November 18 Galaxy and Newspoll in the marginals

Here's the graphic, posted by Pollbludger. And here's yesterday's Newspoll, also in the marginals. Newspoll is more optimistic for Labor, and part of the reason might lie in the different seats surveyed.

In Queensland, for example, they both phoned voters in Bonner, Moreton and Herbert, but Galaxy also included Longman while Newspoll did Blair. 

No-one seems to think Mal Brough (Longman) is in trouble, although Blair is one Queensland seat believed to be in play. 

Polling in Ryan and Leichhardt would be interesting.

Newspoll and Galaxy marginal seat swings (to Labor)

Still, all good (and both surveys have large samples). On the other hand ...

Taverner poll of mortgage holders In Sydney and Melbourne only. It's 57 to 43, apparently. We know nothing about how the survey was conducted, eg sample size.

Also a reference to those on under $80k per annum: "Howard's battlers .. who have loyally delivered the Coalition four terms but look set to deny it a fifth." Yet (apparently) a 2004 Taverner poll had them voting Labor 53 to 47 (compared with 66 to 34 this year).

About Boothby In a recent Canberra Times piece I wrote off, with a flick of the pen, Labor's chances in Boothby. An Adelaide friend chastised me at the time, suggesting I knew bugger all about the seat. Quite true. It is a possible Labor gain.

Nielsen demographic data

Here's the latest Nielsen demographic stuff. A bit more recent than Alan Ramsey's version.

November 17 "John Howard is a conviction politician."

It is highly unlikely that you will ever again experience these words from Paul Kelly's lips or word processor. Some people will find this a relief, but it may induce sadness in the others.  Either way, it's the end of an era. (Ok, it might get into a political obit or two - but in the past tense.)

The Oz now seems to be reconciling itself to the seemingly inevitable, with straight analysis of the latest Newspoll marginal seat situation and Sol Lebovic finally (no doubt reluctantly) unclutching his beloved "who is more capable of running the economy?" straw.

Meanwhile, Professor Jackman notes that betting markets are blowing out. 

It feels like election day, but there's still a very long week to go.

November 16 Nielsen reckons 54 to 46

See Age and SMH. There was a time when these papers published tables online.

Cumulative Newspolls

The Oz has the latest cumulative Newspoll, combining the most recent 55 to 45 and previous 53 to 47.

Decreasing order of Newspoll state swings to Labor from the 2004 election are: Queensland (10), Victoria (10), SA (7), NSW (6) and WA (6).

Kevin Rudd's evasiveness This catchingly titled piece in Crikey yesterday actually dealt with Mr Rudd's refusal to appear on the ABC's Insiders during the campaign. Presumably the same applies to all "hard" interviews. You can understand why. Barrie Cassidy can be quite gruelling; he made Swannie shift in his seat a few times last week.

Apparently, they're sending Julia Gillard instead for the final Insiders interview.

In the dying days of the campaign, this, I reckon, is a mistake. 

Gillard and Latham

I've made the comparison before: Gillard is much, much better than Latham - more cluey and articulate - but she represents something similar to the electorate. They like her, they admire her, she's got gumption, but ... she embodies risk. 

She does not have the calming effect Rudd has; she has the opposite.

In government she would add breadth and appeal, and she does that in opposition too, but for oppositions being safe is (almost) everything.

Closing the deal?

And she'll probably be grilled about Kevin's no-showedness.

Maybe they reckon it doesn't matter because only pointy-heads watch shows like that. But if support is weak in the marginals and strong elsewhere, Labor might need every pointy-head vote it can get. 

In 1993, Dr John Hewson didn't make a final week Press Club appearance, and this was described by some as his never quite looking the electorate in the eye and "closing the deal". 

(As far as I know Rudd is going to the Press Club next week.). 

My bets: two shock Liberal wins

Speaking of the marginal mortgage-belt, I do expect, as you know, these folks to perform weakly (swing-wise) for Labor tomorrow week. Looking at this table, I can imagine that over the last three years lots more Liberal-voting young families have moved into the top two seats, which are both marginal Labor.

So I placed bets. Yes, Holt is a long shot, but the odds tempted. Cowan odds not too bad either.

