email: elect AT

Crikey election book chapter

seat 2pp graphs '07

Seats to watch!

Federal pendulum (old)

margins since 1983

Historical results

Margins 1949  2001

See the lemmings!

two decades of Newspolls

state votes at federal elections

Votes and seat
1949 - 2001


Newspoll &
Morgan graphs

preferential voting

Newspoll Opposition leader approval ratings

Newspoll Opposition voting intentions




Bill Leak in the Oz

welcome to             



about this site

November 30 Memo to Kim (or Kevin): move into the Lodge

I've worked it out - what Mr Beazley has been doing wrong. He hasn't moved into the Lodge! If you go back over all previous Prime Ministers, each has lived in the Lodge.

 Beazley has never done this, and that's why he never wins elections.

The purpose of the above bit of silliness is to point to confusion of cause and effect contained in much popular political analysis. Labor's problem, commentators say, is that it is no longer seen as the party of economic reform and is not the favoured party on national security. Philosophically, it's not sure if it's Arthur or Martha; there's intermittent leadership sniping; structures are moribund; too much dead-wood. The PM has authority and is seen as a 'conviction politician!' - the opposition leader is a hand-wringer and a wimp.

In reality all these things - or perceptions of them - are part and parcel of being in opposition.

If an opposition leader tried to move into the Lodge before winning an election, he or she would be sent to the funny farm. If they set out to be a 'conviction politician!' the electorate puts them out to pasture - the Latham experience. 

And the opposition as the party of economic reform? That's John Hewson's Liberals.

Related to here and, perhaps, here.

November 29 Victoria: learning the lessons

When an opposition loses an election, or performs more poorly than expected, everything they did during the campaign instantly becomes a textbook example of what not to do. 

Mark Latham's 2004 Medicare Gold falls into this category: after election day the received wisdom became that Australians won't cop the idea of universal, non-means tested health care for old people. (Leave off the 'old people' part and haven't you  got most people's idea of Medicare?)

So it is with Ted Bailleiu's Victorian effort: whatever you do, don't impersonate Elvis if you want to win an election. Give the people substance, not fluff. 

Two weeks ago I suggested that:

'if on election day it is closer than about 56 to 44, I reckon we can say Baillieu did quite well (and Bracks didn't).'

Latest numbers seem to suggest it's more like 55 to 45, or even a tad closer. (The Age's Tim Colebatch calculates it as 54.8 to 45.2 and narrowing as counting continues.) The Libs took 5-8(?) seats. In states and territories across the country (apart from WA) Labor governments are being returned in landslides. See table below: Victoria 2006 was the closest two party preferred Labor re-election result over the past decade outside WA.

Election night in Victoria might have been a disappointment for Libs because expectations had been raised by some good opinion polls. If the polls had been like, say, those in Queensland during the last fortnight of the September campaign, they might have thought they got off lightly last Saturday.

Those good poll showings were indicative of something, and in the end the result was pretty good in the circumstances. Whatever Bailleiu did, he did rather well. But he'll probably never be Premier; his timing's all wrong.

November 28 Newspoll says 51 to 49

In The Australian. 

Table: current governments' two party preferred votes in decreasing order [as at 28 November 2006]






Feb '01




Jul? '05




Nov '02




Mar '06




Mar '03




Sept '06




Mar '99




Feb '04




Nov '06




Mar '96







Feb '02




Oct '04




Feb '05




Jun '98




Nov '01




Sept '99




Feb '02




Mar '95




Oct '98




Sept '01


A reader asks: how does Bracks' vote on Saturday compare with other recent state ones? The answer: 56 to 44 is 'about usual'. [update: as counting continues it seems more like 54.5 to 45 so have adjusted table entry] At left, in decreasing order, are all two party preferred votes recorded by current governments in Australia, apart from the two elected under PR. (Some numbers are rounded estimates from memory.)

(Slightly raised numbers in brackets) next to votes are which election government was facing: eg the top four numbers were all at elections going into a second term. Federal and Queensland - the only jurisdictions with three year terms - have four entries each. 

