Go to next posts   

+

=

        Mumble elections 

July 11 Morgan polls in NSW and Vic have governments way ahead in both

A New Shadow Cabinet

Meanwhile, in the federal arena, the opinion poll trend has settled in for a Howard-slide.

How can Labor turn the tide? What follows is a possible (and very partial) shadow ministry for a federal Labor team that actually saw winning the next election, rather than loyalty, sucking up to Sydney's west or doing things "the Labor way" as its top priority.

Opposition leader: Kevin Rudd. The best communicator of the lot, with verbal dexterity. Currently, for example, plugging away at PM's predilection for the porky - Howard's electoral Achilles Heel -  with some success. (Presentation is Crean's biggest problem.) Plus Rudd's from Queensland, which might be worth a few net seats. (More on that last idea here.)

Deputy Leader and Education: Jenny Macklin. Quietly making hits while avoiding the Lawrence/Kernot syndrome of huge expectations and disappointments.

Treasury: Wayne Swan.

Foreign Affairs: Bomber Beazley! Traditional spot for ex-leaders, suits his interests and knowledge, good for a chatterbox, allows him to visit his mates in the US. 

Health: Simon Crean. Good old-fashioned Labor cause with a position that, today, almost argues itself. Just needs a hard working thick-skinned shadow minister to keep plugging away with message and aggressively asking questions.

Immigration: Julia Gillard. (Obvious reasons)

Defence: Bob McMullan. 

Veterans Affairs: Mark Latham, to keep him out of trouble.

Said it was partial. Might add to it.

July 8 Yes, there it is. 53 to 47. 

July 7 Newspoll out tomorrow. If it's still stuck at something like 54 to 46 then we can call it a trend, towards a big Coalition win.

July 6 Immigration to Sydney was topic of article in yesterday's SMH (no link) and question from Barrie Cassidy to Simon Crean on "Insiders" this morning.

Migrants flock to Sydney in disproportionate numbers. The SMH piece predominately deals with demographer Bob Birrell. This is a problem, he says, for federal Labor.

Says the SMH

"[Birrell's] paper points out that if you  superimpose the federal electorate map of Sydney on the ethnic make-up, you get a perfect match: Labor contols all 14 seats with high NESB ratios. John Howard's Liberals dominate the rest ... 

"While the federal Labor Party has a firm fgrip on Sydney's ethnic heartland, this is hardly a satisfactory situation from the point of view of the larger electoral prospects," Birrell argues.

This, according to 2001 Census data, extracted by the parliament house library, is almost true. The table below shows only Parramatta and the PM's seat of Bennelong breaking up the Labor hold at the bottom. 

July 4 Morgan has 51 to 49, for the second time in a row. 

July 2 Quarterly Newspoll II - states

The Australian is going over the last three months of Newspolls again, this time with a looksie at the states. The headline: "Voters desert Crean on home front", ie ALP support in Victoria - with Tasmania the bright spot at the last election - has collapsed.

Once again they're not bothering with two party preferred. My calculations (as always, Coalition in blue, Labor red) are: NSW 51 to 49 (last election was 51.7 to 48.3), Vic 53 to 47 (last election 47.9 to 52.1), Qld 53 to 47 (last election 54.9 to 45.1), SA 57 to 43 (last election 54.1 to 45.9), WA 59 to 41 (last election 51.6 to 48.4)

The blurb says "individual state sample bases range from 1663 to 700". Presumably 700 is for SA, the smallest state. (Tasmania's number is so small the state is left out altogether.)

Do these numbers mean much? Possibly not. Read this, on Newspoll's state by state performance during the last election campaign. The figures are remarkably similar to those above, especially for the two bigger states, and they were, to varying degrees, wrong - most wrong for Victoria.

Admittedly, those figures, published two days before the 2001 election, were based on two surveys, not about six, as today's are, but here [warning: PDF, will open in new window - toggle or close to return] is the Newspoll record during the full campaign and the Victorian numbers were consistently way out.

On the other hand, a drop in Victorian support would be consistent with a theory about state federal voting habits that sees Victoria, having recently re-elected its Labor government with a whopping majority, now entering the phase NSW was in at the last federal election. 