And Peter Costello losing his seat? You're right, it's not likely, but ... stranger things have happened. 

(The other four bets seemed like good ideas at the time; obviously the amount wagered on Boothby was not large.)

November 15 The stories we tell

Some press gallery types are still excited about Rudd's self-description as an "economic conservative". What an amazing thing for a Labor leader to say! Presumably Beazley, Latham and Crean all reckoned the economy didn't matter. (Latham was probably the economically driest of the lot.) Not to mention Keating and Hawke.

And perhaps these writers were elsewhere in the '80s and early '90s, when Labor ran surpluses (the first in generations, according to Treasurer Keating) and warned - successfully for a while - that the Liberals couldn't be trusted with the economy.

Column inches must be filled somehow. And after Saturday week the stories will really take off.

Geoff Lambert's prediction: 44 seat Labor majority

A detailed number-crunching exercise, posted at Pollbludger.

November 14 Me in Crikey: that 30 seat majority

Recently an emailer requested I divvy up my prediction of Labor winning 90 seats (ie adding 30 onto their current 60) into seat gains by state. 

This I have done today in Crikey. 

(By way of credentials, I present the Latham experiment.)

(Nominated almost the same number in a competition a few months ago.)

November 13 Belting the mortgage

Have constructed another electorate table from George Megalogenis's excellent Oz data. This one has percentage of households paying off a mortgage. Am using for a piece I'm writing for the Fin Review.

The percentages range from 52.5 percent in Holt in Melbourne (which also tops this list, if you sort it, of couples with dependent children) to 18.7 percent in Sydney's Wentworth. I have also included current margin, AEC socio demographic classification and the swing in 2004. 

Large-ish post, moved to here.

November 12 Newspoll says 55 to 45

Here are the tables, here's Mr Shanahan.

Note the Greens are at six, their highest Newspoll since April.

Gerard Henderson's crystal ball Gerard Henderson's penchant for making fun, after the fact, of other people's election predictions has often been noted.

A few weeks ago on Lateline, Andrew Bolt quite reasonably suggested that, since Gerard never chances his own arm at election results, he should forfeit the right to "be a smart arse afterwards". Henderson declined the offer.

Gerard is actually a little naughty about these things, at times favouring the retrospective prediction. For example, I heard him on Radio National last year (or so) explain that unlike most people, he had always expected the Latham experiment would "end in tears". He just didn't say so at the time.

Similarly, he expected John Hewson to lose in 1993, but only revealed this afterwards.

Anyway, a few months ago, Henderson did make a prediction - of sorts. He told Insiders that at the next election (after this one) people will pay less attention to the opinion polls. The clear implication is that the polls this year will be proved wrong.

So Gerard does chance his arm at times. Let's see how he goes. (Elaborated here.)

[Update: for the first time in living memory, certainly since the early '90s, Gerard did not produce a post election piece in 2007.]

Prediction still stands

By the way, my prediction of a 30 seat majority for Labor still stands. Will give a final one on election day morning; it could go up or down.

November 11 Galaxy says 50 50 in Wentworth

Read Mr Milne in the Sunday Telegraph. 

Galaxy found primary support of 44 for Malcolm Turnbull, 36 for Labor's George Newhouse, 14 for Greens and (by process of elimination) 6 for everyone else.

As you know, Galaxy estimates its two party preferred numbers based on how they reckon minor party and independent preferences are likely to flow. So they might have taken the Greens' 14 and split it (say) 11 to Newhouse and 3 to Turnbull (ie about 75-80 to Labor, 20-25 to Liberal), and divvied the 6 "other" down the middle.

Primary numbers like that could really end up as anything from 51 to 49 to 49 to 51, and so 50 50 sums it up very well: it could go either way.

Note that Galaxy's recent NSW marginal seats polling showed an eight percent swing from 2004, compared to this four or five in Wentworth (based on Turnbull's probable "real" margin of about 4.5).

For more on Wentworth, see Pollbludger here and here, Antony Green and Crikey. See also my Wentworth - Dobell comparison.

Higgins and Dunkley Along the lines of this, this and this, graph at right has two party preferred Labor votes in the Victorian seats of Higgins and Dunkley from 1983 to 2004. Victoria is the red line. Click for larger

One is a posh inner city leafy Lib, held by the Treasurer, the other a classic outer suburban mortgage belt.

