Victoria state election pendulums
Sunday December 8 2002
It's moot now, given the eight percent swing to the Bracks government, but assuming a uniform one, from the results if the swing had been under two percent (that is, taking the post-election pendulum and counting back by 6 percent) the government would have fallen short of a majority in its own right. So pretty well every commentator was correct on this front.
Funnily enough for such a big overall swing, five seats (apart from Benalla, which came from a by-election) actually swung to the Coalition. They were all rural; none resulted in a change of hands.
The biggest swinger to Labor was Narre Warren North. East and south-east Melbourne seats figured predominately in the biggest swingers. Before November 30, these seats had looked beyond reach if the swing wasn't substantial and had stubbornly failed to move to Labor when others did in 1996 and 1999. With 20-20 hindsight it's now blindingly obvious that they had to catch up some time, once the memory of Cain-Kirner had been exorcised.
Friday December 6 2002
There's this Age article. The numerically oriented can also read this version, originally written for the Fin Review's Lies & Stats column but pulled after the Age one went ahead.
Saturday night, 9pm.
The polls were right again. Looks like a 15 percent two party preferred win for Steve Bracks, which is huge. And Green got no seats.
Bracks' landslide conforms with recent history: lineball Labor wins in NSW and Qld followed by landslides. Mike Rann in SA can take heart. (WA wasn't lineball, but the same probably applies to Geoff Gallup.) The Liberal bitching has begun already, but you have to wonder how much difference any different decisions would have made.
Bracks, if he's interested, can look at Bob Carr's fate next March for an idea of what lies in store.
Burwood's been held by Labor, despite a Liberal swing. Ow.
Saturday November 30.
And they're off (for real this time).
Well, the gap hasn't halved but the last Newspoll has it down to 15 percent. I still reckon it'll come in at about 10 or 11 percent.
Only contentious prediction I'd chance at this stage is the Liberals winning back Burwood. (That is, I'm for it, despite Centrebet giving it a 7 to 2 outside chance).
Thursday November 26.
As we speak, Newspoll employees are asking Victorians "thinking about election this Saturday, which party do you intend to ...?
Final opinion polls will appear in Saturday papers. Would be very surprising if the 20 point gap hasn't about halved in Newspoll by then.
Tuesday November 26.
The polls can't be right.
Granted, Victoria is now a Labor jewel in the crown - along with Tasmania the only state to return federal Labor two party preferred majorities at every election since and including 1993. And although in 1999 the polls had Kennett winning easily, it wasn't by the margins predicted here - 60:40 seems the consistent two party preferred. (The Newspoll published on election day three years ago showed the parties neck and neck.)
Still, I reckon the most likely outcome is roughly 55 to 45 two party preferred, giving Labor about 51 seats, independents 2, say two Greens and the rest - 33 - Coalition. Unlike NSW, which gives Labor good seats for the vote at state level, the electoral demographics favour the Coalition in Victoria, so a decent two party preferred win can lead to a relatively modest seat majority.
Thursday November 21
The latest Newspoll this week puts Labor ahead two party preferred 60:40;. Saulwick last week put it at 62:38. Numbers like this render ruminations on Labor needing a swing of two percent just to hang on technicalities. A ten percent swing is a landslide.
Worse for the Liberal Coalition, Newspoll shows few undecided.
And they're off.
The Victorian state election will be on November 30. Forget the opinion polls that show the parties neck and neck on primary votes. After preferences (and at least two thirds of Green voters, irrespective of how their Green candidate's "directions", will give their preferences to Labor), the Bracks government is comfortably ahead - 53 to 47 is the estimate of Sol Lebovic of Newspoll.
But where are the seats?
So Steve Bracks still sits pretty in the polls. But the feeling in election-watching land seems to be that with such support Labor should win the state election; and yet, and yet .... it's not readily apparent where the seats will come from.
