June 15 Prospect: going, going ...?

Peter Hartcher in SMH unveils Kevin Rudd's Russian doll metaphor ... again. I think he likes it. Others might find it glib.

Peter's main subject is Chris Bowen, the member for Prospect in Sydney's west. 

This graph, of Prospect two party preferreds from  1984 to 2004, is adjusted for redistributions. As you can see, Prospect took two large swings in 2001 and 2004.

Former member Janice Crosio's personal vote would have contributed to the 2004 movement. Coalition 2pp is now higher than at any time over last two decades, probably longer.

Maybe in a decade or so Prospect will be Liberal held and North Sydney Labor. (Also depends on future redistributions.) Seats change.

Another Westpoll Reports of a Westpoll in the West Australian today showing ... something or other in marginals. We await details. If it's about 400 respondents spread over three seats then appropriate response would be chappy at right. [Update: see below, it's about 400 in each seat.]

[Update: Here it is in The Age, complete with a reference to Labor needing "17 seats" to win. This bit of Prime Ministerial miscalculating has been dealt with by Bryan Palmer and Charles Richardson. It's now become a "fact"; this is how rumours start.

The article alleges that WA will be "pivotal to the election", and

"Labor strategists believe the party must not only hold its five seats in the west, but win two others to have any chance of winning the election."

But the chances of this election coming down to two seats are miniscule, and WA is one of the least important states.

All seats are equally "pivotal", but statewise SA and Qld are much more important than WA. The true importance of these recent Westpolls is if, like recent Galaxys, they are revealing national movement back to the government. That's the worry for Labor.

[Update 1pm: Westpoll data

I now have the Westpoll data.

Samples 415 in Cowan, 427 polled in Hasluck and 409 in Stirling, for a total of 1251.

The results were Cowan: Coalition 51.5 to 48.5; Hasluck: 52 to 48; Stirling 53 to 47

2004 results were, respectively, 49.2 to 50.8; 51.8 to 48.2; and 52 to 48.

Because of small individual seat samples, I'd be inclined to add them all up and say Westpoll gives a status quo situation in those three seats overall - or small movement to the Coalition - compared with the last election. Which is what their last statewide one said. More Westpoll data here.

After all that: pollbludger has the data, much prettier, here.]

Australian Commissioners

Commissioners are, in theory, supposed to be independent of the government of the day. They occupy statutory positions and so differ from public servants. Federal Police head Mick Keelty is a commissioner; so is Ian Campbell at the AEC.

But in the end they're probably only as independent as the current government allows them to be - and how willing they are to stand up to them. There seems, for example, to be a history of State Auditors-General (also statutory positions) being immune to pressure and giving governments grief; perhaps rotten relations with governments doesn't matter much to Auditors-General.

One problem with electoral commissioners is the method of appointment: see the fourth column in this table. Appointment by Governor-General/Governor just means, of course, the PM/Premier/Cabinet chooses someone. 

Appointments that legislatively involve the parliament and /or consultation with other political parties are better. Of course, it depends on what you call "consultation" and the makeup of parliament, but South Australia's appears the best: advertising the position, consultation, a committee, both houses of parliament agree (and PR in the upper house, so not usually a government majority there). 

At the very least, a Commissioner who doesn't feel psychologically beholden to the generosity and wisdom of the current PM is preferable.

June 14 More on Galaxy

In Tuesday's Crikey, pollsters Irving Saulwick and Denis Muller flayed the issues questions in Galaxy's recent polls. They didn't pussyfoot around.

(They and others are calling it "push-polling", but push-polling is something else: purporting to survey but really just spreading rumours about someone/something. It aims to directly influence the respondent, and through her/him others in the community. Under push-polling, the actual results are not necessarily published; they is secondary. See Wikipedia.)

I have Crikey's permission to reproduce the Saulwick/Muller item here.

And this: my contribution on the same topic the following day. 

June 13 Strange substances remain in WA water It takes a while for news to travel, but apparently a Westpoll in the West Australian shows the Howard government ahead in that state 56 to 44. Similar to the 55.4. to 44.6 at the last election.

(I've rounded Westpoll's numbers, because the paper is still ...  well, see (2) here.)

The sample size is 400. That's pretty small, but doesn't mean we should dismiss the numbers entirely. The government's support seems to be holding up in WA.

