With two more weeks to go:
• The ballot paper draws were conducted today, and my Victorian election guide has now been brought up to speed with full candidate lists in ballot paper order. I have also added updates to individual entries, where applicable, with the information you will have read on my blog posts, assuming you’ve been following them. As Antony Green’s number-crunching demonstrates, the 545 lower house candidates represents 6.2 per electorate, up from 502 and 5.7 in 2010. Family First, the Democratic Labour Party and the Sex Party are contesting fewer seats, but the void has been field by 30 Australian Christians candidates (compared with two in 2010), 32 from Rise Up Australia (who were not around four years ago), Voice for the West (14 candidates in Melbourne’s western suburbs) and more independents (up from 75 to 91). This growth pales in comparison to that for the Legislative Council, where an average of a little over seven columns per region in 2010 has swollen to over 17. The deadline for submission of group ticket preferences is noon on Sunday; they will be published by the Victorian Electoral Commission shortly after.
• The most notable late entry to the contest is former Labor MP Craig Langdon, who will run as an independent in his old seat of Ivanhoe. Langdon bowed out in 2010 after losing preselection to present incumbent, Anthony Carbines. John Ferguson of The Australian reports that Langdon will direct preference against Labor in what is described, a little excessively I feel, as a “crucial marginal electorate”.
• The Australian today reports that a Liberal internal poll found 54% of respondents supported the East West Link project, with 24% opposed and 22% undecided.
• I have updated the BludgerTrack Victoria poll aggregate on the sidebar with the Ipsos and Morgan SMS poll results, not that you’d notice the difference. Note that the date identified on the display is November 9, as none of the actual poll results have been any more recent than that.
• Kevin Bonham has an extensive post on the finer points of the horse race, including modelled result probabilities for individual seat outcomes using much the same method as BludgerTrack: an assumed uniform swing, but with adjustments made at the individual seat level for retiring sitting members and sophomore surge. Whereas I’m lazily using the adjustments from my federal model, Kevin’s are based on historical state-level observation, and thus reflect the undoubted fact of sitting member factors being greater in smaller state electorates. I thought it might be instructive to lay out the Coalition win probabilities for the key seats alongside each other, together with the implied probabilities of the betting odds from Centrebet/SportingBet (who were offering the same odds on every seat I looked at) and Sportsbet (who weren’t – arbitrage hunters take note).
Some big caveats need to be applied here to both my figures and Kevin’s. Our objective is to reach accurate totals, and in this we trust that local peculiarities will cancel each other out. Neither of us, I’m sure, would be taking our probability estimates for Ripon to the bank. Our probabilities are also based on the proverbial “election held today”, and do not factor in extra uncertainty for what might transpire over the next two weeks. Note also that the betting market probabilities are boiled down to the two-party contest, excluding the allowance in the bookies’ odds for seats being won by minor parties and independents. Blue indicates a seat defended by a Liberal sitting member, red by Labor, black by neither. All told, the picture that emerges is that the betting markets are taking a more moderate view of Labor’s prospects than the polls.
“The Australian today reports that a Liberal internal poll found 54% of respondents supported the East West Link project, with 24% opposed and 22% undecided.”
I really think these state wide polls about the East West Link are meaningless. The East West Link may decide the minds of some voters who are directly effected by it, but people living in other part of Melbourne or the state don’t give it much of a thought. People living in south east Melbourne for instance are more likely to be attracted to Labor’s idea or removing 21 rail crossings along the Frankston and Pakenham lines. A tunnel that they rarely ever use is unlikely to save the Libs in seats which aren’t helped by the tunnel. If The Australian thinks the tunnel is going to save the Libs they are dreaming.
Something else I need to look at in my model is whether my extensive use of personal vote effects (including rural/urban differences) means that the standard deviation I apply to each seat (currently 3 points) is too high.
If I did a forecast probability model rather than a nowcast I would probably get more similar probabilities to the bookies but to do that I would need to dig up massive amounts of data on poll-to-result changes in state election campaigns. The bookies factor in the possibility of a statewide swing from the current polls (eg that the state 2PP might actually end up being 50 or 56 rather than 53), a nowcast doesn’t.
