July 16 pm George and house prices

George Megalogenis today did his usual fine analysis in the Oz, this time on house price movement by electorate. Simon Jackman has looked at it.

Mr Morgan's data

As you know, Morgan has been polling since 1941, under Roy and then from the early 1970s his son Gary. Last week Mr Morgan dipped his toe in a certain skirmish by claiming in a press release that "[a]ll pollsters know voting intention is the real guide to how electors will vote". 

He's also scanned a few old newspaper cuttings. Here, in September 1968, PM Gorton is approved by 63% and Whitlam by 53%. In March '69 it's very similar: 62 to 50. Gorton scraped home in the election in October that year.

And in November 1971, a year out from the government-changing election, Whitlam is only on 36 and McMahon on 37.

Hopefully we'll get old preferred PM numbers later in the week. It'll be interesting to discover when they were first gathered.

July 16 More excellent news for Howard: 58 to 42

ACNielsen in SMH. However, look at this page in The Age: if you divide the 62% who want troops home by the 22% who say housing affordability has declined because etc, you get almost exactly three. This, of course, was the 1996 height difference in inches between John Howard and Paul Keating.

Those who really understand opinion polls would realise that this indicates Howard is likely to make a come-back, based on the fact that Kevin Rudd and Howard are of a similar height and the PM's birthday is approaching. 

Only mad ultra left-wing Howard-haters would disagree with this analysis, surely.

July 15 A new shadow cabinet?

Four years ago, almost to the day, when Simon Crean was ALP leader, I pondered a possible new opposition line-up. Rudd and Latham entries particularly prescient. See top of this page.

July 13 Consolidated Newspoll: the states

• Am formulating a response to yesterday's Oz editorial - not to the hyperbole, but the polling/election data assertions.

Today, meanwhile, the paper gives us consolidated Newspoll numbers over April - June, extracted by state. The PDF is here, analysis here and two party preferred stuff below.

.

Because Newspoll is generally registering about two points lower Green support than we saw at the last election, and their 2pps don't take this into account, we should knock about a point off each Labor number (and give it to the Coalition).

(The order of state support is as anticipated last December, but not the swing order.) 

Comparison 2004 v 2007

How good a guide to the approaching election result was Newspoll's quarterly analysis this time three years ago? Below is the table.

Newspoll state by state 2004 quarterly two party preferreds

The first thing that jumps out out is how narrow Latham's leads were compared with Rudd's. Beazley was generating better numbers than this when he got dumped last year. 

In addition, Newspoll's now rightly abandoned "just ask for second preferences" strategy overstated Labor's lead even more seriously than their current one does now.

Latham's electoral buoyancy came at the overlap of these two periods, around March-April.

Order-wise, the prediction was mixed. At the 2004 election, Victoria was indeed the most supportive, but the rest don't really work. 

The most overstated Labor vote in the 2004 table was Queensland's; whether Rudd really does boost his home state this time around will be one of the more interesting items on election night.

July 12 They do take prisoners in the Newspoll Wars

In the end the Oz has given a very long editorial spray at biased, clueless, out of touch online parasites in general, with a mention of me in particular near the end. 

Not personal, and could have been much worse. That Mr Mitchell is a lovely fella.

And so is Dennis!

The paper's blogger Tim Dunlop wraps it all up. [Update: he did; now it's gone.] [Up-update: Tim absented himself from his blog for 24 hours but is now back. As he briefly notes at bottom of this, the post was indeed pulled. You can read what he wrote at Guido's place and/or Larvatus Prodeo.]

July 11 (pm) Casualties in the Newspoll Wars

I think Dennis Shanahan wrote this this morning (as opposed to yesterday). The "PhD" mentions refer, I believe, to me.

A courtesy call from Editor-in-Chief Chris Mitchell this morning informed me that the paper is going to "go" Charles Richardson (from Crikey) and me tomorrow. 

