October 2 Me in Crikey yesterday

On "the sixteen seats Labor needs to win!" Here.

October 1 Newspoll says  56 to 44

In The Australian, tables here.

Out a day early. Rumour has it they threw in some extra questions which we'll read about tomorrow.

September 29 Innumeracy in the ALP?

Christian Kerr in Crikey  (sub/trial required) and Patricia Karvelas the Oz are reporting that Labor is considering preferencing Family First in the Senate because, apparently, a Rudd government would find them easier to deal with than the Greens.

Has the Grand old Party of blokey number-crunchers forgotten how to count? Or is the story a try-on? Charles Richardson explained the flaws this week, also in Crikey (sub/trial required), but it persists.

There is no way on God's earth that Labor plus Family First can get a majority in the Senate at this year's election. So in simple terms, a Rudd government would be faced with one of three Senate scenarios:

1. A Coalition majority (or half the seats)
2. Labor plus the Greens have a majority. 
3. Labor plus the Greens plus Family First (plus, very remote chance, Pauline Hanson and/or someone else) have a majority.

So the question for Labor is whether they would prefer to deal with the Greens, or deal with the Greens plus Family First. The answer is obvious: trying to please one of them would be hard enough, but both at the same time?

The lower house, which we looked at last year, is a different kettle of fish, but even that analysis rested rather heavily on some assumptions.

[Update: Malcolm Farr seems to buy into the same muddle, writing:

"The worst case for victorious Labor would be that the balance of power was concentrated in one party, and that party wanted to use that power to rewrite Labor policy."

But in Malcolm's favoured scenario, of both the Greens and Family First holding the balance of power, the ALP won't be able to tell the Greens that if they don't play ball they'll deal with Family First instead. 

They'll have to get both of them onside. That means even more "rewriting".]

[Up-update. According to Christian it's about "flexibility" and "an ability [for a Rudd government] to move left and right."

"Political parties always like to keep their options open." 

But putting two hurdles in your way instead of one does not create options. Once more: even in the rosiest of scenarios, Labor will not be able to pass legislation without Green support.]

September 28 Me in the Fin Review on Wednesday

Partly on the pointlessness of conjuring seat-vote scenarios without giving any reasons as to why they might happen.

Page dummied up here.

(Must admit to playing the game myself a little, eg at bottom of this three year old piece.)

(Obscure note: re the 50.7 number in both articles. It was surprising that after allowing for redistributions since 2004, that number didn't change.)

Historical Labor votes needed to win

While we're on the topic, this graph shows the two party preferred votes Labor would have needed to win at elections from 1949-2004, assuming uniform hypothetical swing. Taken from this table.

The elections at which Labor was incumbent government are red. (1975 not strictly true, as Fraser was caretaker PM.) The large red dip down and then up is the Hawke-Keating period.

Notice that at the three elections with Hawke as PM the vote needed was below 50, while both the Keating numbers are above. 

Labor's vote-seat success in the 1980s largely came from alienating the "heartland" - mainly around western Sydney, but also up to Newcastle and down to Wollongong - but not enough to suffer seat losses. But Keating's 1993 GST election saw these very seats swing back dramatically - again with little seat gain.

Not sure how useful this graph is for pondering the future. But I suppose my AFR piece implies the next dot might be a little under 51.

September 27 South Australia 

South Australia is the second smallest state, with only 11 seats, but it gets lots of attention because it is seen as pivotal to Labor's election chances. Labor holds only three seats there and, so the theory goes, the only way is up.

Below is a graph of national and SA two party preferred votes at federal elections 1984 - 2004. Colour code: Labor (fed) Coalition (fed) Labor (SA) Coalition (SA) 

From 1984 to 1990, Coalition support in SA was a little better than their national vote, but the difference increased big time in 1993 - when there was a national swing to Labor (fuelled by NSW, Vic and Tas) and SA went the other way - and the gap has remained large since then. It did narrow a little in 2004 when the Coalition got 52.7 percent nationally, with an SA component of 54.4.

