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



Nicholson (clipped) in The Oz   2007 results->  AEC    ABC  

 Qld election: ABC Antony Green Crikey 

March 25 5pm: Me in Inside Story

On the Liberals letting go of Mr Howard, here.

Antony Green posts on the Queensland result

[Update: Antony has added an update of Brisbane only seats (with 2006 votes very similar to Galaxy and Newspoll, so probably the same rough area). Not much of a story there sadly, and indeed it seems Brisbane seats did swing a little more than the rest of the State.]

Here. His latest reckoning of the two party preferred vote is 50.5 to 49.5, rather closer than it appeared on election night and indeed more like what the polls were saying indicating estimating suggesting. As Antony notes, Newspoll's final survey was very close to the election-day outcome in primary votes. (Note that if Newspoll had rounded to nearest integer as they usually do they would have looked even better.)

(All of this is wholly separate to the Possum issue, which I will revisit at some stage.)

The lower that ALP vote gets, the more folks who said that the LNP needed significantly more than half the vote to win turn out to be correct, and the less correct my rough belief that if they got over 51 percent they probably would.

Late swing?

Antony looks at the question of whether there was a late swing. The question itself of course contains a few semantic questions, but he looks at the postal, pre-poll, absent and ordinary votes for Labor. As Antony writes, there is no great evidence there; indeed pre-poll votes would if anything indicate a swing away from Labor.

Anyway, as Labor's vote lead drops with counting, the question of an overall 'swing' disappears - but perhaps not a regional Brisbane relative to the rest one.

Antony's vote-type comparison is problematic: more and more people are availing themselves of non-ordinary ways to vote at every election, and have there been any changes to the rules since 2006? Still, I would like to see that table split between Brisbane and the rest.

A tentative thought bubble: 'LNP' on the ballot paper

I have a tentative theory about the abbreviation of the new party. Pollsters included the words 'Liberal National Party' in their questions, while the ballot papers had 'LNP' next to candidates' names. It seems likely that on election day at least some people looked at their ballot paper and thought 'who the f$#k is LNP, and where's my Liberal/National candidate?' But for this to account for a (relative) pro-Labor 'swing' (vis a vis the surveys) in Brisbane you would, I suppose, have to suggest the Liberal supporters were more susceptible to this than National ones.

March 25 Possum still stuck!

Possum still hasn't dislodged himself from that position, and sadly his fanclub isn't helping at all. I've only skimmed through it as it is long, will read in detail later. But basically he seems to be restating all that über statistical stuff that led him astray in the first place.

To be continued.

In other news

  • Mr Morgan takes an opportunity to beat his competition around the head.

  • For the second federal Newspoll in a row, the Oz helpfully includes a Peter Costello question. Malcolm Turnbull is surely on the phone thanking them as we speak. Dennis here and here, table here.

March 24 Me in Crikey re Possum Commitatus

This morning I woke up still cranky at getting the election result wrong, so I sucked a lemon. Then I fired bunger at Crikey's marsupial critter.


Federal Newspoll says 56 to 44

Tables here, Dennis here.

The write up of the Afghan war portion presents a false comparison: attitudes to sending the troops there in 2001 versus increasing them today. A question about about bringing them home would have been more comparable.

March 23 Qld election result bits and pieces

Using ABC results at end of Saturday evening.

  • Here is a rough table of seats in order of two party preferred swing to LNP. I said 'rough', it probably has some mistakes due to non-major parties in final count.

  • In uniform terms, what might be called the 'terms of the election' didn't move much. That is the LNP still needs in uniform terms around 53 percent to take government in its own right, and about 51 percent to deprive Labor of its majority.

  • Here is Pauline Hanson in Beaudesert and Anna Bligh in South Brisbane at ECQ and ABC.

March 22 Qld election: 17 seat Labor majority, 51.4 to 48.6

(As of last night.)

This is the sort of result most commentators (but not me) had expected when the election was called. By last night most had came around to approximately my position for one reason only: the opinion polls - both public and, apparently, private party ones.