Chris said by all means criticise the paper, but my "personal" attacks on Dennis had gone too far, and the paper will now go me "personally".

No, I'm not making this up.

If they only get as personal as I get with Dennis, then it should be tame, as I don't believe I've ever criticised anything other than his writing.

And to think I described Dennis, in a chapter in a book being launched this month, as (with no sarcasm) "a fine journalist".

All very strange. And - I'd be lying if I didn't admit - a little stomach-churning.

July 11 (am) Oh Martin, what are you doing?

The Oz used to judiciously enlist Sol Lebovic to the Newspoll Wars. Today his replacement as Newspoll boss, Martin O'Shannessy, defends Dennis Shanahan's enthusiasm.

The graph Martin refers to, which I've stretched for easier reading, is here.

It is silly to linger over the entrails of one single opinion poll, but let's deal with O'Shannessy on his terms. 

Martin sets the question as:

"whether the data supports the view that a turnaround in Mr Howard's better PM rating presages an improvement in the Coalition's electoral stocks. The short answer to this question has been yes in the past three elections."

But the answer - long or short - is "no", because there's a difference between "presage" and "being accompanied by". Martin notes that in 1998, 2001 and 2004, both Howard's "better PM" and voting intentions rose as elections approached. But as the graph shows, the two measures rose simultaneously. Martin seems to acknowledge this, but then jumps to:

"the basic pattern of improved leader's ratings being followed by an improved primary vote appears supportable."

As we know, Howard's better PM numbers last weekend were not accompanied by a vote improvement; in fact the vote gap increased.

In any event, it's not just Howard's' "better PM" numbers that have improved as elections have approached. Incumbents' generally do; see for example Keating v Howard, below.

Newspoll better PM Jan 1995 - March 1996 (Keating v Howard)

What the Coalition desperately needs is better voting intentions. Votes win elections, not all that other fairyfloss.

On the other hand, no-one really expects Rudd to win by 56 to 44, and so unless we're headed for the biggest wipe-out since 1966, voting intentions must narrow before election day.

The last Newspoll also had the paper celebrating a comeback, as did the previous Galaxy. The Howard government is in deep deep electoral doo-doo, and the Australian does it no favours by continually clutching at straws. (It probably helps Howard against Costello, however, which might be against the Coalition's electoral interests.)

[Final line, unnecessarily personal, deleted.]

July 10 2007 Great news for Howard: Newspoll shows Labor landslide

56 to 44, about what polls have been saying all year.

Here are the tables.

No surprise that most people (61%) support the government's NT intervention. The Oz's home page headlines imply wonderful results for Howard. Meanwhile, Newspoll's Martin O'Shannessy believes it inappropriate for a pollster (Galaxy) to probe attitudes to the PM's motives. A very Australian way of looking at things.

• And here's me in Crikey.

July 9 Newspoll tomorrow

Musings this time last week on the next day's Newspoll proved premature, because there wasn't one. They apply today instead.

Electoral roll update

Simon Jackman's and my paper on the state of the electoral roll, published at the Democratic Audit a few weeks ago, has been getting a bit of publicity, including mentions in this and that article.

Here's a wee update. Our paper stopped at 30 April, but the AEC website now has 31 May and 29 June (the page says 'July 29', a typo) enrolment data. The graph below shows total national enrolment over the last year and a half, including those last two numbers.

In the paper we noted the recent impressive monthly roll increase, with the caveat that the AEC has been holding back about 180,000 'objections'. (For a definition, see pg 7 of paper.) The Commissioner told a parliamentary committee in May that these objections would go out in early June, which we suggested meant they wouldn't impact on the total roll numbers until the new financial year.

However, the roll increase in June was small, only about 7k, compared with 47k in May. (See the line flattening at the end in graph above.) Was this due to some of those objections taking effect? Can't say at this stage. 

And perhaps we never will know, because the annual reports in the past have only given total annual objections (and re-enrolment), not monthly.