While Howard was elected in a landslide in 1996, and hasn't come close to repeating that performance, a distinguishing feature of the current state (& NT) governments is that initially they just squeaked in, sometimes with a minority of the vote, and have been returned with thumping, usually record-setting landslides. (WA is an exception, with a poor re-election effort.)

November 26 Victoria, the morning after

ABC's final estimate is Labor with 56 seats, one more than my Saturday prediction. If they edge up to 57 I'll get a small Centrebet payout, so fingers are crossed.

Perhaps the most interesting feature of the evening was former leader Robert Doyle on the telly with Kerry O'Brien (and John Brumby and Antony Green). Robert was terribly forlorn at the result, repeatedly noting that the Libs' primary vote had fallen, keen to offer kindly advice to the party: get behind Ted, work as a team etc. 'I wish I could say it's been a pleasure, Kerry' was his parting line.

The words 'crocodile' and 'tears' came to mind, and 'schadenfreude'. A very human former leader, but no doubt a chirpy one this morning, as anticipated when he resigned in May.

November 25 pm Election snaps from two polling stations

Town Hall and Brunswick Street. Click on a picture for enlarged version (a couple also have hover-over text); use back button to return to here.

November 25 am Final Victorian polls and prediction

Newspoll says 56 to 44 while both Nielsen and Morgan have 53 to 47.

My final prediction: Labor to win 55 seats. (Too late to place a bet.) Also tentatively reckon Greens will take no lower house electorates. [Update: both, I think, spot on.]

  • Note: In the Fin Review today on page 6 a graph of Vic opinion polls credits this site and a bunch of pollsters. Missing from that list is McNair, whose single survey result appears in the graph and whose name I included in the data I sent the paper.

November 24 4:47pm: another Vic wager

Today's Galaxy helped me overcome yesterday's bout of equivocation and I now, at least until I see tomorrow's polls, expect a biiiiig result, in the order of 56 to 44.

So have now put a few bucks each on 57 and 58 seats for the ALP, which is at the higher end of expectations. Respective odds were 9 to 1 and 11 to one.

Tomorrow: final guesstimate and perhaps, if up early enough, a final punt.

November 24 Victoria

Brian Costar's nice summary on Labor in 'the bush' in The Age.

Galaxy in the Herald Sun says 55 to 45.

Money thrown at worthy charity

> 62 SEATS

Centrebet is now taking bets on number of seats Labor will win. (Currently they hold 62.) In absolute contrast to yesterday's ponderings, I've snapped up big odds for results at higher side of expectations (at left). Just a few dollars. Yes, my betting track record has been poor lately; no, I'm not saying any of these results is more likely than not to happen, just that they are good odds. 

Bob Carr, for example, kept his majority in tact at his second re-election in 2003.

William the Conqueror has put out a detailed prediction which includes Labor winning 53 - more likely than any of my three, but only paying $9.

November 23 Scenes from an election

Brunswick candidate Christian Astourian shopping in the CBD

Victoria: a lay-down misere ... or not?

Is this election campaign strange? In middle of last week I wondered whether the upcoming result might be closer than had been expected - expected by me, anyway.

Bracks has rampaged over all the published data since the 2002 election; more important is the behaviour of Australian electors at state level over the last several years. We should be in for a thumping Labor re-election. We probably are, and yet ...

Mr Baillieu seems, for whatever reason, to be going well, doesn't he? Since a week ago we've had one pollster, McNair (a bit small sample-wise, admittedly), putting it dead even and another, Nielsen, with 54 to 46

While the uniform on-paper swing required to change government is very big, Victoria's electoral architecture was in the past generally considered friendly to the conservative parties (unlike NSW, believed Labor-friendly). Part of the reason for surprise at the 1999 result was the belief that the ALP needed at least 52 percent two party preferred to win. They won with 50.2 because of an an unexpectedly excellent showing in rural areas.

The bush gaveth in 1999, might they (or another geographic) taketh away?

Look, Bracks will probably win handsomely on Saturday. But .... am eagerly awaiting the final opinion polls, usually taken Wed-Thurs and published in Saturday papers.