Why did the polls all get it wrong? Was there something that swung it in the last few days? A swing back to the government in the dying hours can happen when the opposition is presenting something scary, but that doesn't seem to have been the case here.

(Ignore the 'West Wing' tales of polls tracking such and such in the final days/hours that the parties feed to commentators. They are usually self-serving and selective.)

Maybe Campbell Newman playing politics with the oil spill annoyed some voters. Maybe something else did it. Or perhaps the pollsters are simply doing something wrong.

Best early evening off-mike exchange

Antony: 'I think the government's back'
Kerry: 'Do you?'

Things we won't hear any more
  • How hopeless the Labor campaign was and how incompetent all those involved are.
  • How foolish it is to go to an early election. (Presumably there will now be more talk of an early federal one. )
Mumble prowess update

Not in the goose category, but my viewing apparatus proved faulty. Will not flay around for mitigating circumstances; some consideration is required, followed by adjustment. Views on likely outcome of future State elections will probably be affected.

To be continued.

March 21 Queensland election morning bits 'n pieces

  • Election addicts across the land can watch the Kerry and Antony show online at from 6pm Queensland time (7pm AEDT).

  • A swing of 5 to 7 percent two party preferred looks on the cards. This is certain to contain components over 10 percent (perhaps as high as 15) in some seats and of course much smaller ones elsewhere. This will mean an 'ohmygod' seat loss or two.

  • My initial expectation ('prediction' being too strong a word) remains: LNP to form government. (That of course includes a hung parliament scenario.)

Queensland Newspoll: 50 50

Or 49.9 to 50.1 to be precise, because as a one-off special Newspoll is reporting to one decimal place. In the Oz; tables here.

Like Galaxy, Newspoll has split the State into Brisbane v the rest, but a total sample of 1328 would likely put each portion above the Mickey cutoff of 600. It says much the same as Galaxy: a big swing - ten or so percent two party preferred - in Brisbane (however defined) and only a few percent in the rest of the State overall.

I've thrown the latest Newspoll and Galaxy numbers together, weighting for sample size, and get 41.1 to 42.4, Greens on 8.0 and others 8.5. [Two obvious issues: one set of results is to one decimal point and one is rounded to the nearest integer; and (I think) Galaxy includes 'Greens' in its 'who will you vote for?' question but am not sure at time of writing whether Newspoll does.]

Antony Green's recent post on the effectiveness of Green 'how to vote' cards shows overall a relatively (to other State and federal election Green flows) low flow to Labor in 2006. And of course OPV makes it lower overall still.

All in all, 49 to 51 is a fair estimate of two party preferred from Newspoll and Galaxy combined.

March 20 Galaxy in Queensland says 49 to 51 (again)

In the Courier Mail from primary support of 43, 40 and 9. Tables here.

There's a breakdown that has the swing much larger in Brisbane than the rest of the State, but of course the samples are too small for us to make a lot of. Given the consistency of Galaxy's overall numbers throughout the campaign it might have been worthwhile to add the whole lot together and then break that several thousand down geographically.

March 19 Queensland election

Brian Costar in Inside Story thinks Anna Bligh will hang on.

Once again, I am surprised at the emphasis folks are giving the required uniform swing on the pendulum. As I wrote before the last federal election, we shouldn't take it literally. As well, the concepts of swings and two party preferred are less meaningful in Queensland than in other States. A vote like 49 to 51 could go either way.

It's all in the seats, of course. Brian is doubtful that Brisvegans will be able to embrace the L and ignore the N in LNP.

But I think they will, because they are still Queenslanders, after all.

Two more polls

On past experience, Newspoll is surveying the State over Tuesday to Thursday and Galaxy Monday to Wednesday, and we'll get Galaxy perhaps tomorrow and Newspoll on Saturday (or leaked late tomorrow).

Then we'll know more.