Patterson's numbers

But check out the strange data under 'Sample Error' at the website of Patterson, who conducts the Westpoll.

(This slightly numbers-heavy post can be found here.)

Betting market update In the latest in a series of coincidences, a relatively good poll for the Coalition is followed by movement in puntersville. Centrebet now pays $1.85 for a Labor win v $1.90 for a Coalition one.

June 12 Galaxy in Queensland

In Courier Mail, gives federal two party preferreds 52 to 48. Here's the data. (Note the numbers given for the 2004 election are wrong; they appear to be the 2001 results.)

The last Galaxy Qld poll, in February, said 55 to 45, a swing to Labor of 12 percent on the 2004 Qld component. At that time, national polls generally showed about a ten percent swing.

Today's survey describes a Queensland pro-Labor swing of 9 percent, against last week's Galaxy which had a national swing of 6. 

So you could interpret this and last week's Galaxys together as saying (1) Labor's national vote has dropped; and (2) the swing in Qld is still higher than the national one.

(The Courier Mail still has trouble counting; plotting these numbers against the pendulum would give Labor 17 out of 29 seat - which is "at least double" the current 6.)

June 11 Senator Peter Andren? William "the Pollbludger" Bowe has posted and opened a thread on Peter Andren's clever/ brave/ interesting/ seriously loopy (we'll know which later in the year) decision to run for a NSW Senate seat.

(William describes me as "... excellen[t]". Cool.)

Pollbludger's Andren post here.

Milne on Malcolm

Glenn Milne in the Oz nicely summarises Shane Easson's Wentworth paper. (However, Glenn's claim that Turnbull will "certainly" lose the seat at either this election or the next is not made by Easson, who anticipates the next NSW redistribution will be after 2010.)

But there is something else that is likely to work Malcolm Turnbull's way.

Another Wentworth thing: the new electoral laws

In the paper, Shane notes, apropos of Turnbull's potential to build a personal vote, that the 2001 census ranked Wentworth fourth in NSW in the proportion of people moving residence. 

Nationally, the ranking is a less stark 27th - see this sorted excel file from aph website (2006 data out soon!) - but that still puts it in the top fifth. And as Easson says, "the recent redistribution which added Kings Cross to Wentworth can only have added to the turnover rate".

Two of the big items in the government's new electoral laws are (1) closing the rolls on the day the writs are issued (and three days later for people changing enrolment details - but not those who have dropped off the roll); and (2) heightened ID requirements for doing these things at any time.

Drivers license

The latter, as I understand it, makes getting on the roll or changing your details easier for driver's license holders (80-90% of electors?), but a significantly increased hassle for those without. I don't have data on licenses, but the Cross, Paddington, Darlinghurst, Surry Hills etc are, like many inner city areas, not parking-friendly, residents often have little use for vehicles, they're not economical.

Folks in these suburbs are disproportionately prone to getting hit by either close of rolls or new ID requirements (or both), and it's fair to reckon a decent majority of them would vote left of centre if they made it to the ballot box.

Usually around the country, these sorts of areas are in safe Labor seats, but not in this case.

Attempting to quantify this is too difficult, but if Labor just misses out on Wentworth, expect a big stink.

Here again is Shane's paper.

June 8 Malcolm in a muddle? Is Malcolm Turnbull in trouble in Wentworth? Peter King's 2004 run shaved the margin to 5.5% and the subsequent redistribution knocked another (estimated) 3% off. But surely sunny Wentworth will remain as blue-blue Liberal as that sparkling harbour it overlooks?

Last year NSW Labor fellow Shane Easson penned this piece on John Howard's chances in post-redistribution Bennelong. I think it remains the most detailed published analysis on the topic.

Now Shane's taken his fine tooth comb across to Turnbull Territory to assess the ALP's chances there.

His main conclusions: (1) Future redistributions are going to continue unloading Liberal voters into Peter Garrett's rather safe Kingsford-Smith and picking up Labor-ish ones from Tanya Plibersek's much safer Sydney. (Both are bad news for Malcolm.)

(2) Secondly, Shane attempts to extract the Peter King and other effects to estimate a notional margin for Malcolm going into 2007. He puts it at 4.5%. 