November 22 Another Australian Newspoll editorial

Another week, another self-congratulatory Oz Newspoll editorial. It's one thing to use that space to promote a newspaper's causes - eg a republic, IR or whatever - that's what editorials do. But to throw all that rhetorical muscle into insisting that it is possible to make a judgement on a political party's fortunes from one survey only - and one that is out of sync with all other recent ones - is to argue that statistical black equals white, and is just sad.

(The bulk of the press gallery running, on cue, whatever line Dennis Shanahan puts in his headlines and lead-in ain't much cause for happiness either.) 

In both the editorial and opinion page cut and paste, the paper sees vindication in the results of their efforts: because Beazley's leadership is now wobbly, they were correct to read so much into one opinion poll.

What to make of this reference to last Friday's Michael Costello column?

Mr Costello errs in assuming the two-party-preferred measure is the one that matters. While he is correct that Australia has compulsory preferential voting for federal elections, the Newspoll preference allocation is artificial, based on the previous election result. 

So because the Newspoll two party preferred is unreliable, we should look only at primary support instead? Let's hope they're not serious.

What would be the result at an election if votes were cast as the latest Newspoll (the one so diabolical for Labor) showed: Coalition on 41, Labor on 37 and Greens on 9? If we eliminate the Greens and send 80% of their preferences to Labor and 20% to the Coalition - as they went at the last election - we get Labor on 44 and Coalition on 43. Who would win from there? Who knows? It would depend on the rest of the votes and where their preferences flowed (and of course how votes were spread across seats). In this instance, that 50-50 two party preferred is a pretty good descriptor of the situation: it could go either way.

As the Oz believes in primary support only, perhaps they should adjust their view and ask themselves whether Howard can win with just 41 percent, down from the 47 he got at the 2004 election.

November 21 Blood in the water and a suggested ticket

As the SMH repeats those very shonky Galaxy Qld survey results from Sunday, as the commentariat slips into a cult-like mass fantasy that the ALP has been doing badly in opinion polls (nearly every survey over the last six months has had Labor in a comfortably winning position, some by landslide amounts), the blood is in the water. We've seen it before, no-one said politics was fair, and Kim's days appear numbered.

But spare us the 'generational change', 'dream team' tag - smells like Downer-Costello. 

Is Julia really the second fiddle playing type? Do they need two motormouths out front? And her tactical view - Latham-like, raring to go on 'values' and 'culture wars' - is still a worry. 

The biggest barrier to a Labor win next year will once again be misgivings about economic competence, and a grounded, thoughtful finance type might complement the spiritual, expansive Rudd nicely. Bob McMullan also has ministerial experience but lacks the factional backing, so ...

How about a Rudd-Tanner ticket?

November 20 ACNielsen in Victoria

Says 54 to 46; no primary numbers in online version except Greens' 12.

Galaxy poll in Queensland

Came across this via Bryan Palmer: odd behaviour from pollsters Galaxy in yesterday's Courier-Mail with this question:

"If a federal election was held today, with John Howard and Kim Beazley as the two leaders of the major parties, which one of the following would you vote for?"

What on earth did they think they were doing inserting leaders' ratings into a voting intention question? Was there possibly an additional question - eg with Kevin Rudd instead of Beazley - which they've omitted from the results? (More likely, the newspaper did.)

(It was also preceded by another question, which lessens its quality a little.)

[Update: reader alerts me to report of another question in same survey.]

November 19 Victoria neck and neck?

According to a McNair poll in the Herald Sun of 609 voters, it's 50.5 to 49.5. That's unexpected enough, but the primary support levels of 39 to 46 are stranger still, and it's hard to see them translating to such two party preferred numbers. With Green support at nine, something like 49 to 51 would be more reasonable. Don't have hardcopy; I wonder how they did preferences. [Update: they notionally allocate them by minor party - full marks.]

It's difficult to believe anything remotely like this will happen on election day - and, yes, it's a small sample - but other polls have shown it closer than expected. We'll have to wait for final surveys this week.

  • A Baillieu win would be great news for federal Labor. It would show that incumbents are beatable; more importantly, the last time Victorians tossed out a state Labor government (1992) they swung to federal Labor by over four percent the following year.