I pondered on this in The Age after Brack's big win, saying: 

 "... something is afoot in the Australian collective psyche.

It might, just might, have something to do with Prime Minister John Howard. And either way, it is typically good news for him.

New South Wales followed up Carr's big win with a 4 per cent swing to John Howard on November 10, 2001, the largest of any state. The second biggest swinger was Queensland.

Don't be surprised if Victoria does the same at the next federal election."

The Victorian Newspoll component is wholly responsible for the change from a close two party preferred contest last year to a large government lead this year.

One shouldn't be overly deterministic about these things, of course, but still ..... see the following June 28 entry.

June 28 

Paul Kelly talks to Labor's "most successful leader, NSW Premier Bob Carr". Bob kindly provides guidance to his federal colleagues on what it takes to win elections.

Carr's advice may or may not be sound, but, being human, he probably thinks his electoral success is wholly due to his own astuteness at reading the voters' mood.

Now, he is a smart political operator, but his record should be read in context. That context is that voters in the four eastern states - Queensland, NSW, Victoria and Tasmania - have performed almost identical feats in the last few years. That is, ever so narrowly tossing out a Coalition (Liberal in Tasmania) government in favour of a Labor one, and then at the next poll returning that Labor government with a thumping, huge (I can't stress that enough), record-making (in Queensland and Victoria) majority. (Tasmania has semi-proportional representation, so while the seat majority appears modest, the vote was big.) 

The other two, SA and WA have only so far reached stage one: narrowly electing Labor, but it's a fair bet they'll follow suit.

A couple of months ago Carr had his second big win. We can expect the other three eastern states to do the same when it's their turn. (Published polls have them a mile ahead).

So something's going on, but what is it? 

Who knows? Perhaps the Howard government is satisfying urges in a large section of voters that would otherwise lead them to turn to conservative state parties. That's my vague theory, anyway. 

(It's not that Australians are sensibly, consciously splitting their ticket - giving Labor control of essential services but entrusting national security to the Coalition. For one thing, most Tasmanians and Victorians voted for a federal Labor government at the most recent opportunity.)

Within that context, looking at each's first re-election, Carr was the least successful and Peter Beattie the most. That might be consistent with an estimation of their respective political nouse and"it" factors.

Anyway, we're talking about winning from opposition, not keeping government. Can anyone remember Bob Carr the opposition leader?

This was touched on after the Victorian election here. 

June 27 Quarterly Oz consolidation of three months of Newspolls. The Australian sees Labor losing 11 seats. 

Just looking at marginal seats - where the action is - Newspoll has, for the period April-June '03, 46 for Coalition, 37 for Labor, 2 for Dems, Greens 7 and 8 others. They don't give two party preferred, but I calculate it (based on preference flows at the last election) as rounding to 53 to 47.  

Newspoll defines marginal seats as those with a less than 6 percent margin. There are 50 of those at present, and the average two party vote at last election for those was 49.6 to 50.4. (Labor's number is bigger because it has more seats with under six percent margin that the Coalition. See pendulum.)

That rounds to 50 50, from which Newspoll's 53 to 47 represents a three percent swing to the government. Track that against the pendulum (ie count all Labor held seats with a three percent margin or less) and you get eleven. That, possibly, is how Newspoll got that figure.

Anyway, it's a realistic number. It would increase the government's majority from sixteen to about 40.

The latest quarterly Newspoll consolidation is easily the worst for the ALP since November 2001 election. And it's not just one poll, it's half a dozen. The trend is setting in, and it's looking fine for the government.

(As an aside, the eleven figure wouldn't have come from individual seats. Newspoll surveyed 1874 marginal seat voters between April and June. There are exactly fifty of those seats, so the average sample size was about 40 people per seat. Without going into margins of error, this is way, way too small for any meaningful individual interpretation.)

June 24 Newspoll has 54 to 46. Yet another big Coalition lead and, importantly, becoming a trend that really does show the Coalition way ahead.

June 21 In The Age, straightening record of  former Labor leader

June 20 ACNielsen’s questions

Here are ACNielsen's questions. They get preferences.