So if the NSW swing to Labor is greater than this, Turnbull is in trouble.

This is detailed stuff but step by step and easy to follow. 

Read it here. [PDF]

No Newspoll on Tuesday: 2 weeks in the sun

This weekend is a long weekend. Therefore, I have been informed, there will be no Newspoll published next Tuesday. Pollsters often don't survey over long weekends because many people are away and the sample is skewed.

This means that Newspoll (and Nielsen, due the week after next) will arrive a little over a fortnight after Monday's Galaxy, which showed 53 to 47. That poll altered the narrative, stopping the "Costello for PM" campaign dead. The word came that Howard was back in the game, and all reporting since then has been through that prism.

You're right, it's silly, but that's the way it is.

The latest Morgan, to be released later in the day [update: here, taken last weekend], has, I hear, a small movement back to Labor. (A few days before Galaxy, Morgan too contained movement to the government.) But since 2001 commentators, rightly or wrongly, don't pay Morgan as much attention.

The next regular polls will tell if 53 to 47 is the new ballpark.

In the past six months, the "Rudd's honeymoon is over" yarns have only lasted a few days before being pushed over by some poll or other. The week after next, Howard will have enjoyed two weeks as "the Master" who may well back back from behind once more. (Note: Centrebet's Labor win payout has continued to move up, now to $1.78. [update 7 June: $1.80.])

Whether all these things favour one side or the other is anyone's guess. The momentum effects of Galaxy were bad for the Treasurer's ambitions, but not necessarily for Rudd's. 

(Of course, factoring in your own views on whether Howard or Costello would be harder for Rudd to beat adds another dimension.)

Mr Keating's advice

Paul Keating's performance last night on Lateline was funny, effervescent, choc-full of chutzpah and with lots of those word pictures. (He seemed at pains to, uncharacteristically, share credit with Bob Hawke for all those economic reforms.) His reference to Joe Hockey was laugh out loud funny.

But is his advice to today's ALP of much use?

One of his pet recommendations (although not last night) is that Labor take massive personal tax cuts to the next election - put a three in front of the top rate. 

As noted last year, this would surely lead to disaster - "where's the money coming from?" "here come higher interest rates" etc.

The tax idea says it all. Keating has never been opposition leader. His stardom came in government, as Treasurer and PM. With the gravitas and legitimacy that come with incumbency, he hammered away, beat up opponents, displayed flair, finesse, flicked switches etc. 

But only incumbents can behave like that. An opposition leader who tried to  would look pathetic and deranged.

Being in opposition is different to being in power. The one Labor win of the 1980s and 90s which potentially holds lessons for today's Labor is the first one - in 1983. But Keating only looks to the later ones.

(He did have a point though about the need to address interest rates.)

June 6 Mug punters: me in Crikey

Here. 

Andrew Leigh, ANU economist and blogger, has a few words to say in response here.

(Andrew notes that I said much the same (including allusions to sheep) here a few months ago).

Update: "Leopold", who has obviously been following the betting numbers more closely than me, notes at Andrew's blog that I

"made a small error - Centrebet had $1.63 [not $1.69] before the Galaxy was published, and had moved Labor’s way for nearly two weeks before that poll was published."

But after Galaxy's publication on Monday Centrebet turned around and is now at $1.72. Furthermore,

"there was a ‘budget bounce’ in the betting market, which unravelled with the emergence of polls showing no such thing. ... The punters initially thought the budget was great for the government, then, coincidentally, reversed their view at the same time as the polls suggested otherwise ..."

June 5 Labor team: be afraid? As you know, the government's penny has dropped that every time it puts the boot into Kevin, his popularity rises. So they have turned the spotlight to his team: if elected they'll run amok, do unspeakable things to the economy.

It's not a bad tactic, except the other half of the story is missing: that Rudd is a weak leader who would let his ministers push him around.

Such a thing might have worked under Beazley (I must admit). But while Rudd is seen as both safe/conservative and tough/ruthless, it's not likely to yield many dividends. 

(Recall immediately after December's leadership vote, Gillard and her backers attempted to create a double-headed leadership. Such talk stopped pretty quickly.)

Mind you, I still hold to this.

June 4 [update] More on today's Galaxy poll

Got hold of questions and data for today's Galaxy. The Terror hardcopy didn't bother with them, maybe the other two News Ltd tabloids did. (Nothing online.)