These graphs are adjusted for redistributions, and Labor's actual 2pp in Higginsin 1983 was only 41.5; the party has gained about 5 and a half percent in redistributions since then. The green line roughly moved roughly in line with the red (state) one, but at the last few elections the gap has narrowed a little. 

Dunkleywas created in 1984 (the first graph point is its notional vote before that election), and while redistributions initially favoured Labor, overall they have favoured the Libs by about one percent. 

In classic outer suburban style, Dunkley has moved substantially towards the Libs over two decades, and the sitting member now has a larger margin than Peter Costello in Higgins.

Just as I favour Labor to take North Sydney before Hughes, I also favour them to take Higgins before Dunkley.

November 10 Nielsen online poll: 56 to 44

In The Age

Former leader offers thoughtful advice Mark Latham used to flay Kim Beazley with the argument that the electorate won't vote for you if you don't stand for anything. He received a lot of support for this view both inside and outside Caucus. But then he got his wish and became leader, clocked in the worst election performance by any federal opposition since 1977, had a hissy fit, explained that the world revolved around him and resigned.

His mantra was then tweaked to: "what's the point of winning an election if you don't stand for anything?" and despite being a little to the right of Genghis Khan on most issues, now portrays himself as the "progressive" conscience of the party. 

But policy-wise he wasn't much different to Rudd, he just made a song and dance about everything. He believed opposition leaders have to muscle up, force their way into middle Australia's living room, fight on values and so on.

In this way he bought into the commentators' narrative, and this was why people such as Paul Kelly adored him so.

November 9 Sad, delusional left-wing bloggers

In the Oz, Samantha Maiden includes me in a gang of pathetic, deluded ... well, see headline above.

Over at ABC website, the lads from Youdecide hit back, noting some projection.

The odd couple From Garrett-Turnbull debate.  Click for larger.

November 8 Newspoll in late October/early November

Here's the PDF, combining the two most recent Newspolls.

The previous one had aggregated a 58 to 42 and a 54 to 46; this one includes the same 54 to 46 plus the most recent 53 to 47. 

Note the last two combine to (rounded) 54 to 46.

Punters still know best (world still mad)

The betting markets shtick reared its head this morning on Radio National, with Centrebet's Neil Evans telling Fran Kelly that, based on current betting trends, Labor will gain 17 to 20 seats. 

Almost a hung parliament! 

Hopefully after election-night we can stop this "punters know best" charade.

The "blabbermouth journalists election"

Another journo heard from.

November 7 Newspoll economic competence

In The Oz, here. 

Back in March, in Canberra Times (first pars), I noted that fortnight to fortnight polls showing big movements in voting intentions are likely to be accompanied by similar changes in secondary numbers - economic competence, approval, nicest haircut etc.

Two weeks ago, the first Newspoll of the campaign had a surprising 58 to 42 two party preferred, with the economic/security (and other) numbers also (relatively) good for Rudd. This survey generated long faces on Coalition supporters; some commentators claimed to base their new-found pessimism not on the 16 percent vote gap but on economic competence data (last part here).

The most recent Newspoll had 53 to 47, and it's not surprising that other numbers have followed. 

If the next Newspoll shows, say, 55 to 45, Rudd's economic etc numbers will probably go up again. (And if they don't, with only one and a half weeks until election day, will anyone pretend anymore to care about anything other than voting intentions?)

State by state, seat by seat - 1996 and 2007

Also in The Australian today, Sol Lebovic goes through each state. 

The headline is correct. At the last change of government, in 1996, seat swings went from almost 13 percent to the Howard opposition to 1.4 percent to Keating. (See this table, ignore Oxley and Wills where non-major party candidates polluted the 2pp).

And while all states went to the Coalition back then, their swings ranged from almost nine percent in Queensland to one and a half (Victoria). New South Wales, with a third of the seats, went by a little under seven.

Sol's final sentence is perhaps his wisest:

"If present polling levels hold through to the election though, not much analysis of state differences will matter with Labor romping in."