On national grandeur

Here's George Bush on November 17 in Hanoi

... when our [Iraq] deliberations are complete - and as you may or may not know, we've got a lot of people looking at different tactical adjustments - once I make up my mind what those will be, I'll share it with [John Howard] right off the bat.

And here's Paul Kelly's interpretation, Insiders ABC November 19

I think the clear message from [the Hanoi meeting] is that Howard and Bush will work together in any subsequent change in policy, in any change in tactics. The trust between the two leaders remains in place and the two countries are going to work through these changes together. I think that's the real point of this meeting.

For a perspective of APEC that isn't Aussie parochial, visit White House November press releases

November 18  "Karl Rove"

With those two words, Bomber shifted his chances of contesting the next federal election back below the 50% mark. But his party's odds of winning next year remain about what they were before - say 2 in three - if we exclude the re-erupting Lemming scenario.

What I mean is:

(1) If Beazley remains as leader all this be forgotten and, assuming his health is indeed ok, he'll probably win next year.

(2) If Rudd becomes leader, he stands a ... possibly a bit better, possibly a bit worse chance of winning than Beazley. On the big plus side he would enjoy the same bonanza as Howard and Latham in their days - just a year in the job before the election, extended honeymoon, 100% party loyalty. (I don't subscribe to the 'the opposition leader must be a household name to win' theory.)

(3) If Gillard gets the job (Lemming scenario) the Coalition will probably win. Like Latham was, she's apparently the branch members' choice, and no doubt some in Caucus (particularly the youngies) are - as they were three years ago - thinking: 'Yes - finally someone to give the show a good shake-up! She understands the secret of Howard's success and will fight him on values, out-Howard the master himself and get him on the run! Do somersaults and win points at election time! Cut through! Create momentum! Triangulate!'

But surely nothing like a Caucus majority this time?

Deja vu

Here's what was happening three years ago. (Funnily enough a Newspoll published in early November 2003 contained a jump in government support which - on top of everything else - led to Simon Crean's demise. Back then, from a status quo of about 50-50 the government increased to 53 to 47. This year the jump was from about 52 to 48 in Labor's favour to 50 50. That is, Labor was in a much worse opinion poll position three years ago.)

Boom boom Harry

There's this short letter from Harry Quick MP to The Australian this morning.

November 16  Newspoll and Beazley: all about Dennis?

Several weeks ago Dennis Shanahan came out blasting in the Oz about criticism of his Newspoll reporting - 'pissing in the wind' he called it, and more. This week he and his paper's editorials (the second today) are going to town on the basis of ... one survey taken over the weekend (reported in two instalments). For a reminder that journalists in all outfits take their cue from Dennis's Newspoll interpretation, see Kerry O'Brien v Beazley last night

The latest barrage is to do with personality traits. Full Newspoll results are here, the bits that Dennis & co are hot and bothered about - health and education - here.

(Probably the most important bit of the Newspoll was something that hardly changed, an area where Howard leads Beazley by a mile. It's the reason the Coalition has remained in power for going on 11 years, why state Labor governments are finding re-election so easy, and why Tony Blair, Helen Clark and Jean Chrétien (unlike Howard) all clocked in record election wins over the last decade (with the lowest interest rates, unemployment etc in decades). The wall of Bill Clinton's 1992 'War Room' famously had a post-it note about it - it's in the first table here.)

All from one survey that itself, in voting intentions, looks a bit rogue-ish. It's tempting to think there's a bit of ego involved, and all this is not just 'I told you so', but retribution: 'go me and I'll bring Beazley down'.

November 15  "Middle Australia"

Mr Beazley goes on a lot about 'middle Australia', doesn't he? It is of course perfectly sensible (nay, necessary) to pitch at people who might be described with those words, but it's a clinical, descriptive phrase, a little condescending in the wrong hands. His repeated and deliberate use of it must mean that Labor research shows lots of people identify with it. 

It's no surprise that many folk reckon they're a 'typical Australian' or an 'average Aussie' - but 'middle Australia'? I guess it must be so - otherwise he's confusing backroom terminology with public communication. But does it alienate people - eg those who have climbed a few rungs of Latham's ladder - who don't see themselves in those terms?