Q.1a If a Federal Election for the House of Representatives were held
   today, which party would receive your first preference vote?
   READ OUT ALL. ROTATE 1 & 2
 
   1 Labor Party
   2 Liberal Party
   3 National Party
   4 Australian Democrats
   5 The Greens
   6 An independent candidate
   7 One Nation Party
   8 Some other party
   DO NOT READ OUT
   9 Don't know
 
   IF Q.1A= CODES 1 - 3 SKIP TO Q.2a
   IF Q.1A = CODES 4 - 8 SKIP TO Q.1C
   IFQ.1A= CODE 9 CONTINUE TO Q.1B
 
   Q.1b Which party do you have a leaning towards at present?
   DO NOT READ OUT
 
   1 Labor Party
   2 Liberal Party
   3 National Party
   4 Australian Democrats
   5 The Greens
   6 An independent candidate
   7 One Nation Party
   8 Some other party
   9 Don't know
 
   IF Q.1B = CODES 1 - 3 SKIP TO Q.2a
   IF Q.1B = CODE 9 SKIP TO Q.2a
   IFQ.1B = CODES 4-8 CONTINUE TO Q.1C
 
   Q.1c At a Federal Election you will be required to vote for all
   candidates in your electorate in order of preference. Given this, will
 you
   give a higher preference to the Labor Party candidate or the
   Liberal/National Party candidate?
   DO NOT READ OUT
 
   1 Labor Party
   2 Liberal/National Party
   9 Don't know
 

June 18 Saulwick's preferences

Another carry on about preferences! It's a long one, heavy on the numbers and not for the easily bored or those to whom too many figures gives headache. (That second part probably describes most people.)

The news is that the latest Fairfax poll overstated the Coalition's lead by six percentage points.

Last Saturday a Saulwick poll in the Fairfax papers, which dealt mainly with leadership, gave a whopping 57 to 43 two party preferred lead to the government. That's easily the biggest gap in any poll since the last federal election.

Here is the table. This will open a new window. It might also take a while. Close it or toggle to return.

What raised the eyebrow was the Coalition's primary vote lead actually increasing after preferences. In the current electoral landscape, this seemed most strange.

On Monday and Tuesday I spoke to Irving Saulwick. He explained his opinion poll preference strategy. His is yet another variation.

Let us recap. We've seen two approaches to preferences so far.

(I) Roy Morgan and ACNielsen's.

All pollsters, of course, ask: if a federal election were held, who would get your first preference vote? Nielsen and Morgan then ask a further question of those who have nominated someone other than Labor or Coalition: which of the Coalition or the Labor Party would probably get your preference?

(II) Newspoll
Newspoll only measures primary support and then estimates a "notional" two party preferred based on flows at the last election.

Regular visitors will know my interest in Newspoll preferences borders on obsessive. Read more - if you're game - on that here.

and now there is ... (III) Irving Saulwick & Assoc
I've discovered a much worse preference distribution strategy: that of Irving Saulwick & Associates.

Let us pause
Let me remind you why preferences are more important now than they've ever been. Put another way, the primary vote per se is less useful in determining probable electoral outcomes than it has ever been. Major party votes are shrinking. More importantly, Labor's are leaking to the Greens, but mostly coming back in preferences.

If you're not convinced, read the Newspoll blurb. See also preferential voting made easy.

Once upon a time it didn't much matter how pollsters got their two party preferred votes. If the gap kind of resembled the primary one it was pretty ok. But no more.

Continue ...
Here's what Saulwick does. They ask, of course, for first preference support. Then they ask the non-major party voters who their second preference will go to. That's it; they ask no more. Then they totally ignore all those voters who didn't indicate a major party as either their first or second preference.

This is a shocker.

Let's look at what they did with last week's survey, published on Saturday.

There were 1,000 respondents. Twenty four either refused, didn't know or intended not to vote. The rest responded, and below are the (weighted) numbers indicating who their primary vote would go to. (They add up to 1,000 minus 24 = 976.)

So far so good. Now, to get two party preferred vote - which is, remember, how individual seats are won in this country - those intending to vote for one of the last four (Dems to Other) were asked who would get their second preference. This is how they answered:

That is, of the 42 people intending to vote '1' Democrat, 8 said they'd put a '2' next to Coalition candidate, 15 said they'd put a '2'  next to the Labor candidate, and so on. 