If another week passes and the Libs strategy of firing negativity from other the horizon fails to get traction. I wonder if Napthine will go crazy brave in the last week and accept a debate with Andrews on free to air TV.
other = over*
Kevin – do you think that the massive increase in pre-poll voting in recent years changes that sort of analysis anyway? Because surely now the people voting today are voting in the “nowcast” ratios, compared to say 1979 when nearly everyone would have voted on the day and those longitudinal changes in polls could be factored in.
Rocket Rocket@5
Yes this is a good point. Even though the data we have from last week predates the start of voting, there is less opportunity for change in voting intention from then to the early votes being cast, than there is for change from then to polling day. So all else being equal the predictive reliability of the polls a few weeks out (and closer) should be better than it used to be – maybe not by much, but to some degree.
We didn’t really see this in the federal election but in that case there was a leadership-change bounce still deflating between the time the early pre-poll votes were cast and polling day.
William,
According to your election guide, Frankston has two Liberal candidates, and no Labor candidate?
William,
It’s no accident that Centrebet and Sportingbet offer identical odds; they are both owned by the same British parent company, William Hill. Tom Waterhouse has also sold to William Hill, and it is offering seat betting, and it will be the same odds.
Sportsbet and the previously independent IAS (initially owned by Australian bookmaker, Mark Read) are now a single entity – just the one website. They are now both owned by Paddy Power.
Ladbrokes is a separate company but is not offering seat betting, it only appears to be offering a market for the winning Party, as is Bet 365.
There do not appear to be any other companies offering individual seat betting, so your comparison covers the options.
Well, there you go.
Thanks for pointing that out, Agoo44 – corrected.
Matt Cunningham @MattCunningham4
Interesting poll result in the marginal seat of Bentleigh in tomorrow’s Sunday Herald Sun #vicvotes
======================================
(Not stated who poll is commissioned by.)
10
Or what the interesting results are.
I have good reason to strongly suspect that the pollster for Bentleigh is Galaxy and that it was in the field Thursday.
Tom the first and best@11
Yeah, it doesn’t much narrow the field; pretty much anything is interesting. That said if it showed ALP up 60:40 or Libs up by anything much I’d probably expect stronger language.
Par for the course in Bentleigh as of last weekend by my model was 52-48 ALP.
[Big #vicvotes story in SHS. Lib candidate quits over domestic violence claims.]
https://twitter.com/MattCunningham4/status/533587555929702400
Front page of The Herald Sun
Which says the Galaxy poll finds the Liberals have “drawn ahead” in Bentleigh.
“But its hopes of taking a seat off the Labor Party in Ballarat appear to be fading two weeks before polling day.”
The Lib who has quit is in Sydenham which is safe ALP, and just got safer.
http://www.heraldsun.com.au/news/victoria-state-election-2014/candidate-walks-over-claims-he-assaulted-wife-in-us/story-fnocxssc-1227124292814
[Candidate walks over claims he assaulted wife in US
James Campbell
Herald Sun
November 15, 2014 11:00PM
A LIBERAL candidate at the state election quit last night after the Sunday Herald Sun prepared to reveal he had once been charged with assaulting his wife.
The revelation comes as a Sunday Herald Sun/Galaxy poll found the Government has drawn ahead in the key seat of Bentleigh.
But its hopes of taking a seat off the Labor Party in Ballarat appear to be fading two weeks before polling day.]
Polling article & figures still not online yet though.
[Matt Cunningham @MattCunningham4
SHS/Galaxy poll. Lib 52 ALP 48 2PP in Bentleigh. ALP ahead 54-46 in Buninyong. Full polls in Sunday Herald Sun #vicvotes #springst
11:30 PM – 15 Nov 2014]
New thread.
Antony Green seems to have the Group Voting Tickets hours before they went on the VEC website. Not sure how. Indicating Labor is going to DLP, Country Alliance ahead of Greens in Western Vic.
More detailed analysis going up seat by seat, so far just Eastern Met http://www.abc.net.au/news/vic-election-2014/guide/emet/