October 24 The power of polls II: Sun King As you know, Rupert Murdoch likes to back winners in advance to create the impression that he helped get them there. See, most obviously, Tony Blair in 1997. And as you also know, the Sun King recently declined to support either side in the Australian election race. Presumably his Australian intelligence told him it could go either way.

But yesterday's big fat  58 to 42 Newspoll shocked everyone, and perhaps jolted election expectations amongst The Australian gang. 

Dennis and Sol - Sol in particular - all but ignored the bleedin' obvious 16 point Labor lead, pretentiously basing their analysis on "crucial" esoterica - security and economics numbers. But both were very gloomy about Howard's re-election chances.

We don't know who Rupert gets his Oz election analysis from, but if he publicly falls off the fence between now and election day, perhaps we'll be able to trace it to yesterday's Newspoll.

Those "crucial" economic and security ratings

And the tables below show Newspoll's readings of who is better able to handle the economy and national security over the year. Do you think the change in either from March to October is worth getting excited about? 

Or is it, again, the fact that the opposition is 16 points ahead in voting intentions one week into the campaign?

WHICH OF (LEADER) DO YOU THINK IS MORE CAPABLE OF HANDLING AUSTRALIA'S ECONOMY?

	JOHN HOWARD	KEVIN RUDD	NEITHER	UN-COMMITTED
	%	%	%	%
16-18 March 2007	45	34	7	14
20-22 July 2007	53	28	3	4
17-19 August 2007	47	31	7	15
28-30 September 2007	48	33	5	14
19-21 October 2007	46	37	5	12

WHICH OF (LEADER) DO YOU THINK IS MORE CAPABLE OF HANDLING AUSTRALIA'S NATIONAL SECURITY?

	JOHN HOWARD	KEVIN RUDD	NEITHER	UN-COMMITTED
	%	%	%	%
16-18 March 2007	44	36	6	14
20-22 July 2007	50	30	5	15
3-5 August 2007	45	33	7	15
19-21 October 2007	43	39	3	15

October 23 Me in Crikey I: the power of polls

Poll-mix, including today's Newspoll.

Me in Crikey II

On Newspoll's economic rating.

Newspoll surveys the worm: 58 to 42

Tables here, Dennis here.

October 22 Win the vote, lose the election

The scenario of Howard losing the vote but winning the election has had a good run lately. (Gave it a bash myself recently.) It's not terribly likely to happen given the opinion polls, but if Labor's two party preferred vote drops down to between 50 and, say, 51, it then becomes very likely.

(Doubt Labor will need 52 percent to win. That's taking uniform swing too literally.)

Phil Coorey in SMH addresses it today, mentioning of course the 1998 result, which sits at the bottom of this table as the most extreme example in federal history. 

1998 result

The national swing to Labor in 1998 was 4.6 percent. As Malcolm Mackerras has noted, his pendulum didn't quite work, because plotting that swing would have given a Labor majority. 

Reasons include that Governments can pork-barrel the marginals and so on, but there's also individual seat incumbency. Here's a quick sum I'm quite sure someone has done before.

The Coalition took 31 seats from the Labor Party in 1996, and the average swing back to Labor in those seats two years later, in 1998, was only 3.6 percent - a percent under the national figure.

Why? At least partially because of personal votes of incumbents - working both ways. Labor lost the personal vote of their previous member, and the Coalition had a personal vote for their new one. And most of these seats, of course, were marginal.

(Not all those 31 seats had sitting Labor members in '96. For example, Kevin Rudd was contesting the Labor seat of Griffith for the first time. But unfortunately I don't have the data to extract all those out.)

So that's one reason marginals swung to Labor less than the rest in 1998. But as noted here, it's not a trick you can keep pulling forever.

And it has not gone unnoticed that the Coalition has quite a few retiring members this year.

October 21"Great" debate tonight 

Everyone rightly says Kim Beazley and Mark Latham won the debate hands down in 2001 and 2004. It might have been helped by the fact that both are big blokes with presence; they were well-prepared, on top of their facts and arguments and relaxed and revealed their warm sides. 