No further questions were asked about who will get '3', '4', '5' etc.

Saulwick then added, for each major party, their primary vote and all the stated second preference ones. The remaining votes - eg Democrat voters who will give second preference to Greens or others - disappeared into the ether.

So Coalition's is 451 (from the first table)+ 8 + 18 + 20 + 5 (all from the 2nd table) = 502

Labor's is 316 + 15 + 33 + 10 + 5 = 379

Add 502 to 379 and you get 881.

379 is (rounded) 43 percent of 881, and 502 is 57 percent.

So there's Saulwick's two party preferred support: 57 to 43.

This preference distribution method is dreadful

To see what's happening, imagine a polling company that calculated two party preferred solely from primary votes - a kind of Saulwick strategy without the second preferences. You'd add 451 to 316 and get 767. Then you'd calculate each party's share of that, which comes to 59 to 41. Everyone would agree this is preposterous.

Taking this methodology to second preferences but no further is better - but not much. And if they took the second preference flows, and extrapolated them to all preferences, that would still be pretty unsatisfactory, but it would be an improvement. They don't even do that.

Saulwick overstates the Coalition's lead by between four and six percent.

If you're still not convinced, look for example at the Green second preferences preferences. The ALP gets its 33 vote share, and the Coalition 18 but the 18 that go to the Democrats and the 24 to either another small party or independent are just totally ignored.

Irving Saulwick explains that they don't want to make any assumptions about the flow of the rest. But assumptions are unavoidable, and ignoring the other votes is itself an assumption: that they will flow to the major parties in the same proportion as the votes already looked at, most of which are first preference ones.

This is worse than splitting them 50 50; it assumes about three out of every five will go to the Coalition - and is why Saulwick's 13 point primary lead to the government actually increases to a fourteen point one after preferences. 

On the contrary, they will favour Labor. Almost certainly, Green votes will strongly, and Democrat votes solidly, flow to Labor. (At the last election a five point Coalition primary lead shrunk to two after preferences.)

And with One Nation now flat-lining, the Coalition has precious few places it can expect preference from.

The human angle: Phillip the voter

Let's try to put a human face on this. Imagine you are a Green voter. Your name might be Phillip or Adam, or perhaps Phillip Adams. You're deeply angry with the ALP - mainly because of asylum seekers - and swear you'll never vote for them again.

It's election day, and Phillip goes into the cubicle and ticks '1' next to the Green. That part's easy. Now, does he put a '2' next to Labor? Not on your life - not those bastards, if something more palatable is on offer. So he goes through all the cuddlier ones - Democrats, Save the Whales, Legalise Marijuana etc etc until he has to make a choice between Labor, the Coalition, Shooters, One Nation and so on.

Now, Phillip hasn't gone mad. He's not going to vote for John Howard just to spite the ALP. So of course he puts Labor ahead of the others.

Saulwick ignores Phillip. But there are hundreds of thousands of him.

Nearly finished

Ideally, pollsters should ask respondents which of the major parties will get their preference - wherever on the ballot slip that is. As mentioned, ACNielsen and Morgan do this. But failing that, they should be allocated as they flowed at the last poll - as Newspoll sort of does.

If Newspoll had done the Saulwick survey - assuming identical polling methodology (apart from not asking the second preference question) - they would have calculated a notional two party preferred of  55 to 45.

And if preferences were distributed according to their individual flows at the last election, the result would be  54 to 46.

(This is the calculation:
Coalition vote = 451 + .36*42 + .25*93 + .51*(56 + 18) = 528
Labor vote = 316 + .64*42 + .75*93 + .49*(56+18)  = 449
Which gives two party preferred of 54 to 46.
Compared with Saulwick's 57 to 43.)

(I get those ratios here)

Ok, so we're talking about the difference between a wipe-out and train wreck, but when the vote's closer it makes a big difference.

End of lesson.

But one side issue from this is that preferential voting gives scope for plenty of mischief in presentation of polling results. None of the public pollsters have malign intentions, but plenty of others do. Specifically, clever party apparatchiks might tweak them to suit their purposes.