Crunch-time will come after the election, when we see what the roll was like at the time it truly mattered.

July 8 The stories we'll tell in 2008

Around this time six  years ago (although much closer to election-day), when this site was even rougher than it is now, I anticipated various commentators' interpretations of the upcoming 2001 result. Some entries worked better than others.

What stories might be told later this year?

A Coalition win

If Rudd loses, the Latham experience provides an ok approximation. Media fans will initially explain that he put up a good fight, it's always difficult in such economic conditions, it's a two term strategy, it's all Beazley's fault, back-bench deadwood, membership base too low.

But the hard logic of the news process will eventually demand they turn on him: he had no policies, didn't move to the centre, this and that policy (ok, he had a few) were real stinkers. He's a bit of a weirdo.

And he lacked the guts to reform the party!! How many times must we tell these guys?

The flipside story, of course, will be that Howard is still the Master and we were wrong to doubt him.

A Labor win

But that scenario is unlikely, because Labor will probably win. Revised history has it that Latham imploded before the election, but in reality he put in a good campaign; it was generally judged superior to the government's at the time.

By contrast, it's difficult to see Rudd losing via anything other than a dramatic pre-election implosion, or a really stupid policy release (one that is discernibly so at the time - eg a huge tax cut comes to mind.)

How will history treat a defeated Howard? Again the past is instructive: look at Fraser and Keating. Lots of people are still fond of them, but the official writing is unkind. Compare that with Hawke, who gets a fine writeup. 

Once the electorate is seen to reject you, you're unrehabilitatable.

July 6 On "charisma"

Apparently, Gordon Brown is not as "charismatic" as Tony Blair. Is there such a thing as "charisma"?

Sometimes it just seems men in politics with tall hair are deemed "charismatic". Bob Hawke, Andrew Peacock, Boris Yeltsin, Junichiro Koizumi and Mark Latham all attracted the label.  (On the other hand, no-one accused John Kerry of this, so maybe it's height of hair relative to length of face.)

But charisma is very largely context dependent, a point not unrelated this one: many commentators seem unable to separate the authority, take chargedness and - yes - charisma that come with the most powerful job in the state/country, from the current occupant's inherent qualities.

Yes, the camera likes Peter Costello, and Peter Beattie has something. But when John Howard and Bob Carr are described as "charismatic" (as I've heard at least once), we might as well just throw the word away.

Perhaps one test is if they seem charismatic even when in opposition. Jeff Kennett, another tall-hair, was considered a clown and a lair as opposition leader, but charismatic in power.

Maybe here's an ok description: occupying a position that is held in high esteem by the community, but doing it with a touch of self-assured unorthodoxy and perhaps informality - for example by having big hair.

In any event, while "it" can help propel individuals to party leadership (hullo Andrew, hullo Mark), it has little relationship to actual electoral success.

End of today's thought bubble.

July 5 (pm) Individual seat betting

Via Andrew Leigh, a betting company has opened books on all federal seats. Surely a first. Assuming they stay open until election day, it's great for people like Andrew who follow the odds, and of course those who fancy a punt.

Apparently when they first put them up (this morning?) they had Labor favoured in Greenway (notional Liberal margin 11.4%), which would have been a bargain. This arvo the odds there were still pretty good, and even better was Macquarie, notionally Labor-held  by half a percent. I  threw a few dollars at each, at prices below.

July 5 My first gig

In early 1992, my first (meagrely) paid published piece appeared, in "Monday Bloody Monday" in the Sydney Morning Herald. This was a section readers sent yarns about their jobs to, and my pocket money-earner was as an "extra" for the Australian Opera Company .

Here is the hardcopy scan and here the is text. Yes it's over-written, but I still like it.

(Originally this post segued into political stuff, but it didn't really work so it now ends here.) 

July 3 Newspoll - there ain't one

Curious. Will attempt to find out why.