John Howard seemed to have nothing to reveal; instead, robbed of the crutch of prime ministerial authority, he just looked scared and shrivelled.

(Can't remember 1998; neither it seems can anyone else.)

Wonder how Kevin Rudd will go tonight? Given his determination never to say anything contentious, it might be more evenly matched: both sides rehearsed and charisma-free. But Rudd does do a decent line in self-deprecation. [Update: not evenly matched.]

Making a difference?

As everyone also says, these things don't make a difference to the election result.

Perhaps the only scenario that might make a difference would be an opposition leader cracking under the pressure and having some sort of melt-down on stage. That would be disastrous. (A similar mistake from an incumbent would be less serious; voters know they're up to the job of running the country.)

October 20 Reasons for Labor cheerfulness I

Mr Carney in The Age writes:

"Is the media's interpretation of the polls [as showing a significant movement to the government in the first week of the campaign] a genuine problem for the Labor Party and an encouraging sign for the Coalition? You bet."

I reckon the opposite, agreeing with folks such as Richard Farmer that widespread anticipation of an inevitable Labor victory would be bad for Rudd. Anything that dampens such expectations is good for the ALP. 

That is, I subscribe to the "underdog" theory. The "Howard's back!" narrative is a plus for Labor. (This is aside from any consideration of what the polls themselves do or do not say.)

Government members may indeed be high-fiving this weekend, but it will be a distant memory in five weeks if/when they're turfed out.

And speaking of what the polls do or do not say ...

Reasons for Labor to be cheerful II Here is poll-mix sans Morgan, who nearly always has Labor a point or several higher than the other pollsters. (Morgan may or may not be right.)

 Click for larger image.

Rather than the 'halving of the gap' that Michael Kroger spoke of last night, this week's polls represent a more modest improvement for the government of a point and a half compared with the fortnight that ended last weekend. 

That's because poll-mix previously has not had the opposition as far ahead after preferences as the pollsters have, but this week comes in the same as the pollsters' 53.5 to 46.5 average.  (More methodology here.)

October 19 Coalition reasons to be cheerful II

[Update: me in Crikey today.]

The two best performing pollsters at the 2004 election today show decent movement back to the government in the first week of this campaign.

Nielsen says 54 to 46 and Galaxy 53 to 47. Galaxy results here.

Under normal circumstances this would be a dreadful position for a government to begin a campaign in, but everything is relative these days, and it's all about the movement.

More polls please. (Yesterday's prediction still stands.)

That two party preferred of 51 to 49 - about which such a fuss was made - appears difficult to get from those primary votes. As you know, Galaxy estimates notional 2pp from its primary data, and at face value Labor would more likely get, after preferences, 52 or 53.

But it might be in the rounding. If we assume Libs' unrounded primary vote is 44.49, Labor's is 44.5 and Greens 6.5 ... it works, just.

Prime Minister Costello would very likely be today in the same dire straits said boss is (perhaps a little less so).

Everyone would agree it had been a terrible mistake: oh, if only Howard were still there, he'd have this Rudd upstart for breakfast, please come back conviction politician etc. 

But I do worry about Peter the opposition leader - he'll go the way of other first-off-the-rankers Beazley, Peacock, Sneddon ... a disappointing end to an illustrious career.

Election predictions

In December, Gerard Henderson will poke fun at pundits' spectacularly awry election predictions. Such egg-on-face considerations are one reason people go for  "it'll be close" scenarios - no matter the result they won't look too silly.

Gerard will be ridiculing either Neil Brown or Malcolm Mackerras.  Malcolm's prediction is much bolder, so he has more to lose.

My early prediction: about 90 seats to Labor

I reckon that people who aren't prepared to have a go - who cleverly hedge their bets all year - lose all bragging rights. I earned mine three years ago but, I believe, exercised them with great decorum and moderation.

Anyway, I'm in the Malcolm ballpark at this stage. He sees Labor winning 89 seats, but I'll up it to 90, which is a nice round 60% of total HoR spots - exactly what Bob Hawke won in 1983. (John Howard got 63.5% of them in 1996.)