Here they are. As you know, Galaxy takes their preferences seriously when calculating a notional two party preferred. But a 2pp of 53 to 47, from primary support of 44 to 42 and Greens on 10 looks a bit narrow. Maybe 2 or 3% of the 4% "others" were Family First.

On the other results, eg that 42% are afraid of too much union influence under Rudd, and the same number worried about interest rates, I heard Fran Kelly on radio this morning noting that wouldn't you know it, these were the same issues the PM has been hammering.

No coincidence, that's why Galaxy asked them specifically, extracting a 'yes' or 'no' to the propositions (scroll bottom). It's rather leading, and I'm surprised the levels of agreement from Coalition supporters weren't closer to 100 (rather than 73 and 68 respectively).

Still, just what the Howard doctor ordered after the weekend's bout of Costello-itis. [End update.]

Hope glimmers for the government

Galaxy in the Tele says 53 to 47.  That's the closest anyone's given it since, I think, December. Future polls will tell us if this means much.

(Malcolm Farr incorrectly claims Galaxy was "the only poll to correctly predict a comfortable Coalition victory in the 2004 federal election." The other, of course, was ACNielsen.)

On switching horses About a year ago I gave these reasons in favour of a change to Costello, and then these against. (In between I made this other leadership post.) I think they still stand.

I would also add in favour of a change the 'honeymoon': wall-to-wall attention, Tanya on the cover of the glossies, Peter doing the Macarena again, showing he has interests apart from numbers etc. 

At the very least it would shift attention from the Ruddernaut.

June 3 2007 Independence of Australian EMBs (from my other hat)

Australia today has nine Electoral Management Bodies - the federal one and one for each state and territory. 

(By comparison, India's Election Commission runs all elections, state and federal. Something like this was contemplated here - and pretty quickly dismissed - after federation in 1901.)

A couple of years ago I began a paper that I thought might also evolve into a thesis chapter. It looked at independence of Australia's EMBs, specifically the on-paper independence of the Electoral Commissioners - method of appointment, tenure, how easy to sack.

It no longer fits into my thesis, and remains uncompleted, possibly never to be completed.

At the time, I spoke to a few Commissioners and others, and the consensus was that the law itself was secondary, more important was the inherited set of practices - how far ministers were prepared to push things and how ready the Commissioner was to give minister the finger when appropriate.

(For an early, fascinating example, see section 4 of this extract from the 1914 Royal Commission in Election Law. The Labor Minister King O'Malley tried to dictate who would and would not work at particular Tasmanian polling booths at the 1913 election. While the Chief Electoral Officer rolled over immediately, officials further down stood their ground and O'Malley eventually backed off.)

This table lists features of each Commissioner's statutory independence, and attempts to determine which provides more most independence. It also doubles as a handy link to current bodies and legislation.

• Go to paper

• Go to table

May 31 Waiting for 'iggins

SMH analysis the other day of consolidated Nielsen data revealed, unsurprisingly, that a ten-plus point swing to Labor would see buckets of Coalition seats fall, including (in uniform terms) the Treasurer's.

Although as pollster John Stirton said: "this is not a prediction". 

Higgins has been going the way of North Sydney - slowly to Labor, but probably too slowly to shift across the line this year.

Below is the (unadjusted for redistributions) Higgins bar chart. In 2004, the 2pp was about the same as in 1984, although Labor's national and Victorian votes were both down about 4 percent points. 

But getting over the 50% mark in 2007 looks a Herculean task.

On the other hand ...

Are Higgins and North Sydney the next Lindsay et al?.

Let us go back in time. Three seats - Hughes, Lindsay and Macarthur - went dramatically to Howard in 1996. The outer south-west Sydney neighbours were the biggest swingers in the country outside Qld. Commentators whacked a silly "battlers" (and allegedly high quality local candidates) label on it, but they were (and are) mainly distinguishable from other western Sydney seats by their accumulating wealth.

And with hindsight, their behaviour at the previous election was an indicator of things to come.

In 1993, Keating's last hurrah, the country swung to Labor by 1.5 percent, and NSW by 2.2 percent. But these three shifted only a little or moved the other way. Hughes and Lindsay went to Labor by half a percent each, Macarthur went to John Hewson by over a percent. Safer (and poorer) Labor seats further in swung to Keating big time, eg Watson by over 6 percent.