I'd love to see those internal party findings. Can't anyone anonymously send them to me?

June 15 

Yesterday's Saulwick poll in Fairfax papers has huuuge 57 to 43 lead to government. Primary support was 45 to 32. That means a gap of thirteen to the gumint is increased to 14 after preferences. 

Hullo hullo: this is peculiar in the current electoral landscape.

What did the minor parties get? The Age doesn't say. SMH doesn't even give voting intentions.

Answer must be one of 3 things.

• Saulwick is only measuring primary votes, and then just splitting the rest fifty fifty to get two party preferred. This would be too silly to believe so can be discounted.

• Green voters are indicating they'll be giving preferences to Coalition ahead of Labor. This would be a big, dramatic new development.

• The poll showed not many Green voters. That means there are a lot more of someone else whose preferences favour the government. One Nation? That's an even bigger development. 

Hmmm.

June 14

Two new tables: opposition leaders ranked on 

(1) Newspoll approval ratings and
(2) Newspoll voting intentions. Some surprises

June 13 

New Morgan poll on voters' attitudes to the Labor contenders

June 12 Death to the referendum

My opinion is in the AFR today calling for the referendum to end all referendums - changing the mode of constitutional alteration to something like the American model.

A couple of points. 

At last count Switzerland is the only other country silly enough to do constitutional change by referendum. Only they're not silly about it, they approach it with respect and maturity. Australians think it's a partisan game.

This is how the Yanks do it:

"The Congress, whenever two thirds of both houses shall deem it necessary, shall propose amendments to this Constitution, or, on the application of the legislatures of two thirds of the several states, shall call a convention for proposing amendments, which, in either case, shall be valid to all intents and purposes, as part of this Constitution, when ratified by the legislatures of three fourths of the several states, or by conventions in three fourths thereof, as the one or the other mode of ratification may be proposed by the Congress; provided that no amendment which may be made prior to the year one thousand eight hundred and eight shall in any manner affect the first and fourth clauses in the ninth section of the first article; and that no state, without its consent, shall be deprived of its equal suffrage in the Senate."

And nobody could accuse Americans of disrespecting either their constitution or federal system.

Canada, also with Her Maj as head of state, changes it through a request to Westminster (ie the Mother country).

Excising the referendum process (by referendum of course) is an extreme suggestion that' will likely get few takers, but constitutional reform is a farce in this country.

What, for example,  happens if the Brits cancel the institution of monarchy and Australians refuse to pass the necessary alterations?

Gough Whitlam in the Age supports the Howard Senate proposal, as you might expect.

June 10 Trashing Sydney (approximately part VIXX)

Steve Loosley, former NSW Labor Party boss and senator, on radio this morning, repeated his conviction that (paraphrasing) "Federal Labor only wins government when it does well in metropolitan Sydney." (Ie it's time to reconnect with battlers in Sydney's west).

Loosely seems to reckon that you build up such massive support in Sydney that it kind of over-flows into NSW regions and beyond, to the rest of the country.

Who said Sydneysiders are self-obsessed?

I tweaked a few numbers. The Labor two party preferred vote for all 27 Sydney seats at the 2001 poll was 50.5 percent, historically low.

However, the 1998 Sydney Labor vote - also a losing election - was 53.5%. That's higher than either 1990 or 1987 (53 and 51 respectively) which, for those with long memories, were winning elections for the ALP

In 1987 the ALP suffered a one percent national swing (most of it in Sydney) but ended up with a bigger majority (thanks mainly to Queensland) and 1990 was the only election Labor ever won with a minority of the two party preferred vote.

The idea that Labor only takes government (from opposition) when it wins huge two party preferred majorities in Sydney deserves two responses.

(1) It's happened so rarely (1983, 1972, 1929, ..) that if it's a strategy to take government it ain't a very good one.

(2) There have been plenty of elections at which the federal ALP got whopping two party preferred majorities in Sydney but failed to get enough seats nationally. There have also been instances of mediocre Sydney votes (specifically in 1987 and 1990, as above) in federal election wins.