I of course reserve the right to update my prediction at later date (but will not alter the numbers in this post). Labor could still conceivably lose the election - although it's highly unlikely.

October 17 Electoral roll             

We should, I suppose, be thankful for small mercies: that the government left the electoral rolls open for three days after announcing the election (by delaying the issue of writs).

But it totally negates their phoney argument for making the change in the first place: that the AEC couldn't handle tens/hundreds of thousands of last minute enrolments/changes to enrolment. 

Lots and lots of last minute, filled-in enrolment forms is exactly what the AEC has been trying to generate since Sunday night, through television ads etc.

They obviously feel they can handle the work.

Same Newspoll, different questions

When Newspoll employees were asking folks how they would vote last weekend - the results of that came out on Monday - they tacked on a few extra questions.

That extra data, published in today's Australian, shows ... something or other. (What they show is that the only thing that matters is voting intentions. But not without interest. Tables here.)

SBS Insight program on yoof

A month ago, an SBS Insight producer emailed: they would be devoting a program to young people at the election, and would I have time for a chat; did I think this would be an important demographic?

The bulk of my email response was:

'I'm happy to talk. I must warn you though, I'm a sceptic about beatups of this and that demographic "holding the key". (Also not very youthful.)'

And that was the end of that.

Caught the last 15 minutes of the show last night, and sure enough it seemed to rest on a couple of beat-ups: that young people voted right of centre in 2004; and that they "hold the key" this time. (But perhaps this isn't fair, as I didn't see the whole show.)

The first belief, though widely held, is absolute nonsense, and comes from one not very scientific, unweighted survey of 127 young people. Nielsen (and probably Newspoll), with decent sample sizes, say the opposite. (Much more here.)

A million potential beat-ups

The second is perhaps semantic. Members of any large-ish demographic you can think of - young people, middle-aged women, people over 55 - could sway the result if they all voted one way or the other. But they all split in some way - for example young people in Labor's favour and old people in the Coalition's favour.

But if you measured any group at the moment, they would generally show a large jump in Labor support from the 2004 election. 

This gives you a million possible beat-ups - eg, stock market yuppies/people with freckles/deadbeat dads hold the key!

(The program was still rather interesting.)

Election links

Am constructing a page of election links. Still in draft form, but a useable one-stop shop.

October 16 Me in Crikey today

On those Liberal tax-cuts.

My bets: Dobell

Dobell, north of Sydney (see Antony Green, Crikey, Pollbludger and adjusted 2pp graph) has a margin of 4.8 percent - exactly the uniform swing Labor requires to form government in its own right. It is seen by most observers as a "must win" seat - or close to it.

But I reckon Dobell is just the sort of electorate - mortgage belt etc - that will not swing very much in 2007. 

Therefore, I've thrown a few dollars on the Libs holding it - at quite good odds of two to one. (If the odds were only even, I wouldn't take the bet. I expect Labor to take the seat they win the election big-time, but I don't put it in the top 16.)

Malcolm Mackerras in The Australian

Malcolm is out early with a characteristically bold prediction: a 28 seat majority for Labor, with 54 to 46 two party preferred.

(It's actually a reasonable assessment at this stage, and assumes the government makes up a couple of points in the next six weeks. Similar seat-wise to Hawke's 1983 win, although a bit bigger vote-wise.)

On personal votes According to Michael McKenna, also in the Oz: "Queensland University of Technology professor of politics Clive Bean said the high local profile of MPs, like [Mal] Brough and [Gary] Hardgrave, could defy a uniform swing against the Coalition."

That reasoning works in the case of Brough, but not Hardgrave. 

Current margins reflect how people voted in 2004. Brough was then Minister for Revenue and Assistant Treasurer, but has since swapped the dreary suit for denim and is forever on the nightly news, dashing around the country, ciggie packet tucked into shirt-sleeve (not literally), barking orders. 

From carrying the Treasurer's luggage to fully fledged Action Man: that's got to be worth a few percent.