Update: Newspoll says there'll be one out next week.

I reckon there are three possible explanations: [Update: sadly for the conspiratorially minded, Explanation 1 is correct.]

Explanation 1 (80% probability): Nielsen sync

My guess is that it's to get out of sync with Nielsen in Fairfax, which runs every 4 weeks. A recent long weekend, on which Newspoll didn't poll, put them in sync. When they're published in the same week, The Australian has to second-guess whether Fairfax will publish on the Monday to get in first. To be safe they (the Oz) publish on the Monday too (as happened two weeks ago).

Which is a lot of hassle. Newspoll staff have to crunch the numbers on Sunday night, which perhaps costs money in overtime and presents a greater risk of error; everybody at all stages of the process, eg Dennis/Steve, is rushed. Much better to have the whole of Monday to do it.

That's my theory, anyway.

Explanation 2 (10% probability): they didn't like the results

On the other hand, there's a miniscule chance Matt Price's tongue was not in his cheek when he typed, on the Oz site this morning: "The results didn't conform to News Ltd policy ... so we're doing another for publication next week." [Link from pollbludger.].

Explanation 3: (10% probability): any other reason

Computer crash, a dog ate the results etc.

July 2 (pm) Polling: what's in a question?

Answer: lots. Galaxy found 58% of respondents reckon the NT intervention a political stunt. But what if they had instead asked whether people thought it was a good idea? My guess is lots - probably an even bigger majority - would say it is.

Suggested this mix of cynicism of motives and approval of plan here (fourth par) and here (third par).

I doubt The Australian would disrespect the PM so much as to ask such a question as Galaxy's. More likely we'll get "do you approve of the PM's plan ..?" tomorrow. A nice counterpart to today's poll.

Galaxy poll says 55 to 45

Malcolm Farr's intro in the Telegraph says:

"Most voters believe John Howard's push to improve Aboriginal communities is nothing more than a vote-grabbing stunt, the latest Galaxy poll figures reveal."

Yesterday on 'Insiders', Malcolm and Glenn Milne were getting terribly cross with anyone, anywhere, who had the poor taste to suggest such a thing. It now appears the list of heretics is very long indeed.

This of course lengthens the odds for an August poll, especially if Newspoll tomorrow has a similarly big 2pp gap.

Galaxy's questions

Here's the data; also a question about union influence. Note their questions have changed since the mauling they got last time. Two alternatives on the NT question, three on the union one.

Three is what pollsters tend to favour - a kind of neutral one in the middle. Should they have included "both because of the upcoming federal election and because he cares" in the NT question? Such a broad range would probably have attracted a large number of responses - and told us very little. 

June 30 August election bet?

Bryan Palmer wonders whether I'd fancy a wager on that "early August" election. 

Bryan notes that, according to this timetable, a minimum of 33 days must pass between the dissolution of parliament and the election, and so an August 4 poll would have to be called in the next couple of days. Must admit I hadn't known this, and was thinking it could be "about four weeks".

So, my response: I never said it was more likely than not, just more likely than people seem to think. In fact the last line here logically implies a less than 50% chance of an election in that month.

Still, putting my money where my mouth is: I put the chances of an August election at about one in three, and one in six for an early August one (4th or 11th). Modest bets accepted along those lines - but after Monday the odds of course change. (Given the timetable, the chances of August 4 are now miniscule.)

Electoral bounce

An important variable in the election timing equation is how much bounce Howard gets out of the exercise. Voter cynicism would I imagine be the order of the day, but in the last 24 hours land rights has been thrown into the mix (surprise!) and other juicy issues will presumably follow, at least subliminally. Does Abstudy still exist?

(It's curious how the people employed to watch and analyse politics go on and on about how brilliant our PM is at politics, but have difficulty, at any given time, actually perceiving that he is playing politics.)