Makin and Grey

Mr Bludger posts on Makin today, and the comment thread of a recent post Professor Jackman post visits the likelihood of Grey changing hands. One is seen as a likely Labor win, the other a very long shot (or no shot at all). 

Below are graphs of 2pps in those two seats 1984 - 2004, adjusted for redistributions.

Makin 1984 - 2004 (horizontal lines at 40, 50 and 60%)

Grey 1984 - 2004 (horizontal lines at 40, 50 and 60%)

Grey is to SA what Kalgoorlie is to WA, accounting geographically for the vast majority of the state. On paper, with a margin of 14%, it looks a jolly unlikely prospect for Labor, but recall John Howard's 5% swing in 1996 produced individual seat components as high as 13%, and in 11 electorates the swing was greater than 10%. 

What might a national swing to Rudd of, say, 7 percent (to take Labor to 54) contain? 

I reckon the swing to Labor is going to be smallest in the outer suburban mortgage belt. That's another way of saying it will be greater than average outside the cities (as well as towards the inner cities). 

It's also reasonable to vaguely expect that, coming off a low base, South Australia will swing by a larger amount than the national number. (Well, I'd put the chances at slightly better than even. The recent quarterly Newspoll consolidation of 699 South Australians has the swing bang on the national one, 9.)

All rather fuzzy, I know. But with the sitting member retiring, ... you never know.

September 26 Oz editorial: here we go again?

The Australian delivers another odd editorial on opinion polls. In similar vein to the last instalment's not quite substantiated historical assertions, we get this: "History shows the polls will narrow before election day". (Let's insert a "probably". The polls didn't narrow in 1996.)

And this:

"Labor is certainly keen for the perception it will have a landslide win to take hold to help build a community acceptance for change."

We know The Australian believes perceptions about likely outcomes have a self-fulfilling influence, but the ALP? Let us hope not, for their sake.

Online polling

I was in Crikey yesterday, on an area I've only recently begun looking at, online polling. 

Simon Jackman, on the other hand, is an expert.

September 25 Antony's election site

Over at the ABC, Mr Green's election '07 site went up last Friday. The usual brilliant stuff, plus a spiffy calculator. Please note (and Antony warns the same) not to take the seat by seat swing results (as opposed to overall seat numbers produced by such and such a swing, which while not always accurate, often are) literally. On election night seats will fly all over the place - as they always do. 

In 2004, for example, a 1.8 percent national swing to the Coalition included everything from 9.1 percent to the government to 4.5 percent to Labor. But as Megalogenis, G noted the other day, the 1.8 percent swing produced almost exactly the net seat change the Malcolm pendulum would have predicted.

Other links

This looks like an interesting project.

And this.

And if the betting markets turn you on, you should regularly visit Simon.

September 24 Poll-mix: me in Crikey today

Here.

Galaxy says 56 to 44

Here. Not for first time, none of the tabloids has the tables online. Will attempt to get them this morning. 

[Update: here they are.]

September 23 2007 US Prez election

Went to see a talk by a bunch of visiting young US state politicians last Thursday. They had just been to the final(?) House of Reps question time before the election, and still couldn't quite believe their eyes and ears.

The Republicans said the one thing they were hoping for in 2008 is Hillary Clinton winning the Democratic primary. Most of the Democrats agreed: "it would kill us". 

The Oklahoma(?) Democrat explained that in 2004 lots of Republicans, who otherwise would have stayed at home, lined up for hours just to get a chance to vote against John Kerry. While they're in the compartment, of course, they vote for Congress, state and local legislatures and every other position under the sun. That's what the Dems are afraid of next year.

"She's a cold woman", said one of the Republicans, which is probably a good summation of mid-western, middle America's attitude towards her. (They were mainly from conservative states, I think.)

Still, this person reckons that the Republicans are going to end up with someone so unelectably right-wing that Hillary will win by default. All she has to do is win all of Kerry's states plus Ohio, he says.