Hardgrave, on the other hand, was Minister for Citizenship and Multicultural Affairs at the 2004 election, and is now Minister for nothing. His high profile would already be built into his margin, and the only publicity he's had recently has been negative.

That, as well as their respective margins, is one reason Hardgrave will probably go down and Brough won't.

October 15 Newspoll in the Oz 

Says 56 to 44, tables here.

Galaxy in four Qld marginals

From this and that piece in the Courier-Mail, a Galaxy poll of 800 people shows about a 5 percent swing (they say 5.6, which presumably comes from subtracting a number to one decimal place from a rounded one, and also perhaps uses pre-redistribution 2004 results) to Labor across four seats - Bonner, Moreton, Longman and Herbert. 

Not sure why they didn't survey Blair, which is slightly more marginal than Herbert.

Anyway, a five point swing would put Labor on about 48 to 52 in that state. If we don't get too anal about uniform swings, we could expect such a result to give them, perhaps, five more seats state-wise. (That would still leave them with only 11 out of 29.)

October 14 Electoral roll closes this week (and next)

If you've never been on the roll but are eligible, or if you have dropped off, you have only until 8pm on Wednesday, the day the writs are issued - to rectify situation.

That is, the "close of rolls" date the PM stated as Monday 22 October is only for people who are on the roll and wish to change details, eg address.

Once again: rolls close, for people who aren't on it, on Wednesday evening..

See if you're correctly enrolled here. Enrolment forms here.

"Here we go again"

So began Prime Minister Paul Keating in late January 1996 as he fronted the media after popping in on Sir William Deane.

As you know, I expect a similar result in 24 November 2007: government tossed out, and not in a "close" way. But of course anything can happen in the coming weeks.

I'll be regularly filing for Crikey throughout the campaign, beginning tomorrow, and, space etc permitting, should appear some time in the Fin Review and Sydney Morning Herald. 

And of course I'll be posting here.

What's gonna happen?

I reckon tax policy will be a big headache for Labor. They can't promise anything big and sexy, because that will make them look intemperate and "unsafe", but if the government does it (and governments never look unsafe), Labor will have to fall into line. 

Also, the Industrial Relations focus is likely to switch from government to opposition, and it'll be Rudd and Gillard in the spotlight, doing the equivalent of calculating that GST on those cake candles.

The second last week

If there were bets going on it, I would put dosh on the government dropping something big and contentious - about "values"/people "not like us", naturally - in the second final week of the campaign. Will explain the reasoning for the timing (has to do with opinion polls) tomorrow.

Why so glum? Have you noticed that for the past month or so the man at right has, when fronting the media, looked like there's been a death in the family? Very flat and contained. Presumably it's about not wanting to appear to believe the election is in the bag; Lord Downer has been particularly effective with the "cocky" message.

I guess we'll have to wait until the evening of November 24 to see that smile again.

Howard's last hurrah

And finally: as you know (eg post below), I don't tolerate "Howard is great" narratives. However, it must be said that, being a "whatever it takes" politician, he is willing to try anything, think outside the strategic square.

So he won't go down quietly. Stuff will happen, and it should be "interesting" - in both the Chinese and normal uses of the word.

October 13 Maestro-itis

In the last few years a disease did the rounds of the Australian political class. You could call it Maestro-itis: the belief that John Howard is politically invincible, flays opponents with ease, has a special relationship with middle Australia and so on. No one can beat him, he is the Maestro.

Just this time last year it was a still common affliction.

These days, however, most people have not only ceased expressing such sentiments, they've forgotten they ever did. Human nature.

Return of the Maestro?

However, the the PM's re-entry in the reconciliation debate has probably brought a return of Maestro-itis, at least a low dose, in some quarters. A stir, that indefinable quiver at the back of the neck: could history be about to repeat itself?

There were similar rumblings after the NT intervention announcement in June, with one Labor staffer emailing, practically in tears: "Howard's so clever"; he was about to do them over again. 

But when the next Galaxy opinion poll showed most electors thought the whole thing a pre-election stunt, Labor support went up and the invasion dropped off the front pages.

