December 9 NSW cumulative swings 1993 - 2007
Further to Friday's post, have extracted NSW seat 1993-2007 swings. That state's component of the Labor two party preferred vote was almost identical in 1993 and 2007 - 54.4 and 54.1 percent respectively.
But the total seat hauls were very different: 33 out of 50 in 1993 but only 28 out of 49 in 2007. Table below only contains seats that existed at each election over the 14 years, and never had an independent member.
NSW electorate net swings from 1993 to 2007
These are the addition of 1996, 1998, 2001, 2004 and 2007 swings to Labor. Table shows: Seat current margin, AEC category (r = Rural; p = Provincial; o = Outer metro; I = Inner metro; see here and here), cumulative swing 1993-2007 to Labor, [2007 swing to Labor] Click any seat for Antony Green's page, including results.
Eden-Monaro still bells, but not Macarthur
First thing to note: Eden-Monaro (rank 26) continues to take its bellwether status seriously, recording nearly zero net swing over the 13 years, while everyone's former favourite bellwether, Macarthur (no 2), has shifted nearly 8 percent into Liberal territory.
The Coalition appears to have done well out of seats' shifting alliances (having held only 17 seats 14 years ago and 21 now) - the top five blue seats were all Labor-held in '93.
Bennelong and Richmond
At the other end of the table (the bottom, of seats that have moved towards the ALP) only Bennelong (rank 40) has gone Labor. The rest are too safe (either Liberal or Labor) to matter.
While Bennelong is probably the sort of seat that will tend to go with whoever wins government from now on, Richmond (42) is more solidly Labor, it was won by Labor in 1993 and has moved by over 5.2 percent since then.
For all Australian seats, see post below.
December 7 Cumulative swings 1993 to 2007
The last time the ALP won an election was in 1993. I've calculated the cumulative swings in each seat at the five elections since then (for seats that existed at each election) and put in this table.
Perhaps most interesting is Sydney's outer fringe. While the Labor NSW two party preferred votes in 1993 and 2007 were similar - 54.4 and 54.0 respectively (ie it was actually higher 14 years ago) - Hughes, Macarthur and Greenway top the list (if you discount seats that have had independents, which mucks up the 2pp) of swingers to the Coalition. Lindsay and Dobell are also near the top.
What happens at the next election in these seats will be interesting. But that's a long way away ....
Queensland and South Australia are well-represented at the other end of the table - biggest 14 year swings to Labor - but then the ALP did poorly in those states in 1993.
December 6 Tim Gartrell's Press Club talk
The main theme is that the opposition won it - with a plan for the future - the government didn't lose it.
If I were to summarise the election with two cliches they would be "it's time" and "small target" (the latter is aka the silly and thankfully now gone "me-tooism").
But that is not to take away from all the hard work, good campaigning, strategic cleverness, 24 hour working days, discipline etc.
On the other hand, I refuse to believe the Liberal campaign was the shocker that is now widely believed. They changed direction several times, but that's what good campaigns are supposed to be able to do.
Brian Loughnane & co just had a dreadful hand to play, and they probably made the most of it. Like Gartrell & co in 2004.
About the pendulum
Results update: big vote but modest majority
The latest numbers have Labor on 83 seats with 52.8 percent of the two party preferred vote. A comfortable 16 seat majority - but a piss-poor seat return for a vote that would normally give a landslide.
By comparison, in 2001 John Howard got 82 seats with just 51 percent of the vote (and three of the seats he didn't get would have gone to the government if independents hadn't won them) and in 1998 a little under 49 percent support got him 80 seats (out of 148).
This low seat for votes equation is generally an opposition's lot. And while the on-paper required swing for a change of government in 2010 is now only about 1.7[?] percent, the general benefits of incumbency - including personal votes of new Labor members in marginal seats - should push that margin out.
Howard found the same at the '98 election, when he survived a swing that according to the pendulum would have seen him lose.
December 4 Benefits of incumbency pt xxiv
Here's a post-election Newspoll in the Oz.