Then in 1996 the country swung to the Howard's Coalition by 5 percent, and NSW by almost seven percent. The trio went by more than 11 points each.

Pattern?

If one is overly interested in patterns (as one should be), then one might ask: are seats like North Sydney (notwithstanding this) and Higgins the next big swingers? North Sydney moved to Labor in 2004 while NSW did the opposite; Higgins swing just .4% to Libs while Victoria went by 3%.

Are they (and others like them) preparing to bolt in 2007?

Higgins probably won't be publicly polled before the election, but North Sydney might be. Something to watch out for.

May 30 Me in Crikey on recent polls

May 29 Newspoll, counter-intuitively, says 60 40

 In the Oz.

May 28 AEC launches Enrol to Vote Week

With, for example, a banner ad at top of the Oz.

Earlier today I whinged under this heading that the link only went to the AEC's homepage which didn't, I claimed, have direct links to relevant enrolment pages. 

I was wrong, I hadn't looked close enough, the homepage does have those links.

There they are at top, next to a plug for the budget.

No, not the AEC's budget, but the federal one delivered earlier this month. In fact, it's the most prominent link on the page, and folks visiting to get on the electoral roll may, with a couple of clicks, read things like this:

"This budget includes personal tax relief worth $31.5 billion over four years to reward effort, improve work incentives (particularly for lower income earners) and enhance Australia's international competitiveness."

Do other statutory bodies flog the budget? Not the ones I quickly looked at - AFP, AIRC and  Productivity Commission. 

Curious.

Howard's approval rating

Bob McMullan in the Oz notes John Howard's approval has been dropping since Beazley took over in early 2005. Below graph backs him up.

Howard Newspoll satisfaction ratings Feb 2005 - May 2007

However, that's only part of the story. See his ratings over his whole 11 years of government, below. Apart from an initial honeymoon, the first five years were ordinary, and often awful. 

Howard Newspoll satisfaction ratings Mar 1996 - May 2007

What marked the turnaround? You don't need to ask, it was,  of course ...

Tampa

Graph below has last six years. Howard's approval remained high under Latham, even when he was trailing in the polls. As I've often argued, Latham made him look good because he represented a young, unstable version of the same package. 

Howard Newspoll satisfaction ratings Mar 2001 - May 2007

Under Beazley, and now Rudd, Howard's approval is approaching pre-Tampa levels.

Still, approval isn't much of a predictor of success. Just interesting, that's all.

May 26 Third Newspoll in a fortnight

This one says 55 to 45, compared with 57 to 43 in Tuesday's. Voting intentions in the two are actually identical apart from two points moving from Labor to the Greens in today's. As you (but not Newspoll) would know, this would make only about a half a percent difference to two party preferred numbers.

But 55 to 45 seems about right this time.

May 25 Release Brian? Poor Mr Howard ain't the only one who perhaps should have gotten out early. Spare a thought for federal Liberal Director Brian Loughnane.

In 2004, Brian ran a campaign that saw the best election performance by an incumbent federal government since 1977. Imagine bowing out with that as your record. 

But after this year his legacy is likely to be very different.

Party directors and secretaries take their responsibilities selectively. Labor's Gary Gray blamed all of 1996 on Paul Keating ("Captain Wacky"), while Bob Hogg was eager to claim credit for 1993.

The over-hyped Lynton Crosby performed the allegedly heroic feats of winning in 1998 and 2001. That is, he managed not to preside, in the best economic conditions in generations, over the first two-term non-Labor federal government since WWI.

Then, like many an ambitious Aussie, he took his talents abroad - to elect Tories in London and Wellington. Lynton's home, both parties still in opposition, the end.

But my favourite is Remo Nogarotto (left), who ran NSW Liberal campaigns in the late 1990s. After the October 1998 federal poll he was credited with keeping NSW marginals in the Howard government fold, despite a large state-wide swing to Labor, so saving the day.

They sang songs about him, he was feted as a strategic genius, a man with a great future in politics.

But only for a few months, because in March 1999 came the NSW state election.  An absolute shocker for the Libs; Remo copped much of the blame and if you google him now you'll find him somewhere in the soccer world.