Similarly, my bet is that if Liberal polling reveals that this pantomime has fizzed in Volvo Street, Liberalsville, it too will sink and be rarely heard of again.

"Just keep trying things", seems to be the government's approach, "we'll stumble across something eventually."

Rudd's Libs!

Speaking of "Rudd's Liberals" (and I immodestly lay claim to coining the phrase first) here's the first sighting I am aware of outside this site - in today's Weekend Australian. (They have a small 'l'; a large one is more appropriate.)

October 12 What do we want? Reconciliation!

(Between Mr Howard and his party's leafy base.)

APH Census data: broadband connections

Not unrelated, this table, from APH data, shows electorates ranked by percentage of households with broadband connections. I've also included 2004 swing, which shows a pretty impressive correlation between seats at top of table (many of them leafy Liberal seats) and swings to Labor three years ago; those at bottom tended to swing to Coalition. 

Table here.

[Update: Simon Jackman has, via email, chastised me for using clinical wording, - "impressive correlation" - to describe a superficial, untested observation. True, I should have written something like: "lots of seats at top of table swung to Labor, lots at bottom swung to Coalition".]

October 11 APH Library Census 2006 electorate data

The Parliament House Library has released Census data by electoral division. Against expectations, and perhaps also against what the title seems to imply, the electoral boundaries are actually those going into the next election.

Which is excellent news!

Here.

October 10 "Must win" seats

Whenever you hear that such and such a seat is a "must win", don't believe it. There are seats about which it can be said that it is difficult to see Labor taking government without winning that seat, but that's something different. 

And "difficult to see" is as strong as you can put it. You could, for example, take any particular seat in this this lot, and then reasonably say that the idea of the ALP forming government without winning that electorate would be nothing to be astounded about.

No seat is "must win".

Everyone's favourite bellwether

Everyone's favourite "must win" is Eden Monaro in NSW. You get the impression that the big parties almost think that if they can just get that seat, the other seats will follow. 

If that's the case, they're operating on superstition. 

Anyway, below are Eden-Monaro's swings compared with those of the state and country at the last three changes of government. (That's as far back as its bellwether status goes.)

October 9 Poll-mix

Me in Crikey yesterday.

Because Morgan's results are generally much more favourable to Labor than the rest, below graph excludes that pollster, just to see how things look without them. (Click for better quality picture.)

My two party preferred numbers, estimated from the pollsters' primary data, nearly always put it a point or two closer than they do.

Howard the campaigner: myths and reality

This wise piece by Peter Tucker in Online Opinion deals with the perception that the polls "must" narrow before election day.

One wee question I would add, and I'll return to this at some stage: are we talking about the polls narrowing, or are we comparing the pre-campaign polls with the actual election result?

There can be a difference. For example, Newspoll's 2004 published Coalition two party preferreds went from 48 percent before the campaign to 50, 50, 47.5, 48, 50.5 and then 50. That was a narrowing of only 2 points throughout the campaign, and of zero between the first week and the last. 

However, comparing the pre-campaign 48 to the actual result of 52.8 gives a (rounded) 5 points improvement - a much more impressive effort by Howard. 

Newspoll primary numbers in the last week in 2004 were pretty good, but they overstated Labor's two party preferred vote (and understated government's) pretty well throughout 2004 - until the election, anyway. Then they changed their preference allocation method. 

I think they still elevate Labor's 2pp a little, but nothing like in 2004. 

The Nielsen story

However, if you want a narrative that has the government making leaps and bounds in the 2004 campaign, go to Nielsen.  They went from 47 before the election was called to 54 in the final week. That's a seven percent improvement, which if repeated this year would get the government over the line. 

(Nielsen's final number overstated Coalition support by a percent, but they were second closest after Galaxy.)

October 8 Nielsen says 56 to 44

See Age, SMH and tables.

Nit-picking 1

According to Patricia Karvelas and Matthew Denholm in The Oz:

"Labor must hold Lyons and recapture Bass and the adjacent seat of Braddon - two marginal seats it lost in 2004 - if it wants to defeat the Coalition in the election".