November 6 Newspoll in the Oz: 53 to 47

Tables here, Dennis here. The change from last week's 54 to 46 comes wholly from a one point primary shift from Labor to Greens. Statisticians will tell you this is statistically insignificant, and the fact that at least three quarters of Green preferences go to Labor anyway makes it moreso. 

(Newspoll indiscriminately allocates every minor party's/ independent's preferences 61 to 39.)

Antony Green update

A full set of Senate calculators at ABC election home page; see also senate group tickets and changes to calculator.

November 5 Two mentions in today's Australian

Late in the day, I notice two mentions of me in The Oz - one by Mr Milne, and one in 'cut and paste'. Both quote from this July piece in the same paper about the mixed political effect of interest rate rises.

Good Nicholson cartoon 

Here.

Galaxy says 54 to 46

Read Gerard McManus, Malcolm Farr and Clinton Porteous. Good on the Courier-Mail for generally being the only News Ltd tabloid to publish the Galaxy data online (and in hardcopy too, perhaps.)

Today's numbers here; primary support identical to the last national Galaxy, except Coalition down one and Greens up one.

My bets: Next year's John Kerry? About 4 years ago, I got 33 to one odds on John Kerry becoming the next president of the United States. He was at that stage well down the list of Democrat candidates; Howard Dean had easily the shortest odds (second only to GW Bush overall).

I reckoned that although Kerry was unlikely to get the Democratic nomination, if he did he stood a good chance of winning the presidency.

The rest, as you know, is history - apart from about 60, 000 Ohio voters.

A few weeks ago I saw a television interview with former Arkansas Governor and current Republican presidential aspirant Mike Huckabee. He had a pleasant, comforting, seemingly open manner, a good personal story - about losing lots of weight - and a name that conjures an American fictional icon (Huckleberry Finn).

So I threw some money at 40 to 1 odds on Huckabee to become the next president of those United States. Yes, all from one short interview. With similar reasoning to that which applied to Kerry.

[Update (a slight tweak): having learnt more about him (and his views), I now think he'd be unelectable as Republican candidate, but stands a good chance of becoming just that.]

New electoral blog on the block 

Read Tasmania's Peter Tucker.

Me in the Sunday Age

On the electoral roll, this time with Brian Costar.

November 4 Comparing the polls I: Bennelong and national

Today's Bennelong Galaxy poll in the Sunday Tele has a seven percent swing to Maxine McKew. That's a point more than the swing to Labor in the most recent national Galaxy. (And a bit more still if you reckon Howard's margin was artificially eroded by Andrew Wilkie in 2004.)

Comparing the polls II: Galaxy's and Newspoll's Qld marginals

Yesterday, Newspoll released marginal seat polling in four states. Recently Galaxy did the same in NSW and Queensland.

Newspoll's Qld seats were Blair, Bonner, Herbert and Moreton, Galaxy polled the same minus Blair but plus Longman. Galaxy's sample was 800; Newspoll's (from its 3.5% error margin) about the same. 

Newspoll had a ten percent swing to Labor, while Galaxy only had about six.

Comparing the polls III: Galaxy's and Newspoll's NSW marginals

Newspoll's NSW seats were Bennelong, Eden-Monaro, Lindsay and Wentworth. Galaxy's list only had one of those: Lindsay. The others were Dobell, Paterson and Robertson.

Newspoll had the pro-Labor swing at five percent, Galaxy had nine.

So a mixed bag, with Galaxy favouring, in relative terms, Queensland over NSW for the government, and Newspoll showing the opposite.

Galaxy in Lilley

And here's Galaxy's recent poll data in  Lilley.

November 3 Peter Andren dead: sad news

Discipline Say what you like about Swannie (tune here) - the "white bread politician" from central casting - at least you know that anything that emerges from his mouth will have already been practiced in his head - and practiced again.

And listening to Lindsay Tanner being interviewed these days is interesting: generally a three second pause before every answer, to make doubly sure he doesn't make the next day's front pages.

And then there's Peter Garrett. His press secretary is a Gen Y author who doubtless writes about the zeitgeist excellently and is a fun guy to have around, but ... this is politics. 

Doesn't head office have a spare Swan-clone - they've presumably got cupboards full of them - they can send to mind Pete for three weeks? 