March 18 Stating the obvious: Lib problems in the polls

When Malcolm Turnbull took the Liberal leadership last September, Paul Kelly said on Insiders that he needed "to be rating better in the polls than Brendan Nelson in order to try and exercise his authority in the party and keep the party united."

Indeed. Malcolm hasn't received those ratings, he can't exercise his authority and the party is disunited.

But it is doubtful anyone would be doing better in the position at the moment. All leaders have their strengths and weaknesses; if the polls are bad the weaknesses become evident and if they're good their strengths do. Don't try to work out what Malcolm's doing wrong; he's just in the wrong place at the wrong time.

If Peter Costello were leader right now, he'd be doing poorly in the polls and his particular weaknesses would be an issue. You might remember some of those: unpopular in the electorate; dismissive of colleagues; dismissive of rural interests; dismissive of anyone who disagrees with him; too keen on IR reform; has a tin political ear; seen as mean, tricky and out of touch ...

March 17 Politicians as interviewees

How do our politicians approach media interviews?

* Penny Wong evidently sees them as something to be survived without making a blunder. Probably most ministers approach them like this, on orders from the PM's office. But it makes her boring and uncommunicative in a portfolio where the minister needs to take the electorate into their confidence (or at least appear to).

* Wayne Swan is rather like Wong, plus the rehearsed sound grabs. 'Mate, learn your lines and recite 'em!'

* Julia Gillard has lots of innate oomph and great communication skills but is also now rehearsed and buttoned down. She's supposed to be a bit of a lefty but never gets to say anything remotely left wing. Thank God for parliament.

* Lindsay Tanner tries to channel Peter Walsh (or the received ideas of him) and impress businesspeople and Oz and AFR editorial writers. Tanner is a very good interviewee and gets good messages across. (Note to Wayne: if a message is going to stick it needs some complexity.)

No doubt Walsh's memoirs sit front and centre of Lindsay's shelf. But the persona of the grumpy finance minister who won't suffer wimpy, profligate ministers is old hat and based on a caricature from the past. It will get old soon; this is 2009 not 1989.

* Rudd is of course a very good interviewee. He has the advantage of being Prime Minister, but even as shadow foreign minister he was a doer, energetically getting messages across (usually about himself).

* Tony Abbott: always up for an interview, and while he's always on message you do get the real Tony there.

* Greg Hunt: how he approaches interviews is anyone's guess. Still stumbling over lines after 18 months.

* The member for Higgins sees an interview as a way to remind colleagues of how wonderful he is, what they're missing out on, and why they should still be sorry for not making him PM before November 2007.

He's done more interviews than the rest of them put together (maybe not more than Abbott) and he's in his element, but he does get silly sometimes, pulling funny faces and laughing at his own jokes. Well, usually, actually.

There is no excuse for Swan and Wong being so dull. They are ministers with the power of their departments behind them. I have it on good authority that both have inner lives and that real personalities lurk beneath.

They never 'own' the interview, they are constrained by it because they are so scared. They need to find a way to express themselves.

Free yourselves, Penny and Wayne!

Can it be a challenge without an challenger?

Big danger for Liberals is that Peter Costello's flirtations lead them to topple Malcolm Turnbull and hand him the position and ... he gets coy again.

Repeat: Peter won't touch the leadership with a barge poll in the near future. He would not last eighteen months in the job. (Perhaps if the government calls an early election and the voting intention polls are competitive, then Peter snaps his fingers, hoping to ride the initial wave of enthusiasm.)

March 16 Queensland Galaxy poll: 49 to 51

Tables here.

Polls seem to be stuck on the mid-point of my initial expectations. As is always the case, commentators' judgments follow the numbers: Labor's campaign has been a shocker, they say. The same in Western Australia last year: campaigning geniuses somehow forgot how to campaign. But if the next poll shows Labor ahead again, the ALP's campaigning will have been quite good.

We saw such things at the last federal election. But perhaps some things are just meant to be.

March 12 Will Queensland 2009 join Malcolm's table?

At my request, Malcolm Mackerras sends in this great table of two party preferred votes for all defeated governments in Australia from 1975 to 2008.