Actually, they need to win 75 (preferably 76) seats in total, and it doesn't matter which ones they are. It's possible to win without holding any Tasmanian seats - see 1983 and 1984.

Nit-picking 2

Glenn Milne, on the same issue, reckons a "survey of 300 voters [is] a large sample in a single seat"

No, 300 is a small sample, with the same error margin (about 6%), whether surveyed in one seat or across the country.

But the results Glenn quotes, if fair dinkum (a big assumption with leaked party polling), do appear significant.

October 7 Seven (eleven) year itch

Here's one of the reasons Kevin 07 will probably become reality.

October 5 2007 The power of Tasmania

Twenty-four years ago a Labor opposition quite happily gave the finger to Tasmanian voters over the dams issue - and it worked a treat. Yes, that state swung against them, but it only has a few seats, and every other state swung to Bob Hawke to deliver that big majority.

So what's changed today?

Probably it largely reflects the caution that increasing drives politics today, especially under Team Rudd:. don't set anyone's pulse racing! 

It possibly also comes from a belief that Mark Latham's 2004 forest foray made a significant difference to that election result.

Commentators (and the party strategists who read their columns) mistake good TV news images - a couple of hundred forestry workers cheering John Howard - for things that actually matter. 

I doubt the so-called "forest disaster" cost Labor more than one seat - Braddon.

Sitting members

But perhaps most important was that Labor had no sitting MPs in Tasmania in 1983, while today they have three, and in 2004 they had five. 

That is, it was only candidates they were shafting in '83. That's different to telling a Caucus member: the good news is we're about to win government; the bad is that you won't be part of it.

(In 1983 the country swung to Labor by about 4 percent, but Tassie went by 4 percent the other way. If the state had also gone to Hawke by the same amount, Labor would, assuming uniform swing, have picked up three of the five seats.)

[Update: reader correctly points out that the '83 knocking over of a Lib government presents less problems than doing it to a Labor one. Don't think the state opposition was in favour of federal intervention either.]

Also this.

October 4 "Are you thinking what we're thinking?"
	+		=

(Recall this Spectator piece from a couple of years ago?)

Hockey-stick: a far-fetched scenario?

The big bear of a man might be in trouble in North Sydney, according to the SMH.

Leaked internal party polling should always be taken with a bag of salt or several. Still, a revenge of the leafy streets shouldn't be ruled out. Perhaps it depends on the effectiveness of the anti-union campaign.

October 3 That mooted Family First preference deal

The other day I mentioned the bizarre story doing the rounds that some Labor folks are advocating directing preferences to Family First in the Senate, because they'd prefer to deal with them than the Greens. 

As I noted, this does not compute, as the situation won't be either/or. That is, if the Coalition doesn't retain a Senate majority (or half the seats) after the election, a Labor government would have to deal with the Greens, and the only question would be whether they must deal with Family First as well.

Story within the story

But while the reasoning is flaky, there's probably more to the story.

Labor thinking might goes like this. The preference piece de resistance would be a deal with Family First in the lower house. Everything else is secondary. They believe, with some (but not absolute) justification, that FF (directed) preferences are worth more to them in the HoR than Green ones, even though Greens get three or four times the support.

In return, of course, Family First would expect Senate preferences. 

These pessimistic ALP number-crunchers have pretty well given up on the possibility of having, with the Greens, control of the Senate; at the very least, they reckon, they'll need Steve Fielding. 

I reckon they're being overly pessimistic (if you're in a numbery mood the reasons are here) - but if you accept that assumption, then it doesn't matter how all those minor party spots are divvied up between Family First and the Greens.

The package

The argument that is being put about - about flexibility, keeping their upper house alliance options open - is absolute nonsense, for the reasons already given, but it's just an attractive package to present the proposition in - not least to Labor colleagues who view Family First as evil. 

"We can deal with Family First in the Senate", is the message, but it's really all about trying to get a lower house preference deal.

(Go to the convoluted more numbery bit.)