Victoria: closer than expected?

Speaking of Newspoll preferences below (don't yawn). They are using different methods in the Victorian and federal spheres. In Victoria they're keeping the 'old' method - the one that came a cropper federally at the 2004 federal election - of asking for second preferences only and kind of extrapolating from there. For example, the final pre-2004 federal election Newspoll had 45 to 39, two party preferred 50-50. If they had used the method they have used since, of allocating preferences as they flowed (in total) at the previous election, they probably would have gotten 52 to 48 - much closer to the actual result of 52.7 to 47.3. 

This over-estimating of Latham's two party preferred vote was fairly consistent, and we might approximately infer that people who were going to give their preferences to Labor (eg most Green supporters) tended to nominate Labor as their second choice, while those who would eventually preference the Coalition tended to say 'dunno' to Newspoll's preference question. (Remember, Newspoll's question is different to Nielsen's; Nielsen asks (paraphrasing) 'on election day you must eventually choose between one or the other: which will it be?')

I bring all this up (yet) again because of this week's Victoria Newspoll, which had primary support at 44 to 40 (Greens on 8) and two party preferred 55 to 45. I reckon a notional distribution of preferences would round to 54 to 46 which, along with other recent polls, makes the contest closer than might have been anticipated. As you know, I'm a 'pattern' kind of guy, and if Victoria is going to conform to the state pattern the gap will need to widen over the next week and a half. The pattern guy in me expects this to happen, but if on election day it is closer than about 56 to 44, I reckon we can say Baillieu did quite well (and Bracks didn't).

(Is Newspoll's fascination with second (rather than full) preferences related to Dennis Shanahan's repeated use - eg yesterday - of the phrase 'second preference vote' instead of 'two party preferred' or perhaps 'full preference vote'?)

November 14   That's the way ... Sol does it

Newspoll says 50 50 from 37 to 41, not good for Beazley, no matter how you look at it - and it's a stretch to blame it on 'State Labor sleaze'. But the Oz editorial (penned by Dennis?) again goes over the top in dragging Newspoll into the Howard Wars. Is there really such a creature as an 'over-enthusiastic Beazley supporter'? Or even an enthusiastic one (other than Michael Costello)?

Newspoll's Green support is up two points to 9 percent. At the last election 80 percent of Green preferences went to Labor and 20 to the Coalition, and something like that will happen at the next one. But Sol Lebovic stubbornly sticks to a preference allocation method which doesn't take changes in Green (or any other individual party) support into account, and that extra two percent might as well belong to Family First.

If I was King of a polling outfit I would notionally distribute preferences from each minor party as they flowed at the last election (table 3 here) no matter how small those support levels. Newspoll is getting all that data - Democrats, Family First, One Nation, etc - and is just wasting it.


See Malcolm's Victoria pendulum in the same paper. (Not great quality.)

November 12 2006 Betting on US Presidential election 2008

Americans elect their next president in two years, and Centrebet is currently giving odds. Hillary Clinton tops the list at $2.40 and John McCain pays $3.25. To the casual observer (me) Hillary is too unpopular, surely. On the other hand, a McCain Republican candidacy would be very hard to beat, but those odds aren't attractive enough for a two year investment.

Going down the list of longer shots, most of whom I haven't heard of, Johns Kerry & Edwards are there (Buckleys), as are Jeb Bush and Dick Cheney (yes, laugh). But Rudi Guliani on $11 and Al Gore on $12 must be contenders. 

 I settled on Wesley Clark, Democrat primary contender in 2004, $10 @ $101 (payout $1010). Of course I've almost certainly done my dough, but ... Americans like soldiers.

Continue to previous posts


external links
these open new windows

Electoral Commissions



Antony Green (ABC)
Adam Carr
Simon Jackman  
Oz politics
Poll bludger
Possums Pollytics
Peter Tucker
Upper House


Enrolling the People
Crikey blog
Piping Shrike
Andrew Leigh
Rick Hasen (US)
Oxfam Shop
Australian Constitution
Australian Policy Online
Democratic Audit
Nicholson cartoons
Online opinion

WA Uni
election database

Distance Learning

Yarrh wines