On me-tooism II

Warning: this post has a bob each way.

It's a cliché firmly embedded in the campaign narrative (click cartoon above), and like most clichés contains some truth. Whatever you want to call it - "me-tooism" or "small target" - Rudd is more brazen than even Howard was in 1996.

Does it have a downside? It might. If you want people to believe it really is time to give the current mob the heave-ho, it helps to appear to strongly hold that view yourself.

Every time Rudd agrees with Howard, he further legitimises him (or at least boosts his flagging legitimacy). 

Minimising economic differences was wise, but perhaps at some stage it wouldn't hurt to stand up and have a blue - about something not too alienating, of course. Drawing the ethical line at the Exclusive Brethren a few weeks ago fit this bill.

On the other hand, maybe broadband, climate change, Iraq and IR are differences enough.

Newspoll in the marginals: best/mid case scenarios

Newspoll has surveyed 18 seats across four states - the most marginal in each state. (The tables contain 24 seats, but I think they've included 6 extra Qld seats their swing would take in.)

No online version of tables, but have copied (with permission) here. Read Dennis and Sol. Most dramatically, Newspoll has a much smaller swing in the NSW marginals than in Thursday's state-wide tables.  

Some niggling points:

• They've rounded their data and then calculated swings from the 2004 election to one decimal place, so you get numbers like a 9.6 percent swing in Qld, 4.7 in NSW etc. This is silly, because it implies greater precision than is the case - that 9.6 could actually be anything from 9.1 to 10.1. They should have calculated the swing using their unrounded raw data and published that - either rounded or unrounded.

• More importantly: below each state's table they show two numbers: how many seats would go to Labor with the swing their polling found (assuming uniform), and a "best case scenario" - what would happen with their swing minus the error margin (about 3 percent in each case). But there's no "worst case scenario" number - what the Newspoll swing plus the error margin would produce.

And in the piece, Dennis adds all the state numbers, calling the aggregate of the poll results the "worst case scenario". The "best case scenario" (he says) sees the government losing only 8 seats, while on the "worst case scenario" they lose 22. But that 22 is really the "mid-case scenario", and the real worst case scenario - if you assume all the swings are Newspoll's numbers plus error margin - comes to, by my calculation, Labor adding 45 seats to their 60.

• Numbers for Greens etc would have been good.

Anyway, all interesting. Combining today's tables and Thursday's state ones gives the following table (in rounded integers.)

Newspoll state-wide and marginal swings to Labor

There's little point just hoping/intuiting that the swing will all be in ultra-safe Coalition and Labor seats without reasons for doing so.

If there's a big swing in NSW, then the next most marginal seats - Dobell (4.8) Page (5.5) Paterson (6.1) and Cowper (6.7) - are all vulnerable. And don't forget Joe Hockey in North Sydney

November 2 Nielsen says 55 to 45

In Age and SMH.

On me-tooism I

You don't hear much criticism any more of "small target strategy", because it seems to be working brilliantly for Kevin Rudd. But then it worked well for Kim Beazley in 2001 until a certain boat turned up at Xmas Island.

But by August 2001 the government had already made ground; the final pre-Tampa Newspoll had each side on 40, with two party preferred about 52 to 48. As well, the government was only five years old then; now it's 11.

Warning: broad brushes used

Still, within three weeks of Tampa the government had gained 8 points on the 2pp, and for this reason I recently suggested that the first week in November might be a spot to watch for something big and emotional.

A Tampa-esque incident would involve, of course, people "not like us", with only one man having the courage and disdain for political correctness to stand up to them on behalf of real Australians etc.

But back in 2001 Howard just needed to get the "outer middle" back on board. His task is much bigger now, and that which goes down well in plasma-tellyland might have the opposite effect in leafy Liberalsville. It might get them over the line in Lindsay and Dobell, but give them grief in North Sydney, Wentworth and, perhaps, Bennelong. 

It might even trouble Peter Costello in Higgins.

Damned if he does, damned if he doesn't, perhaps.

Republic 'yes' vote: a measure of "leafiness"?

While we're on the topic, Pollbludger lists safe Liberal seats allegedly (ie according to leaked party polling) in trouble; most of them look far-fetched.