Which seems optimistic on Hillary's behalf, but there you are.

Sol's electoral history

Sol Lebovic in The Oz puts Labor's current poll leads in the context of past election results. He's correct as far as he goes - although comparing raw seats numbers isn't really cricket given the growth in the size of the House - but he doesn't go back past 1949.

That's because the AEC doesn't have two party preferred data before then. (The page, like many, has disappeared with the AEC's revamped site.) AEC has actual data going back to 1983 and estimates from 1949 to 1980. 

The only pre-1949 estimates I'm aware of are Adam Carr's (whose relevant pages are currently unavailable as well), which I used for these tables.

As you can see, there are indeed precedents for very large swings, such as a little over 7 percent to Labor in 1929 (which, as we've been hearing recently, saw PM Bruce lose his seat), fifteen percent to the UAP two years later, and about 9 percent to Labor in 1943. (The last one is based on Adam's 59 percent two party preferred vote, which admittedly looks a bit high.)

It would be good if the AEC went back to 1919, the first general election under preferential voting. Hopefully one day they will. 

September 21 Newspoll quarterly consolidation

Assuming we should interpret these numbers literally ....

Happy news for Mr Howard in the Queensland swing to Labor being only nine percent, and WA going by just four. Another bright spot: outside the capital cities the shift to Labor has been contained to 8 points.

Other than that, things look gloomy for the government.

Tables here and here.

September 20 The "Why?" battle begins

Imre Salusinszky is out early in The Australian with "the reasons" for the election result, and he is correct to say that if/when people toss John Howard out it won't be because they "hate" him. 

But Imre practices denial in totally ignoring the theme he and his buddies have been flogging all decade - that Australians can't get enough of John Howard, his prime ministership represents a victory in the culture/history wars, our everyman PM embodies the values of real Australians etc.

"Rudd only won by pretending to be Howard" will be cold comfort; Howard as opposition leader in 1995-6 similarly minimised the differentiation, particularly on economic policy. 

Once the result is in, a story is written.

(More on alleged similarities between Rudd and Howard.)

September 19 The view from Brazil

Nicked this link from Pollbludger. A commenter translates.

Nexus poll in Sydney and Melbourne

See this, with lots of data at bottom link. Defining what constitutes Melbourne and Sydney is tricky, but the average votes across AEC-designated inner and outer metro seats in the two cities in 2004 were: 43.7 to Libs and 41.4 to Labor, Greens on 8.5, Labor ahead two party preferred 51.5 to 48.5. 

Sol Lebovic in the Australian

Sol explains why we shouldn't write John Howard off. First, he notes that Howard's approval ratings remain respectable, while Keating's in 1996 were in the 30s. 

I think approval ratings are are useless predictor of election-results, but to enjoin Sol's game we can note that Rudd's approval today is also higher than Howard's were back then.

Lebovic also reckons:

"Howard as Opposition leader in 1995 was in a stronger position than Rudd is on the key measure of who voters see as the better economic manager."

True, although as this table shows, Keating was ahead in the final poll. 

Yes the economy is Labor's weak spot (and generally has been when they're in opposition), but to place such confidence in a set of numbers generated by a pollster's question, and on the fact that they don't coincide with the 1996 experience, doesn't convince.

My guess is that Sol will be rewriting his set of rules over Xmas.

Belated response to that Oz editorial

Relatedly, here's my late response to that Oz editorial of a couple of months ago, which thundered many things, including that:

no Opposition since World War II has won government without two key indicators 12 months out from the election. These are that the Opposition Leader has a lead over the incumbent of at least five points on the question of who would make a better Prime Minister and the party has a nine point lead on a two party preferred basis. Applying this historical test Mr Rudd may not have had enough time to cement his claim to the top job, though he leads by a huge margin now.

The facts as far as I can tell - although I'm willing to be corrected - are these. Until the late '60s/early '70s, the concept of national two party preferred support did not exist in the popular mind and was not measured by opinion polls. (That is, pre-Malcolm Mackerras.) Similarly, "preferred PM" wasn't measured by anyone until the 1970s. That knocks off the 1949 change of government.