Most people would expect Queensland's Labor government to get under half the two party preferred vote in nine days, but the question is: will Malcolm be subsequently putting another entry into this table? (I suspect yes, but see the Wayne Goss 1995 result of a very low 46.7; Malcolm explains its inclusion at a note down the bottom.)

Note also that only three governments have won the vote but lost the election (a fate more commonly befalling opposition parties).

How can you mend a broken heart?

To all those praying and dreaming (again) that Peter Costello will take the Liberal leadership and give Kevin Rudd some what-for: you will be disappointed (again). As you were in November 2007 and September 2008. It is not yet time.

Peter will not take the job while the opposition trails so much in opinion polls. And not 18 months before an election. Have you not been paying attention? He is determined not to end public life as a failed opposition leader (as he certainly would have if he'd taken the job immediately after the election).

Perhaps this time next year, if, say, the polls indicate Liberal leader Hockey is a lovely guy but is not 'cutting through', but voting intentions are competitive. Then ... perhaps.

But right now Peter is only flirting. He knows what he's doing to you but he doesn't care; he will break your heart again.

March 10 Federal Newspoll says 56 to 44

In the Oz, Dennis Shanahan here and here, tables here. Malcolm Turnbull won't thank them for including Peter Costello in the best prime minister question.

Queensland Newspoll: 49 to 51

In the Oz, from primary support of 41, 43, 8 and 8, almost identical to the most recent Galaxy, which had the same two party preferred.

Betting market wackiness

According to Centrebet, the Rudd government is more likely to lose the next federal election (Coalition pays $2.90) than the Bligh Labor government is to lose this one (LNP pays $3.20).

And punters are supposed to be wise?

And another thing

What has Christian Kerr been smoking (bottom here)? Malcolm has enough troubles as it is.

March 9 Queensland election: two furphies

As you know, opinion polls up north have the LNP opposition competitive with /equal to /ahead of the ALP government in voting intentions. Yet commentators have difficulty seeing a change of government. You can probably put this down to gut feeling, but they also give two specific reasons:

1. Premier Anna Bligh is easily ahead of Opposition leader Lawrence Springborg in approval/popularity, and ratings on the 'all important [insert issue] issue' also favour Labor.

2. The LNP needs, according to the pendulum, 53.5 percent to win.

Both of these formulations are misguided. All/nearly all State change of government over the past couple of decades has seen the more popular premier knocked off. For example, the final Newspoll before the 1992 Victorian election had Joan Kirner preferred over Jeff Kennett 46 to 38; six years later the final pre-election Newspoll had Kennett ahead of Steve Bracks 51 to 35.

Preferred Premier/PM, approval rating and all the other stuff is interesting, but there's only one set of numbers that can point to an election outcome: voting intentions.

On that 53.5 percent - yes, it's possible that the LNP will need this, but unlikely. We shouldn't take the pendulum literally, and I reckon if the Queensland opposition gets over 51 percent after preferences it will probably win. That turned out to be about the threshold number at the recent WA and federal elections, despite each pendulum predicting a higher one.

Initial assessment stands: the LNP to probably win.

March 6 Galaxy in Queensland says 49 to 51

From primaries of 41, 44, 8 and 7. Am hoping to get tables, but see from the last set that the 2006 election result was 46.9, 37.9, 8 and 7.2, which after preferences was (as estimated by Malcolm Mackerras) 54.9 to 45.1. So a roughly identical Green and other share to this week added 8 points to Labor and 7 to the Coalition/LNP. Do that to today's figures and you do indeed get that 49 to 51.

(A reader has pointed out that the 80 20 Green split I assumed in this calculation was probably too favourable to Labor. The flow of all votes that were either (1) Green primary votes and then went to one of the major parties, or (2) went to another small party/independent and then to the Greens and then to a major party, was about 65 to 35. But we don't know the numbers for (1) only, which is what we need here.

Obviously 'others' significantly favoured the Libs/Nats last time; what they consist of will be important.)