Not for the first time, Ryan in Brisbane gets a mention.

How might one measure/define cosmopolitan Liberal leafiness? How they voted in 1999 republic referendum might be an ok indicator. In Sydney, North Sydney and Wentworth recorded 'yes' votes of 61 and 60 respectively, while Hughes and Dobell registered 49 and 43.

Up in Queensland, only two electorates voted 'yes': the Labor-held Brisbane and Liberal-held Ryan. (All these are under 1999 boundaries, of course.)

Down in Victoria, Kooyong voted 64% 'yes', Dunkley 45%.

Coalition-held seats in order of 1999 republic 'yes' vote

Here are tables of today's Coalition-held seats in order of 1999 Republic 'yes' vote.

November 1 Rudd's required swing 

In the Oz, Black and Lockwood make a few pendulum points I recently made.

They also have a stab at Labor's required swing - they estimate 6 percent rather than the pendulum's 5. This would mean Rudd needs 53 percent two party preferred to win, which is possible - Labor needed 52 percent in 1998, and in 1987 the Coalition needed about 52.5 (see here) - but it does seem high.

One of the things to look for in the election wash-up.

Hughes and North Sydney In similar vein to recent Dobell and Wentworth analysis, graph at right has ALP two party preferred votes from 1983 to 2004, adjusted for redistributions, in Sydney seats Hughes and North Sydney, along with NSW.

Click for larger. 

These two sit near each other on today's pendulum - held by Libs by 8.5 and 10% respectively - but their paths there were very different.

Hughes used to be safe Labor, but moved south, particularly dramatically in 1996. Its change has largely been demographic - this table has Hughes as has the 8th highest median income in Sydney in 2006 - and people there tend to vote Liberal now.

North Sydney, with the highest income in the country, has moved a little the other way, against the state tide. What happens if/when the red (NSW) line moves above 50% this year will be something to watch.

Given the directions of those lines, it seems reasonable to expect North Sydney will move towards Labor more than Hughes in 2007. 

Prediction!

In fact, I confidently predict, Malcolm Mackerras style, that North Sydney will be more marginal than Hughes after this year's election. [Post-election update: dud prediction.]

Richmond and Bennelong This graph shows adjusted Labor 2pps for Richmond and Bennelong and NSW. Click for larger.

Richmond was safe Country/National - had been held by generations of Anthonys - until surprising everyone by going Labor in 1990 (largely on Helen Caldicott's preferences). At the last two elections it has been above the red line. Its change has been largely demographic - lots of "Balmain basket-weavers" moving up north.

The makeup of Bennelong's trajectory is more mixed. Howard's 1983 two party preferred vote was a comfortable 54.2, but adjusted for subsequent redistributions you can see it would have been line-ball. And since then the red and green lines have moved closer together.

It's mostly redistributions that account for Bennelong's marginal status today, but demographic factors - eg lots of Asians moving in - is also part of the story. 

McKew's required swing?

Recall Shane Easson, who reckons Andrew Wilkie's 2004 candidacy didn't really distort the 2pp. I'm not so sure, and would perhaps be inclined to knock a point off the Labor vote, which would mean McKew needs about a  six percent swing to take Bennelong.

October 31 Momentum? Lots of commentators reckon something called "momentum" is important in election campaigns.

Back in 1993, John Hewson's public rallies were considered great stuff because they were generating "momentum" towards victory. That was before the 1.5 percent pro-government swing on election night.

In 2004, people reckoned Mark Latham had momentum. Ok, he might not win, but he'll certainly close the gap. But at about 6:45pm AEST that "momentum" was forgotten too. 

Graham Morris has described how, when on John Howard's staff in the 1996 campaign, he was worried by shots on television news one night of Paul Keating being mobbed by schoolgirls, because they might create "momentum" for Labor.

You can sort of see his point, in that the idea that one side or other is "doing well" takes hold in the news and editorial rooms and generates more good news stories. But to believe that this translates into behaviour at the ballot box is a stretch.

Momentum, from what I can gather, is the general impression that you're doing well. To think this matters is to buy into the bandwagon theory. 