Interesting, but be sceptical of after-the-fact surveys, when result is done and dusted, all is seen through the prism of the outcome being "correct" and in some ways inevitable and the parties and people viewed as they are today, not as they were before 6pm AEST on election day.
For example, it turns out the Rudd Opposition was brimming with ideas; the phrase 'me-tooism' has gone forever.
Exit polls are better.
Narratives old and new - just a little tweaking ...
These warlords obviously can't be entirely ignored.
It shows proximity between The Age HQ and Senator Carr's office. Just up the road really, with lots of coffee shops and pubs in between. You can join the dots ...
Peter Costello, in perhaps his final heavy-breathing Lateline performance last night, noted that the Labor win is not as large as it appeared on election night.
True. The AEC now has the two party preferred votes at 52.9 to 47.1 and narrowing, which is just a smidgin wider than the Coalition's win last time. Antony Green currently has the number of Labor seats at 85. On election night these were looking more like 53.5 to 46.5 and 87.
In this table of national elections since 1919 (introduction of preferential voting) ranked on the winner's two party preferred vote, last Saturday's vote slots between rankings 15 and 16. Labor's 56.7% of seats sits further down this list from 1910 (beginning of two party system) - between 23 and 24.
Reasons for the narrowing?
The final Newspoll and Galaxys had 52 to 48, and Morgan's election eve poll said 53.5 to 46.5, but until the final week the opinion polls constantly pointed to a much bigger result.
It is pointless for us to try to pinpoint "occurrences" in the final week that caused voters to move back to the government. Those undecideds may have simply decided to stick with what they knew. Perhaps the government's messages from weeks one to five finally sunk in, or Kevin should have appeared on Insiders, or perhaps something that happened six months ago kicked in.
Kevin Rudd's high approval ratings
Last year I suggested that, contrary to popular belief, high approval ratings in an opposition leader don't make voting intentions more sustainable. On the contrary, the relationship seems to be inverse, and I suggested a way of looking at it was that high approval artificially inflated voting intentions.
My point was that on election day, opposition leaders with low approval ratings, eg Howard in 1987 and Peacock on 1990, got more votes than the polls had suggested, while those with high approval - Hewson in 1993 and Latham in 2004 - got fewer.
Last weekend's result certainly fits into this thesis.
November 30 Liberal structure
Back in early 1995, new opposition leader John Howard responded to nagging about the poor state of the Liberal structure by observing that there was nothing wrong with the party's organisation that a decent win wouldn't fix.
A year later he was proved correct (in a sense) and from March 1996 until about a year ago, it was the ALP's dreadful organisational structure, miniscule membership and all round inbrededness that meant they would keep preselecting duds and, perhaps, never again take power federally.
True, Labor dominated at state level, which meant giving the story a tweak: the Liberal structure was still hopeless as well, but sort of only at state level; they still managed to get highly talented, knockabout folks to Canberra.
And Labor similarly got top people into the state arena for ... some reason or other.
So here we go again. Labor brims with talent, the Libs are hopeless.
They also, like Peter Garrett, add breadth to the general parliamentary pool, having first carved themselves successful, interesting lives.
But Brendan's chances of ever becoming PM took a dive yesterday, while Malcolm's now look sunnier.
Several years ago, Michael Duffy wrote a book about Mark Latham and Tony Abbott. He called them the "best politicians of their generation" or something like that. This probably meant he liked them both, but what they have in common is that they are energetic, smart, larger than life people who are unelectable as opposition leader.
There is no way on earth the Liberals could win an election from opposition with Tony Abbott as leader. I'm sure you know this.
November 29 Gap narrowing
Malcolm Mackerras mentioned this in today's Crikey. As the count progresses, the Labor lead narrows; the AEC now has 53 to 47 and Antony Green is projecting 85 ALP seats. I think William "Pollbludger" Bowe had a prediction of 84 seats, but can't find it now.
Election day photos
Here at The Democratic Audit of Australia.
If you agree this is highly unlikely, then let's pose the counterfactual: that Howard did hand over to Costello, say last year. Assuming that Costello went on to lose this year, what lessons would the Libs - and the commentators - be drawing today?
