older voters and will only conduct further interviews with the hard to get younger voters.
Newspoll uses a question on education level as one of their weights. But there is a limit to how much weighting you can do when you only have a sample of a thousand, and when you only have a limited number of demographic variables.
When you have a sample of 30,000 and a range of demographic variables, you can apply more weights, though the number of constraints you apply to the data increases the error margin.
As Kevin says, if you have a question which is heavily correlated with motivation to do the survey, then it is hard to unpick the bias that self-selection creates.
Napthine enlisting the aid of Nazis? Seems to bizzare to be true
Read more: http://www.theage.com.au/victoria/victoria-state-election-2014/liberal-party-campaigner-scott-harrison-a-former-neonazi-20141112-11l1dx.html#ixzz3IqB8DAsu
The sad thing about the ex-neo-Nazi is that some of his professed views now are more socially progressive than those of many “socially conservative” Liberals. He’s got there from where he was in just four years; how long will it take them?
Just redid my model after checking some sitting member effect stuff following sceptical input from Psephos.
It’s still saying 48-40 for a 2PP of 52.6 (which I will revise once the Ipsos comes out) and it currently has Labor gaining Carrum, Frankston, Bentleigh and Mordialloc and losing notionally Liberal Ripon.