One way of looking at 1987 was as an electoral turning point. Labor won with a national vote (50.8) that probably would have been insufficient at most previous polls. In 1998 (a loss with over 51 percent of the national vote) the party reverted to type - wasting electoral capital in Sydney. 

Conclusion? One of my hobby horses, that for a given national vote, a low Sydney component is good news. This applies to both sides.

Forget Sydney, look elsewhere - like Queensland.

Phew. That was longer than anticipated.

June 9 I have a contrarian piece in today's Canberra Times. Now, before reading,  remember that equivocations and shades of grey and opinion columns don't mix. I believe that Howard's so-called "stunning" political ascendency is ludicrously overstated, his lead in the opinion polls not as great as widely believed, Labor stands a much better chance of winning than is generally believed and ......

See? Not gripping reading. Much better to go all the way and write something brave and stupid like this.

June 8 Channel Nine's "Sunday" on Federal Labor woes included a not at all pretty glimpse at underbelly.

Said Graham Richardson: "I just can't see Labor beating John Howard". This got me searching the shelves for his book, "Whatever it Takes", published in the Keating government's last term, in which I'm pretty sure he says he just can't see the Liberals beating Paul Keating.

Couldn't find it, but did come across Peter Walsh's "Confessions of a Failed Finance Minister", which contains something similar. This bit, written in late 1994.

The Federal Liberal Party, somewhat like Labor in the 1950s and 1960s, does not know what changes to make ....  John Howard, for a decade the Liberals' best Parliamentary performer and most substantial on policy matters always looks better when he is not actually Leader of the Opposition. Peter Costello ... should probably be Leader now, but may not want to be until after the next election, which the Liberals are more likely than not to lose [my italics].

A decade on and the script remains the same.

June 6

Don't laugh, but Morgan reckons Labor would have won an election held in late May/early June.

They also find that most electors wanted Howard to stay on. Counter-intuitive, but nothing unusual in something like that. For example, this week's Newspoll found 65 percent prefer Howard as PM, but 49 percent would vote for Crean ahead of Howard.

June 5 Where have all the byelections gone?

There might be only one poll that counts, but byelections are invaluable (if costly for taxpayers) pointers to the electoral situation. Interpreted correctly, they can be seen as an 80 thousand strong opinion poll, which means a close to zero error margin.

They nearly always show a two party preferred swing to the opposition - recent exceptions were Jackie Kelly in Lindsay in 1996 and Carmen Lawrence in Fremantle in 1994 - and the only question is how much - 4 or 5 percent is usually expected.

We've had just one in the current term, in Cunningham, where the government chickened out. It's unusual to have so few in nearly two years, and is symptomatic of the tight discipline in the government and Howard's authority within it.

But doing poorly in a byelection isn't in itself a bad thing. To think otherwise is to confuse cause and effect.

 You could argue that the Ryan byelection in March 2001 - which saw a ten percent swing to Labor - saved the Howard government. It woke them up, got that  "mean and tricky" memo up and running and turned Howard's mind to spending and grovelling his way back into contention.

A mid-term kick in the teeth, and public display of government humiliation, is possibly a cathartic experience, not just for voters in that seat but the whole country, that has net benefits for a government - especially if it communicates it's "gotten the message".

With opinion polls showing a tight contest but most observers seemingly convinced that Howard the political colossus would easily win an election anytime, anywhere against anyone, a byelection might sort a few things out.

Ambassadorship, anyone?

June 04 Peter Costello, in refusing to rule out a challenge yesterday, deliberately used John Howard's own words re Andrew Peacock in 1984. Peacock suddenly resigned the next year and the leadership fell to Howard.

The Howard-Costello leadership tension is also compared to Hawke-Keating's.

There's a more apt parallel. In April 1981, Peacock resigned from the Fraser ministry in a destabilising tilt at the leadership which culminated in a spill the following year that Fraser won easily.

The parallel is in Peacock's resignation letter. It included word for word a phrase Malcolm Fraser had used in his own resignation letter a decade earlier to PM John Gorton.

Peacock 22 years ago was, like Costello today, seen as petulant. The third term Prime Minister back then was unassailable, having snatched victory at the previous poll with a wealth tax scare campaign.