We're left with four - 1972, 1975, 1983 and 1996. Malcolm Fraser (1975) and Bob Hawke (1983) weren't opposition leaders 12 months out, so they're gone too.

Whitlam? This late 1971 report of Morgan (on second page) has, by any reasonable distribution of DLP preferences, the Coalition ahead in the 2pp. Don't know if preferred PM was around in the Whitlam v McMahon time. Maybe not.

So our six decades has been whittled down to one election - 1996 - and, again, nit-picking about some differences in peripheral Newspoll data. It has very little power, and we might as well say all successful opposition leaders need to be four inches shorter than the incumbent but with bushier eyebrows.

It's voting intentions wot matter!

Poll-mix

Me in yesterday's Crikey, here.

September 18 Newspoll says 55 to 45

Tables here, Dennis here.

September 17 First 'hung parliament!' sighting for 2007

During the 2004 election campaign, Glenn Milne repeatedly worried himself (and his readers) sick at the prospect of a hung parliament. It was a very, very possible, very, very scary scenario that Mr Milne warned us of in at least three columns. (I think Gerard Henderson christened him "Glenn 'hung jury' Milne".)

Today in the Oz, Glenn is first out of the barriers with this year's scenario. Just a mention in the first par, but it's there. 

(The article itself is interesting, but the Ireland-Australia comparison is dodgy because Rudd and Swan are forever talking about setting the country up for future prosperity: education, broadband etc. And the gambling analogy is all wrong; a bet of E1000 to win E10 is probably more attractive than one to win E10,000, assuming the odds reflect the likelihood of success. Voters prefer small risks to big ones.)

(This could almost re-run word for word.)

On 'Bellwether seats'

Reader sends this sensible email on the Eden-Monaro talk.

Antony Green expressed similar sentiments, apropos of a state election, a couple of years ago. 

Median household income in Melbourne seats

Here's the equivalent table to Friday's Sydney one.

You could say a similar story 2004-swing-wise to Sydney, but overlayed with state-wide movement to the Coalition - that is, the biggest swings were at the bottom and the "outer middle", the famous Aston being the biggest swinger of all.

In similar vein to Sydney ponderings, I reckon Labor has a better chance in 2007 in Kooyong or Higgins (which it has never held) than Aston or Casey (which it has) or even Dunkley.

September 15 Sydney seats 

This qualitative Nielsen stuff in the SMH is interesting. But I wonder whether the "young families in middle-income (western Sydney) marginal seats" demographic is getting a bit hoary.

Table shows Sydney seats ranked by median household weekly income. Note how high incomes tended to swing [the green number in square brackets] to Labor and low ones to Libs in 2004. Lindsay and Parramatta, where Nielsen surveyed, are 16 and 17. (And a memo to folks who still claim John Howard "took Labor's heartland": see the bottom 11 seats, which have never voted for Howard. [Rural-ish Macquarie the exception, although on current boundaries it has remained Labor - just. See '84-'04 graph.])

Red of course indicates Labor-held and blue is Liberal. (Notional numbers.)

As you know, in every state in the country there exists a swag of suburban voters who over the last five to ten years have backed Labor at state level and Libs federally. They've been called "Beattie's Liberals", "Carr's Liberals" etc, but I describe them incumbent-lovers. While they provided the back-bone of Howard's 1996 win, they only went to state Labor once they were in office.

"Rudd's Liberals"

Lots of people reckon the "outer middle" will provide the bulk of the pro-Labor swing this year, but I don't, because they're conservative and (relatively) likely to stick with the devil they know. Instead, I suspect that Libs in leafy streets who have never voted Labor in their lives might be over-represented. 

Might even earn the moniker "Rudd's Liberals". (In fact I'm sure they will.)

This is why I reckon sitting member Malcolm Turnbull is more vulnerable than John Howard, and North Sydney will fall before Hughes. (Both of the last two unlikely.)