Well, they would conclude the handover had been a mistake. They would be now saying that if Howard had remained leader the Coalition would have won. Howard never lost an election before; he would have had Rudd on toast.
This belief, which we know is wrong, would have left the party in a dysfunctional place, trying to recreate the putrid Howard brand of politics.
Instead, because they know the truth, they are able to break with that past.
Our national storytellers would also still be banging away: Howard remains a sublime communicator; only he represents the true Australia; we'll never see his like again.
Costello copping the blame
Peter Costello's brief tenure as PM would be viewed unfavourably, and in fact many Libs would blame him for forcing the leadership issue.
So there would only be one person benefiting today if John Winston Howard had handed over the leadership last year - and that's John Winston Howard.
Everyone else - his party and the country - is better off because of his stubbornness.
November 28 My bets
I had a decent amount riding on a Labor victory, with the fantastic overall payout of $2.60. (Was about $1.20 last week.) They were all placed when Beazley was leader.
My seat-betting over the last few months is mixed. I backed a few dreadful nags but also some good ones. Probably came out a little ahead; too many to mention, but the overall amount is insignificant compared to victory payout.
Also won some bottles of red from a few people you might have heard of. Just getting their permission before posting; will do so in the next day or so.
November 27 Swings by seat by state
As you know, throughout the campaign the government clung to the hope that they could hang on with just 48 percent two party preferred. Did that number turn out to be correct?
So in uniform terms that 48 percent looks more like 49. [Update: final figures 52.7 percent and 83 seats, but 49percent remains about right.]
Malcolm Turnbull didn't get to where he is without having a high opinion of his own abilities, and so probably over-estimates his influence over these things.
But such an election would squash his political prime ministerial aspirations, probably forever.
November 26 Seat swings: business as usual
Three pieces in Crikey this morning. I've sorted the seats by 2007 swing and included a couple of other columns.
What stands out: NSW and Qld did the heavy lifting in both the 1996 and 2007 changes of government, and generally (to my surprise) in the same seats.
Costello bows out
A wise decision by Peter Costello; who can blame him? (See last line here). He retires from public life with the last public memory being a widely respected Treasurership. If he had taken the leadership he would leave politics a disappointed man and perceived political failure. Do you remember Kim Beazley the minister or the later one?
I suspect Peter knows he's already dodged a bullet.
Both Costello and Lord Downer say they're staying for three years, but surely we'll see byelections in Higgins and Mayo next year?
Get used to "oxymoron" jokes about that heading.
Whoever becomes opposition leader will probably never become PM. This is one of the conundrums of two-party politics: the best and brightest becomes leader after the wipeout and so trashes their political future.
A rational approach would be to keep Malcolm Turnbull on ice and let someone else inherit this most horrible of jobs for one election at least. Both Brendan Nelson and Tony Abbott could play the hapless, hopeless, out of touch opposition leader in their sleep, but ... Abbott somehow deserves the fate more.
(Did I mention the flip-side of the pledge?)
Antony Green is currently estimating 86 seats [update: 88; update: back to 86] for Labor; my final prediction was 89.
My prognostications on which sorts of seats would swing where now look flaky. The "outer middle" did shift big-time, and leafy Libs mostly stayed where they were.
Morgan's final poll - 53.5 to 46.5, with very good primary numbers too - appears to be the closest, and that's the figure the final poll-mix churned out as well (see below).
Vote result is bigger than Hawke's in 1983, but is smaller seat-wise (allowing for different size of HoR). Numbers can change over the next week.
Best Freudian slip: Kerry O'Brien's "swing to the ABC".
Best early call: the Mackerras double of overall result and Bennelong.
More, perhaps, later today.
November 24 Election day: Morgan's final poll 53.5 to 46.5
Poll-mix of all final week polls is 53.5 to 46.5
(Same as Morgan's final evening one.)
My final prediction: 89 seats for Labor
That's down from my previous 91.
That is, I'm joining Malcolm.
The final polls in 2004 and 2007
First table below has the final polls at the last election, in 2004. You'll recall that Morgan and Newspoll came a cropper with preferences; today they notionally allocate them. The last column on right is my notional estimate of Labor 2pp to nearest half a percent.