Whoever wins the election will be deemed to have had momentum.

My prediction

By the way, my prediction from a couple of weeks ago still stands: Labor to win 90 seats. May adjust up or down between now and 24 November.

Irony on Swannie was a revelation, weaving the policy and politics, telling a story, walking the tightrope, conjuring the word-pictures. His personality filled the room. Costello was charming and passionate and oozed gravitas, the breadth of his knowledge and interests astounding everyone. And funny too!

Irony off

October 30 Me in the Canberra Times

On whether "quality" candidates make a difference. (Answer: not really.)

Newspoll now says 54 to 46

The roller-coaster continues. See tables, Dennis and Sol.

October 29 Close of rolls: me in Crikey

With Simon Jackman, on the electoral roll numbers. 

October 28 Me in the AFR: ACNielsen in 2004

Last Monday, in the Financial Review, I constructed a straw for Coalition supporters to clutch (although yesterday's online Nielsen didn't follow the script).

Piece dummied up here.

October 27 Close of rolls: two sides to youth story?

The close of rolls numbers were released yesterday, and two newspaper articles today focus on the youth component.

George Megalogenis in the Oz presents the rosy picture; Paul Bibby in SMH is a glass half empty kind of guy. 

The truth, you'll be surprised to know, lies somewhere in between.

Large post, moved off home page, read more more.

Online Nielsen

In The Age say 56 to 44. More here from last time. And here's a comparative table.

October 26 New elections website: Soapbox

This, from Melbourne Uni, looks interesting. Press release here, site here.

12 year old NSW Labor government

This is a good tactic in the state with a third of the country's seats and a government who has been there forever.

The (apparently) relatively high opinion poll support for Rudd in NSW is perhaps surprising. Wall-to-wall Labor governments should in general be working for Howard, and in particular in the states where they've been there the longest. 

Ruminated on this a while ago. 

October 25 Postal vote applications

This Crikey piece today, by Jehane Sharah, on postal vote applications, is interesting.

It's actually got nothing to do with the secrecy of the ballot - ballot papers go straight to the AEC - but it is problematic. In America there are often mutterings about parties and their agents throwing out applications from people they suspect will vote for the other side.

And I must admit that when we received Senator Humphries' forms in the mail I thought, at first glance, from the envelope, that they were from the AEC. Can't recall why, something about the design.

Labor does it as well. It is a bit of a worry.

Galaxy in the Senate

This, in the SMH (and Brisbane Times), is a strange result, with a very low Labor vote. For some reason they've left in the undecideds, which artificially deflates all parties' support, but even extracting them would leave the Coalition way above Labor.

Here's the Getup press release.

Me in Crikey

On punters' seat betting. Related to below post.

My bets: Dobell and Wentworth

Graph at left has Labor two party preferred votes from 1983 to 2004 in NSW (red), Dobell (blue see Antony) and Wentworth (green,  see AG). Both seats are adjusted for redistributions, and Wentworth's most recent number is adjusted from 47.5 down to 45.5 as per Shane Easson's intuitively sound estimate.

Click graph for larger image.

Dobell has gone from being safe Labor to safe-ish/marginal Liberal, while Wentworth has gone from safe to marginal Lib. Employing a very broad brush, Dobell's evolution has been largely socio-demographic - people who live there tend to vote Liberal now. It used to be comfortably above the red line, but is now below it.

But Wentworth's has been largely due to redistributions, which is why the green line in 1983 is a touch above the 50% mark. In reality the 1983 Labor vote in Wentworth was only 40%, but roughly 10% worth of redistributions over two and a bit decades mean that, under current boundaries, Labor would/might have won the seat in 1983.

If Labor wins this election the red line will surely go above 50%, and, to an extent, drag the other two with it.

Dobell and Wentworth have similar margins (to each other) today, but you can get over $3.00 for a Labor win in Wentworth and over $3.00 for a Coalition win in Dobell. 

I reckon those odds are the wrong way round - Wentworth is more likely to fall than Dobell - but that  punters are overly influenced by the fact that Labor has recently held Dobell but never held Wentworth. Elaborated in today's Crikey.

Anyway, I've taken both $3 bets. If Labor wins both seats I'll be a little ahead.