June 29 More Northern Exposure

Me in the Canberra Times (again) on the logical political direction of Mr Howard's intervention. [My two d'ohs: "airwaves" not "airways"; and Brough spoke on '7:30 Report', not 'Lateline'. Plus a sub pluralised my "mob".]

Malcolm Farr would not approve.

But Malcolm illustrates one of the very things he dismisses: the Iraq war comparison. As in wartime, we're being told that this is not the time to criticise the government. 

This sort of bossiness will surely subside after a few weeks and be replaced with questions - including probably from Malcolm.

Another reason not to dismiss the early election scenario. 

June 28 Me in Crikey

On the chances of an August election. Here.

June 27 ABS 2006 Census 

As of 9:30am, census data is available at the ABS site. 

The Parliament House Library told me that the ABS would have electorate by electorate data on the pre-redistribution boundaries, and the Library will compile data for the new boundaries around October. (Surprised me that the ABS would have this.)

At a quick glance I can't see any reference to it. [Update: here. Now to find someone who has purchased the data.]

June 25 Me in Canberra Times

Written before the NT intervention, a general musing on polls etc. 

June 24 State of the electoral roll at the next election

At the moment I seem to be the only person contemplating an August 4 election, although Bryan Palmer has gently suggested people get their electoral roll details in order.

Probably Liberal pollsters are, as we speak, measuring how the NT intervention is going down.

My guess is it won't give Howard much bounce, but if it does, and an election is held in a little over a month, what would be the state of the electoral roll? The best calculation is: terrible. The AEC needs as long as it can get to get the thing into shape, a situation exacerbated by the new law closing the rolls when the writs are issued.

As of a month ago (the latest numbers), the electoral roll had grown by about 1.6 percent since the 2004 election, while population growth had been about double that.

Obviously the AEC hopes to catch up substantially between now and the election, but an August poll would probably see it well short of comprehensive. And don't believe what Gary Nairn or anyone else tells you, such a situation would certainly benefit the Coalition at the ballot box. 

Not a reason in itself for the government to go early. But maybe icing on the cake. 

Mark Textor is a deadset legend

In the SMH, Alan Ramsey recalls a push polling case in 1995 involving Liberal pollster Mark Textor and Andrew Robb. That referred to real push polling, not the dodgy polling Galaxy was accused of recently. Push polling is illegal, I think. [Not really, clarifies Graeme Orr.]

Anyway, Mark Textor, the baldy Liberal pollster, is a deadset legend. This is what we're repeatedly told, anyway. 

His work is of course mostly private, but my recollection is that he was the Bulletin's pollster for the 1998 federal election. And if that recollection is correct, his numbers were generally gave the Coalition comfy two party preferred leads, while the other pollsters had them behind. At the election the government got under 49%

Maybe qualitative work is his forte.

In any event, like his partner Lynton Crosby, there is surely a lot of right-place-right-time involved. We also don't hear much about their involvement in the 20-odd state and federal losses over the last decade or so.

June 22 Tampa and an August election?

Over the last few months, quite a few Labor conspiracy theorists have pondered, via email, the possibility of Howard "doing a Tampa" to get over the 2007 election line.

Could yesterday's announcement of a Commonwealth takeover of NT Aboriginal land be it? Its Tampa-esque qualities have been noted by others. It's big and unorthodox, elicits emotion and turmoil and re-establishes a Prime Ministerial connection with the electorate. 

It provokes human rights lawyery types, could cause the ALP internal grief, is apparently legally dubious and appeals to an instinctive yearning for an older, more orderly Australia. 

It returns the unblinking Man of Steel to our screens, a rock of common sense and certainty amongst the hand-wringing and equivocation. 

And it involves, of course, people of colour.

(It may or may not be good policy; I don't know.)

Anyway, if we let our conspiratorial imaginations run wild, we can imagine writs being issued in the next fortnight for an election on 4 August (the earliest practical date).

But that's surely taking things too far. (Labor would still win, anyway.)