The next federal election: Depression or war?

Wars are good for incumbents, at least for a few months/years, while recessions are bad. Using the most popular template, as a general rule governments in power when the Great Depression began were tossed out, while those that followed them did quite well. A hopeful stat for the Coalition.

I'm not telling you anything when I say Mr Rudd wants this to me more like a war than a recession. But a war needs an enemy - not an 'economic cyclone', but flesh and blood people: depraved foreigners determined to enslave the country and foist diabolical customs onto a peace-loving people.

Evil neo-liberals will have to do for the part, but it is a tricky one.

March 2 Queensland Galaxy tables

Are finally here, along with previous data.

My bets: Queensland

Will Queensland be like the Northern Territory or Western Australia in 2008? A big swing is on for sure: will it be not quite enough to change government, or just enough to do it? As noted at the outset, I slightly favour the latter scenario and have bet a few dollars accordingly @$3.15.

February 27 Galaxy says 50 50 in Queensland

In Courier-Mail. From primary support of 43, 42, 8 and (by process of elimination) 7 'others'. This is the first survey taken since the election was called. Am hoping to get the tables. [Update: Possum has them.]

Like Newspoll, Galaxy estimates two party preferred from primary support, but Galaxy takes this task very seriously. See this two year old Crikey piece (note that Newspoll no longer uses the 'tired old preference strategy' mentioned).

Galaxy's primary support is very similar to Tuesday's Newspoll's, the same in fact except for two points more for the LNP and one more to the Greens. This is not enough to turn a six point two party preferred lead (53 to 47) into zero (50 50), so there is indeed a difference in preference allocation.

In fact, while Newspoll was overly generous to Labor in its two party preferred numbers, Galaxy on the face of it seems to be going the other way, overstating LNP after preference support. But perhaps it's in the makeup of those 7 'others', eg a few points of Family First.

Anyway, yesterday I wrote a rather long-winded 'how to do preferences under OPV' but hadn't yet posted it. It dealt with this week's Newspoll.

Too boring to put on the main page, it is here.

And another thing ...

The boys at Crikey have a Queensland election thing going at the atrociously named Pineapple Party Time.

February 26 Question time in the House of Representatives

I popped into question time yesterday because it seemed like a good idea at the time.

Kalgoorlie MP Barry Haase got expelled for an hour; he gave Brendan Nelson's shoulder a squeeze on the way out. Keepers of the Howard flame.

The opposition went after the defence minister, who did his best to maintain a 'what, me worry?' composure. I left before the censure motion. As did Ralph Willis.

Sat next to several Asian people. A few years ago I reported an Indonesian laughing at our then prime minister's jokes. Yesterday several of the visitors laughed at a Wayne Swan witticism!

We may be close geographically, but at times the cultural gulf seems infinite.

End of report. Normal transmission resumes.

February 25 Brian Costar in Inside Story

On Queensland election, a nice piece with historical parallels n stuff, here.

But have to disagree on two things: that the LNP 'must secure a swing of 8.3 per cent ' to win; and that Tuesday's Newspoll 'was not large enough to tell us anything useful'.

My first objection is pedantic; Brian's statement assumes uniform swing. I reckon it's unlikely the LNP will need 53.5 percent of the two party preferred vote to form government. (But it's possible; that's roughly how the 1995 state election turned out.)

On the Newspoll, a sample of 752 is on the smallish side but still ok. (See Possum on the margin of error etc.) Galaxy's are usually around 800 I think. A pollster in a certain far away state uses samples half that size - now that's too small.

February 24 Two Newspolls: 58 to 42 and 53 to 47

In the Oz, Federally and in Queensland respectively.

On the federal numbers, all we can say is: poor Malcolm. When he was chasing the job last year it wasn't supposed to end up like this.

And look more closely at Queensland. Labor has 42 percent of primary vote, Coalition LNP has 41, Greens on 7 'others' on 10. Under OPV about half the Green votes will exhaust; another question is: what does that 10 'others' comprise? ?