A message there for Brian? Perhaps it's not too late to get out?

Polling out West Have the Oz to thank for this reference to Piers Akerman in the Tele, who quotes a Westpoll showing the federal government travelling well in that state.

I have no doubt that government support is holding up, relatively, in WA. But Mickey gets a guernsey here for two reasons. (1) A sample of 400 is rather small - with a margin of error of about 5%.

(2) Much sillier, though, is the West publishing primary support rounded to the nearest integer, as well as two party preferred support (presumably calculated from the rounded numbers) to ... not zero, not one, but two decimal places. (Eg "support for the ALP dropped from 49.75 to 48.95 per cent .. ")

They do things differently over there.

May 24 Mea Culpas? Leadership and IR

Some readers have invited me to recant on a couple of previous, strongly stated positions. They were: that Labor would do better at the next election under Beazley than Rudd; and that industrial relations would not cost the Coalition office. These are my responses:

1. On the leadership. In a sense a Beazley leadership would have been unsustainable, given its insecurity. But even if he'd had a clear run, do I concede Rudd was a better idea?

Perhaps. Certainly Beazley wouldn't be generating today's opinion poll numbers. But let's wait until election night. I believed then and believe now Bomber would very likely have won. If Rudd gets more than, say, 53 percent two party preferred, which might be the mid-point of Beazley's expected vote, I'll concede error.

2. On Industrial Relations. Yes, IR is hurting the government, but it remains a mixed bag. IR will very probably figure prominently in the campaign - and it'll likely be a negative for the ALP. 

It'll be Rudd and Gillard trying to bat away questions about union influence and how you unscramble the egg. If anything wins the election for the government, it'll be fear of change - in the economy in general, and IR in particular.

In summary: I believe that if Beazley was still Labor leader, and the government hadn't introduced Workchoices, Labor would still be heading for victory.

But I can't prove it, of course.  

May 23 Me in Crikey today re Mike Bailey in North Sydney

May 22 Newspoll says 57 to 43  

Here in the Oz. (Didn't we have one last week? )

Green support is at a low-low low 3 percent. Newspoll's wonky preference allocation - from the last election when Greens got 7 - overstates Labor's 2pp by perhaps two points. 

So 56 to 44 or even 55 to 45 are more reasonable headline figures.

Word of budget caution

The Sunday before the budget, ABC's 'Insiders' had a nice set of tables showing jumps in government support after 2001 and 2004 (election year) budgets. On both occasions they took (I think) six weeks to fully eventuate.

Probably the giveaways and goodies have gone the way suggested in this Nicholson animation. But budgets serve other purpose: reminding everyone of the government's strong perceived economic credentials.

Let's wait another few weeks for any 'bounce'. (If it does come we won't really know whether it was the budget's doing.)

New Crikey blog

Crikey has a blog going. Some good topics already. Here  

May 21 ACNielsen in SMH says 58 to 42

Here's a graph of the pollster over the last several years. It's missing the "Beazley becomes leader" tag in Jan/Feb 2005.

May 20 A tale of two seats

1. More Bennelong follies

Here's the reason for all the Bennelong fuss. The bar graph below is more or less a colourful representation of the table at bottom of last year's Shane Easson paper.

It shows Labor two party preferred votes for Australia, New South Wales and Bennelong from 1983 to 2004. (Unlike this line graph, the one below is not adjusted for redistributions.) 

Clearly, Bennelong (the blue bar) has become more Labor over the last two decades, especially relative to deteriorating NSW and national support. This has been caused by changes in both boundaries and residents' voting habits.

Put another way, the difference between Bennelong and the national and NSW bars is shrinking, and many people expect that the next time the purple and orange bars rise above 50, the blue one will be hot on their heels.

Malcolm Mackerras has reckoned in the past that the next time Labor wins an election, Bennelong will be part of it. (He's now tipping both this year.) Shane reckons it'll happen if Labor gets over 51 percent or so 2pp in NSW.  

2. Addicted to Dunkley 

Bennelong has become more Labor, but quite a few seats have gone the other way.

Some commentators in Victoria are addicted to Dunkley in outer Melbourne, their version of Sydney's Hughes/Lindsay/Macarthur. They reckon it's the "sort of seat" that Labor really "must win" if it is to have a chance. 