After a redistribution, the Coalition starts this election with a notional seat majority
One of the reasons Bracks was written off before the 1999 election was that it was believed he would need a high two party preferred vote - say 52 percent - to win. In the event he scraped in with a touch over 50 percent, courtesy of unexpected and close wins in the bush. Several of these rural and regional electorates have since been redistributed from marginal Labor to marginal Coalition; in any event you might question whether, with the dreaded Jeff gone, bushies will stick with Labor anyway.
So he may indeed need every scrap of his 53 percent to get back.
A word of warning. While the seat margins I use are taken from Antony Green, the Victorian Electoral Commission was raided for the geographic classifications. These were vague in parts, and with my knowledge of Melbourne and Victoria somewhat scrappy, there may be some anomalous entries. I will be fixing this page up and adding to it in the coming weeks.
Also, I've left out the three seats currently held by Independents: Bass, Mildura and Gippsland East. They should either remain independent or go back to the Coalition.
Oz readers on Tuesday November 5 may have noticed Malcolm Mackerras's pendulum and, importantly, his post-distribution notional seat allocation which sees no net change in seat allocation. He has a seat swap - Mitcham becomes notional Labor, Yan Yean notional Liberal. Antony Green, whose numbers I've used, sees Macedon, Narracan and Geelong joining Yan Yean on the other side of the pendulum.
God knows who's right, but it makes little difference to the diagnosis that Labor will probably go backwards in the rural/provincial seats and need to make gains elsewhere (east Melbourne).
There are four mini-pendulums below. Click each to enlarge and read commentary. The figures for the seat margins were filched from Antony Green at the ABC and the locations and classifications from the Victorian Electoral Commission.
Given the Bracks government starts with a notional seat deficit, it is believed they will need extra seats in Melbourne's east to hang on and inroads in the bush are unlikely. I've made pendulums according to these classifications.
The first pendulum is all seats in the state with less than 10 percent margin. The ten percent criterion is for space reasons (it's not likely others can change hands). The second shows rural Victoria, the third Melbourne's east and the fourth everything else (under 10% margin).
The redistribution took several marginal Labor seats and made them marginal Coalition. Most of them were rural and possible losses anyway; the resulting notional seat allocation actually gives a more realistic indication of the Bracks government's challenge in staying in.
With 50.2 percent of the two party vote at the last election, Labor now sits several seats shy of a majority (the exact number depends on how you count seats that changed hands in by-elections in last term and the three Independents' electorates). If you assume Labor doesn't make headway in rural and regional Victoria, the required swing in the east and 'rest' becomes substantial. In aggregate terms, if Doyle's opposition can keep the swing to the government to below, say, one and a half percent, they have a good chance of winning.
Swings are never uniform, of course, but the prognosis remains: a swing to Labor that sees a change in government is a distinct possibility.
Let's make a very rough calculation. Assume the state-wide two party preferred vote on November 30 comes in at 53 percent for Labor, 47 for Coalition. Also assume Labor does poorly in the rural seats.
From the rural pendulum, they lose Ripon, Benalla and Seymour, but gain Geelong and Bellarine, which are more urbanised.
If they do poorly in the bush, the rest of the state must compensate to make up the 53 percent. So we'll add on a percent to the Labor vote in rest of the state. Make that 1.2 percent, so we can apply a nice round four percent swing to the rest of the pendulum. (From 50.2 to 54.2 percent. I said this was rough.)
Looking at the eastern suburbs. I'm going to give the Burwood, won by Labor in a by-election following Jeff Kennett's retirement, back to the Coalition. A four percent swing to Labor just gives them two seats, (Narre Warren South and Monbulk).
In "the rest" a four percent swing gives Labor five seats (from Yan Yean to Eltham).
Where does all that leave us? Most analysts put Labor's notional seat count at 41 out of 88. From the bush they have a net loss of one, from the east a net gain of one. From the rest there is a gain of five.
Which gives the Labor Party 46 out of 88 seats, a majority of four in its own right.
Please note I'm not making particular predictions on results in all the seats mentioned. Pendulums are more general than that. And given my scant knowledge of individual seats, these pendulums are more general than most.
A similar exercise assuming a 52:48 percent state-wide vote gives Labor 44 seats.
See any howlers? Please let me know.