None of which is to suggest Labor won't win Lindsay and Parramatta. 

September 13 "A vote for Howard is a vote for Costello"

... is a crap mantra. This is much better.

• Seen The Bulletin's election coverage at The Bullring?
• In Melbourne with nothing to do this evening? This might be worth a look.

Julia: a touch of the Beazleys?

Was surprised to see on 'Lateline' last night Julia Gillard reminding viewers of things they like about Howard: his responses to Port Arthur massacre (1996) and East Timor (1999). It took one back to Kim Beazley early in the 2001 campaign describing the PM as "the most considerable conservative politician the country has produced and a substantial servant to this nation". The idea in both cases was, presumably, to create the impression of an era coming inexorably to a conclusion.

It's a terrible tactic, born of the fuzzy post-hocery that comes after every election, which sees the result, whatever it was, as having been inevitable. If we can create that feeling of inevitability beforehand, the thinking goes, the result will follow. 

Carts and horses (again)

If Labor wins this year, it will seem with hindsight as the endpoint of unstoppable momentum for change: Howard did some good things but his time was up. (As it would have been had they prevailed in 2004.) And a Howard re-election will be viewed, again, as inevitable: the man is unbeatable.

But trying to create self-fulfilling expectations is pointless - nay, counter-productive. Voters aren't zombies who think "it looks like the Howard era is over, I'd better jump on board". Instead they should be thinking "I'm sick of this bunch and will vote for the other side." 

Try to wrap it in a three-act narrative and you do what Kim and Julia did: provide reasons to stick with incumbent.

September 12 Eden-Monaro: cause and effect?

[Update: reader sends this]

Mr Howard apparently told the partyroom this morning that internal polling shows the Libs travelling very well in Eden-Monaro (NSW). As Eden-Monaro has been won by whoever formed government at every election this side of 1969, QED the Libs are actually doing well overall, etc etc.

I'm sure they don't go so far as thinking that if they can win Eden-Monaro, the rest will follow. That would be in line with Crosby-Textor's apparent belief that if you can get people to expect you to win, it will happen. (Ok, that's overstating it.)

Macarthur used to have the bellwether reputation, having gone with the winner at every election since its creation in 1949. But even though the pattern remains in tact, most people recognise the seat has changed almost beyond recognition. 

New jewel in crown

Apart from the usual - demographic changes, history doesn't always repeat - there's a good reason to expect that Eden-Monaro can stay Liberal while Labor wins government. 

It's that NSW is no longer the jewel in Labor's crown. Labor doesn't need an outstanding NSW result to take government anymore. Victoria, which for decades pulled the other way, has at every election since 1977, with two exceptions (1990 & 1993), given a higher Labor two party preferred vote than NSW. Previously this was unthinkable. 

Maybe Victoria is the new NSW (albeit with three quarters of the seat numbers).

Graph of NSW, Victoria and Australia 2pp 1949-2004

Graph above shows NSW, Victoria and National Labor two party preferreds 1949-2004. The first NSW vote that tips the 55% line is in 1961, then in 1972 it sneaks over 55%, dropping a little in 1974. 1983 is almost as high, and 1993 gets even closer.

The main point is that the red line is sustained less and less by NSW and more and more by Victoria.

September 11 Bits 'n pieces

• Neil Brown in the Oz goes for double or nothing. Fool! He'll leave empty-handed.

• See the final sentences in this three year old Canberra Times piece (with odd formatting) from three years ago. Advice now redundant, of course.

US Presidential bets

Someone who knows these things explained to me the other day why Hillary Clinton will be the next US president: in short, none of the electable Republicans can win nomination. (Previously mused thus.)

He reckoned Newt Gingrich, while an outside nomination chance, could change his spots (after nomination), move to the middle and be contestable.

So I wacked a few bucks on Newt at 66 to 1 (and on Fred Thompson @10 to 1, largely because of his 'Law & Order' gravitas).