A Nielsen which was out of wack with everyone else, and looked at the time too big to believe, proved one of the two best performers (the other was Galaxy), and it pulled the average into the right direction. (The word around the traps is that their respective unrounded Coalition 2pp numbers were 53.6 and 51.8, which makes it almost a tie.)
Applying the poll-mix (which is pretty well the same as averaging the right hand 'me' column, plus a little fiddle for sample size) came to (to nearest half a percent) 52.5 to 47.5, which was of course very close to actual result.
Final polls in 2007
This year, because everyone but Nielsen estimates 2pp, my Labor two party preferreds are close to the pollsters'. In each case mine are the same or smaller.
It also looks like Greens are going to get about 8 percent - up one point from last time.
Satisfaction ratings for opposition leaders
And finally, ...
A year ago in the AFR I noted that, contrary to received wisdom, opposition leaders with low approval ratings tend to do better on election-day than their surveyed voting intentions, while highly approved leaders perform more poorly. I suggested that "high approval [might] artificially boost voting intentions.
Mr Rudd's Newspoll two party preferred support has averaged 55.6 percent, easily the highest of any opposition leader since the 1980s, and probably the highest ever.
An ALP 2pp of 53.5 would provide support for the thesis.
Have a nice day
And don't forget to vote!
Remember, if there's a big swing on, even if it's "only" five percent, it won't be even, and will pop up (and down) in all sorts of places.
November 23 4pm: Newspoll: 52 48
Here in the Oz. That grapevine has it that primary support is 44 to 43, Greens on 7.
Galaxy and Nielsen
Galaxy's neck and neck primary vote is so out of sync with everything all year as to be incredible. But the pollster has a good record.
Here's this morning's Crikey poll-mix.
Portlandbetting on seat margin and two party preferred vote
Maybe I'm slow off the mark, but discovered last night Labor seat number and two party preferred betting. At time of writing (9am) Labor's odds were shortest on a 2pp between 53 and 54, and 85 seats.
Laurie Oakes put the ALP's chances of winning at 60%. I put them at about 90%.
Crikey is collecting its commentators' predictions for today's main edition, and I've upped mine to 91 seats for Labor in the House of Reps, the same as my August entry in The Monthly.
Tomorrow we should see the final Newspoll; I may adjust from there.
Now, on the eve of the 2007 election, I make this pledge.
My pledge to you
If the Howard government is returned tomorrow, I will post the goose here. It will be a great big fat goose, and it will signify me for my breezy, self-assured predictions of a Labor victory - not just this year, but (in less strident language) the year before.
I pledge to be circumspect and humble and not to attempt to bluster my way back to credibility. My confidence in my own psephological prowess will be irreparably diminished, and I certainly won't offer lightning quick explanations for something that I simply did not believe would happen
Newspoll tomorrow; final seat prediction then. Right now it's 91 seats for Labor.
November 22 Newspoll's final cumulative data
In the Oz today. They've polled a humungous number of people over the six-week campaign - 1700 a week comes to about 12000 - and are going to have some great data, for age cohorts etc.
November 21 Richo's call: a 20 seat gain for Labor
Graham Richardson has predicted Labor will gain 20 seats, enough to take government, with four to spare. That would give them 80 out of 150 - a ten seat majority and, contrary to how it's being reported, in the "close result" camp.
Only 1990, 1974 and 1961 were closer in recent memory.
Richo also stated this would require the biggest pro-Labor swing in federal history (or something like that). This is wrong - see 1969, 1943 and 1929.
It's not the first time he's made incorrect numbery statements in this, his first tentative year out of hibernation.
That cultivated "numbers man" persona was probably always rather dodgy.
Costar: election will "go down to the wire"
Actually, Brian Costar doesn't say that at all.
When the died, the SA Public Service Review
described him as “though reserved, … always courteous and affable". It
noted his “prowess upon the cricket field” and “remarkable intellectual
facilities, indicated by his broad, high brow”; he was “a signal example of
the advantage of the great public school and university system of education in
Boothby in 2007: a 'Tiser poll
Boothby was almost certainly the only electoral official in the world to have a seat named after him, and it is there that an Advertiser poll today has the Liberal incumbent Andrew Southcott ahead of Labor's Nicole Cornes 52 to 48.