As this graph shows, Dunkley (light blue) has moved Liberal, also for boundary and demographic reasons. In raw terms, the movement has been more dramatic than Bennelong's, but relative to state/federal support they're probably on a par.  

I reckon Labor has Buckley's of taking it this year, that Dunkleyites will stick with the government who, over the last decade, has given them two four-wheel drives and that second home. But maybe next time, when Labor is the incumbent, will be another story.

People and the seats they live in change over time. If Labor wins the next election, expect post-mortems to declare the Libs in diabolical trouble because they have lost lots of "the sorts of people" they used to be able to rely on, eg those disparagingly called "doctors' wives".

(The 2pp graph, adjusted for redistributions, has Dunkley Liberal held since 1990, although in reality it was won by Labor in 1993.)

*Since 1961, when Labor got about 49% under favourable boundaries

May 18 Me in Crikey on Wednesday

In today's Crikey (sub required), Malcolm Mackerras repeats his prediction of the Bennelong result at the next election: Howard will go down. Malcolm also refers, in generous terms, while disagreeing, to a piece I had in the same organ, on the same topic, on Wednesday.

From the time-capsule

According to the Sydney Morning Herald, "The government is neither tired nor incompetent. Why then is it, apparently, fighting for its survival?" 

Good question. The time was federal election eve, 1996. Here's a Reuters roundup of editorials at the time. All were lukewarm on John Howard (and most had been more enthusiastic about John Hewson three years earlier) but all but two editorialised for a change in government. The two exceptions were News Ltd: the Oz refusing to recommend either, and the Terror backing Labor.

1996 - 2007

The first two Howard terms were unimpressive, characterised by political ineptitude. Howard supporter Michael Duffy, while urging his re-election in 1998, conceded that he had "not grown in the job of Prime Minister. He has shrunk in it ... [he] does not seem to have leadership qualities. He lacks the ability to inspire." (3 October '98, Daily Tele)

Yet despite its widely acknowledged mediocrity (and implementation of an unpopular tax), the government survived.

Today, everyone thinks the GST was a good idea, and few would describe Howard as a weak leader. And of course he and his government are infinitely more accomplished politically. You tend to get the hang of things after 11 years. Even the government's harshest critics would concede they're competent. 

So we can ask the same question: why is the government very probably going to lose?

Golden rule of electoral politics

The main answer, and I've stated it on numerous occasions, is one of the paradoxes of modern electoral behaviour. It is that voters need a good reason to turf young governments out; they need a jolly good reason to retain old ones. 

And to paraphrase the late Mr Packer, you only get one Mark Latham in your life.

May 16 Conviction politician! count

As you know, descriptions of our PM as a "conviction politician!" began with then Lib Federal Director Lynton Crosby's helpful post-mortems to journalists after the 2001 federal election.

A database search of the year 2007 to date counts 26 linkings of "John Howard" and "conviction politician" in our newspapers. Sixteen of them are in The Australian. Paul's byline is there, of course, and Dennis's, but also editorials and headlines to other articles.

Use of the phrase seems to have escalated in the last six months. Maybe they're clutching at straws at Holt Street.

May 15 Newspoll in the Oz

Says 59 to 41. See tables.

Notice the Greens are down to 4 percent, from 7 at the last election. Here's a piece on Greens support I had in Crikey a few weeks ago.

May 14 Another day, another Galaxy poll

This one, in the Hun, taken nationwide over the weekend. It has Labor still a mile in front with 57 to 43, a ten percent swing on the 2004 result.

By comparison, Bennelong (below) looks a bright spot for the government, with Maxine not travelling nearly as well as her party. Taking the numbers literally: if, for example, Labor wins the election 53 to 47, then Howard wins Bennelong 52 to 48. 

Life, of course, isn't that linear and simple. But I've said it before: celebrities and marginal seats don't mix.

May 13 Galaxy in Bennelong

In the Sunday Tele, says 52 to 48, from primary support of 46 to 44, Greens on 6. Based on preference flows at the last election, it seems a conservative 2pp for Labor; perhaps it's in the rounding.

Budget bounce

Taken after the budget, but before the weekend, it's a swing of about six percent from the last election, which compares favourably, for the government, with the ten-plus points that most national polls have shown recently.