There's an even chance (depending on bombs and wars) that Costello will become the next Peacock. Howard will lose the next election, and the number one leadership contender will have no choice but to inherit a defeated party and the thankless opposition leadership against a new Labor government.

The 64 thousand dollar question, of course, is who will be the next PM?

June 03 

Today this site was briefly hacked by Brazilian anarchists. All is hopefully well now.

Polls put Crean in the Lodge!

Newspoll and internal Labor polling

As you know, today's Newspoll has the same two party preferred result as the last election, 51 to 49. Also, apparently, internal Labor polling sees a 10 percent swing to the Coalition in Sydney.

This would see six Labor seats lost to the government! (So Labor's Sydney share would go from 14 to 8.)

Now, there are 27 Sydney seats and 150 in the country as a whole. A ten percent swing to the Coalition in Sydney, with the national aggregate remaining at 51 to 49, would mean a pro-Labor swing in the rest of the country - of about 2 percent.

A uniform two percent swing to Labor outside Sydney would see the non-Sydney share go from 58 to 67. Add the eight in Sydney and ... it's 75 out of 150! If they win Cunningham back from the Green Michael Organ they've got 76, otherwise they could rely on his support.

A deeply silly exercise, but it goes to show how over-rated Sydney seats are.

11:30am Howard to stay! 

Apparently he's told Liberal colleagues he intends to "stay on as Prime Minister". But what on earth does that mean? Does it mean 'til the next election? It doesn't stop him stepping down early next year; it just stops people talking about it between now and then.

I still reckon he'll go. Perhaps we should look at Peter Costello's reaction: if he appears sanguine we can assume a Kirribilli type agreement. 

Newspoll

Today's Newspoll has 51 to 49, which is what ACNielsen had two weeks ago. Can't tell if there has indeed been a change in preference distribution methodology, as using both correct and incorrect on the (rounded) primary numbers gives the same rounded two party preferred, given above.

(Preferences - rather boring - addressed vis a vis an earlier Newspoll here. Also mentioned in this AFR Lies and Stats.)

The Oz headline and lead par emphasise Crean's continuing low approval ratings, cue ABC radio declaring "another bad opinion poll for federal Labor." That's the news process for you.

A six point narrowing of the two party gap from eight to two is "bad".

Had the headline declared "huge leap puts Labor back in contention", you'd get something else. I've said it before and I'll say it again: just look at voting intentions; forget all that approval/disapproval and preferred PM rubbish.

Much reporting - in above Oz, SMH and Age - of a Beazley strategy meeting last Friday.

June 02 Newspoll out tomorrow. Perhaps a  change to notional two party preferred methodology? The Sydney Morning Herald mentions polls plural, suggesting maybe an ACNielsen as well.

More importantly, internal NSW Labor polling is dribbling out that points to big losses in Sydney. "At least six seats", according to ABC's AM, including Banks, Greenway, Lowe and Barton.

That's all Sydney seats with margins 6 percent or less on the pendulum. Werriwa, with margin of 8.5, is also mentioned as copping a big swing; if six Sydney seats are going to go, Werriwa would probably be one of them, and so would Kingsford-Smith (8.9).

It would be interesting to see the details of polls such as these - sampling errors and so on. The published ones - Newspoll, ACNielsen, Morgan - certainly don't show anything like a six percent swing to the government, either nationally or in New South Wales.

 (If anyone wants to send me anything on this, confidentiality is guaranteed)

Anyway, vote for vote, Sydney is probably the last place in Australia worth polling. Victory lies not in minimising damage there but in the marginals, most of them regional and in the other states.  Just look at the seats on the left of the pendulum.

Is the future for the ALP as a national party? Or will it continue to do really, really well in Sydney's west but only cross the federal treasury benches (in the desirable direction) once every several decades (1983, 1972, 1929, 1914 ...).

Big votes in safe seats in Sydney's west and not enough elsewhere was the reason for the 1998 loss with over 51 percent of the vote. Compare that with a 24 seat majority with 50.8 percent in 1987. Now that was an efficient use of electoral capital: swapping a couple of heartland votes for a strategic one elsewhere.

Battlers are over-rated. As was mentioned a year ago.