As election day approaches that 10 will begin peeling off to the other parties and we'll get a better picture. As it is, 53 to 47 (or maybe 52 to 48) is as reasonable an estimate as any of the two party preferred.

Dennis here, Sean Parnell here.

And another thing ...

Gerard Henderson in the SMH puts Labor's current dominance over the Libs down to 'experience':

Nearly all Labor frontbenchers have been active in politics since their teenage years - in the party factions, in trade unions and/or on university campuses. On the other hand, quite a few Liberals wait until their late thirties to get involved in party politics.

This of course was the very reason Labor, a few years ago, was destined to remain in opposition forever: shallow gene pool, party hacks with no experience outside politics, compared to the knockabout 'real-life' appeal of the Libs. (Not sure if Gerard himself ever wrote such things, but it was a common story).

What a difference an election win makes.

Along similar lines to this.

February 23 Afternoon: Queensland election on 21 March

My prediction: the polls will narrow alot over the coming weeks (here's the latest Newspoll) and the two party preferred vote (some people don't believe in 2pp in Qld, but I do) on election day will come in somewhere between 51 to 49 and 47 to 53. So an expectation midpoint of 49 to 51, or six percent swing.

Of course, what happens over the next four weeks will be important. But does that mean I favour the LNP to win? I suppose it does - just. (That is, I just favour them to win, rather than favour them to just win.

If they do, amalgamation will be judged a wonderful thing.

(Update: obviously I'm not interpreting the pendulum literally, as you have to move 8 points on Antony's excellent gizmo to get a change of government. One never should interpret it literally, but I reckon the objections to using 2pp under Qld's OPV, noted above, apply more to discussion of swings.

This is the only election we're likely to get this year. Savour it.)

Betting market: the next federal election

See Centrebet's odds, below. The Rudd/Hockey scenario that I backed the other day has narrowed from $12 (but they haven't moved it up in the list, so it is out of order).

When, upon his elevation last year, I opined that Malcolm was unlikely to make it to the next election, I was in a minority of approximately one. After last week, it's probably a larger minority, but if we interpret Centrebet's odds literally folks still overwhelmingly expect the next election will be a Rudd/Turnbull contest. (Probably also, these are simply to the odds Centrebet started with and they haven't had much traffic).

The general expectation that Costello would take over if Turnbull fell seems to ignore the evidence of his behaviour since November 2007. Peter would not take the leadership unless victory looked likely (and that's not likely).
Life after Kevin?

The government side of the equation is interesting too. Gillard is quite reasonably favoured over Swan to become PM in the extremely unlikely event of Rudd not being around.

But lemming-rooster divisions lurk beneath the surface, and as Glenn Milne reminds us today Cabinet personnel are largely a reflection of Julia's choices. Recall, as Glenn's emailer does, the circumstances of Beazley's toppling.

The Libs aren't the only ones with internal divisions and loathing. As noted below, success breeds unity, but a vacant PMship would unleash all sorts of horrors.

Me in Canberra Times

Last week's Inside Story piece on the Liberals' approach to the stimulus package, reprinted here.

February 20 Liberal party instability: chicken and egg

Last year Tony Abbott quoted Bob Menzies as saying "defeat breeds disunity".

And victory, or its likelihood in the near future, breeds unity. If Malcolm Turnbull was generating great opinion poll numbers, his party would be a happy camp - or at least be disciplined enough to appear that way.

But that's not his fault: they would be as troubled no matter who was in the position.

February 19 Malcolm and the package: Me in Inside Story

I accidentally posted an item late yesterday, on the Libs' position on the stimulus package, and have taken it down. Its sentiments are put (hopefully more eloquently) along with others in this Inside Story piece.

Economy and State elections

Andrew Leigh and Mark McLeish here.

My bets: leadership

A reader alerts me to Centrebet's odds for leadership at next election. I threw a few dollars on Rudd/Hockey, paying $12.

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


Inside Story
House Rules
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