It has an ok sample size of 617; primary support is 41 to 32 (no further detail).
Sometimes 'Tiser doesn't extract 'undecideds/don't knows'. If they have, then that's a lot (about a quarter) of minor party/independent votes, and they seem to have been notionally allocated about 60 40 to Labor, ie the nationwide flows in 2004. But in in 2004, preferences in Boothby flowed about 67% to Labor, which applied to today's poll comes to a 2pp of about 50 50. On the other hand, if there were about 10% that should be excluded, their preference allocation is closer to that 67 33.
But what's Independent Ray McGhee on, and where might his preferences go? And the Greens? Still, it would be difficult for Cornes to win if she lags by 9 primary points.
[Update: a reader primary vote data: Lib 41 ALP 32 Dem 1 Greens 9 Family First 3 McGhee 2 Other 1 Informal 1 Undecided 10.
So, yes, the silly sausages have again left the undecideds (and informal!) in. Extracting them puts Libs on 46, Labor on 36, Greens on 10 (others about the same) - an easy Liberal win.]
Morgan in Western Australia
Survey of 435 voters across 5 marginal WA seats comes to 50.5 to 49.5. Mr Morgan's report is headed "The Swing Back To The Liberals Has Started In WA" and he describes a "swing of 0.8% to the Liberals since the 2004 election".
Morgan estimates notional two party preferreds, and with Greens on 9.5 compared with about 6.5 in 2004, this one seems overly favourable to the Libs (I would make it about 51 to 49.)
"About status quo" would be the safest description of this survey result.
November 20 Me in the Financial Review
On marginal seats and the total two party preferred vote. Dummied up here.
Out on a limb: me on ABC News Radio
Talking, again, about marginal seats etc. Was asked what I thought the result might be, and because I cannot tell a lie, I told them what I thought.
The increase has come at the expense of Labor, but the two party preferred effect is minimal. And whereas Newspoll was previously over-stating the Labor lead a little (from the primary numbers), now it's about right.
In other words, the "narrowing", such as it is, is over-stated.
[Update: the unverified word is that Newspoll has joined the other pollsters in including Greens in their voting intention question. This would explain the increase. ]
November 19 Jackman's poll-pooling
Simon Jackman has pooled those polls; he gets 54 to 46.
The Long March of Sydney's fringe
[Update: a longer, better version of this post has been published in Crikey.]
This graph shows Labor's two party preferred votes from 1993 to 2004 in seats on Sydney's fringe, every one of which was held by Labor 14 years ago but is now held by the Libs. Votes are
Some folks like to call them "battler" seats, but "aspirational" is better, or even just "mortgage belt".
The thick red and purple lines are Australia-wide and NSW Labor two party preferreds, and the main lesson to take from this picture is that every seat was, in 1993, above the Australia-wide one, and all but two were above the NSW line, but today every seat is below both those lines.
That is, they've been pretty consistently moving further to the Libs than everyone else, averaging 14.8 percent over those four elections, compared with 4.2 percent nationally and 6.3 percent state-wide.
I reckon of this lot Labor will probably get Lindsay and maybe Robertson and Dobell (in that order) but that's it; it's hard to believe they would suddenly turn around and swing to Labor in a big way.
But if the NSW vote is big enough, seats will come - but from elsewhere: Parramatta, Eden-Monaro, up the north coast perhaps ( Page, Paterson and maybe even Cowper), Bennelong, Wentworth, perhaps even North Sydney ...
Get the forest right, and the trees will follow.
(Last week I suggested Labor might win nine seats in NSW. I've named 11 possibilities above, plenty to choose from.)
Undecided voters on Radio National
Spoke with James Carleton on Fran Kelly's brekkie show this morning. (Pre-recorded, to be precise. The story starts about eleven minutes into the second hour.) [Update: here's a stand-alone 900kb MP3 of the piece.]