June 21 The incredible vanishing electoral roll!

Simon Jackman and I have written a paper for the Democratic Audit on the current state of the electoral roll. It's a shortish paper, suitable for that forum; we intend to revisit the topic to pick further at a few strands  - probably both before and after the federal election.

(Annoyingly, the amount of data available on the AEC website is also shrinking; in particular the historical monthly enrolment data has vanished.)

Simon has a good summary of the paper at his blog. [Update: with lots of interesting reader comments.]

Read paper here.

June 20 Me in Crikey

On the cost of political opinion polls. Here

More Oz editorial oddness

Yesterday's Oz editorial insisted on doing what Dennis regularly does: treat individual poll movements literally. The 56 to 44 2pp might appear favourable to Labor, they said, but "wiser heads" look instead to primary support. In particular,

"the slump in Labor's primary-vote popularity of six percentage points, from 52 per cent to 46 per cent, could be a wake-up call that the momentum in the Rudd campaign is starting to flag."

But last fortnight they dismissed the then most recent Newspoll, the 52 percent one, as one-off and "inflated". The Newspoll before showed 47 percent. In which case Labor's primary vote has barely moved.

June 18 Mr Shanahan's half full glass

Nielsen in Fairfax papers and Newspoll in the Oz are saying 57 to 43 and 56 to 44 respectively. But the headlines are very different.

Never before in the history of opinion polls have such diabolical numbers for a political party been so spruced up as those in today's Australian. According to Mr Shanahan, "John Howard and the Coalition have got a polling breather."

Two weeks ago an Oz editorial celebrated the return of the Master in the Galaxy poll (53 to 47) by dismissing its own recent Newspoll result (60 40) as "artificially boosted". But now 56 to 44 means "Howard closes gap".

Be thankful for Nielsen; if we only had the Newspoll today, news-writers around the country would be regurgitating Dennis's lead-in.

June 17 Betting markets part LXIII

Jason Koutsoukis in The Sunday Age on superiority of betting markets over opinion polls writes: 

"For the first time in more than a decade, bookmakers have for the past few months steadily rated Labor as favourite to win."

Not true Jason. During the first half of 2001 Labor was ahead in the betting markets - often further than at any time this year.

[Update: Jason's probabilities for both parties add up to more than 100%, which is obviously silly. 

He needs a formula that allows for bookmaker's commission, such as: Probability of Coalition win = Labor price divided by (Labor price plus Coalition price). Swap them around for Labor probability. (Simon Jackman gave me this once; it's presumably how Bryan Palmer does it as well.)]

Big week for polls

Jason verifies we'll get a Nielsen in Fairfax papers tomorrow - a 1400 strong one. And of course Tuesday will see a Newspoll in The Australian.

It's quite exciting really, for two reasons: (a) what the polls inherently say, and (b) flow-ons to the political narrative. You will recall how the Galaxy published a fortnight ago stopped dead most talk of Costello for PM. 

Since then a Qld Galaxy and some Westpolling in WA have been relatively good for the government. Morgan's gap has narrowed in polls such as this (although I read somewhere recently - can't recall where - that Gary Morgan reckons his telephone polls are more favourable to the Coalition than his face to face ones).

Over the last fortnight the story shifted from "Howard appears in big trouble" to "never write Howard off". Commentators have been hedging their bets, as have punters quite literally, with betting markets moving paying $1.62 for a Labor win two weeks ago to $1.85 today. (Coalition now pays $1.90, which is close to dead even.)

The political class wrongly sees Newspoll as most accurate, which makes this Tuesday's headline 2pp the most important. Perhaps 55 to 45 is the cut-off: closer than that and Howard will still be considered back in the game, the political colossus rebuilding support brick by brick etc. 

Much wider than that and we're back to "what're the Treasurer's plans?"

Belatedly, from the 7:30 Report ...

This Clarke and Dawe on Rudd is hilarious.