Go to next posts

+

=

        Mumble elections

December 2003: Go to post redistribution pendulum. It's a corker!  

November 3

What Senator Brandis was up to

What was that George Brandis up to last week? The consensus seems to be that the Libs are worried about defections from their "progressive" base to the Greens, and hence the seemingly bizarre comparisons of the Greens with Nazism. I suspect it's something else: to turn the Greens into Labor's One Nation.

The Green's are, on balance, a plus for the ALP. They mop up disaffected Labor voters and return them to the fold. Without Bob Brown's party on the ballot slip, many of these find the ALP so uninspiring they would be tempted to vote informal. Brown gives them a reason to mark their slip, and once they have, compulsory preferential voting dictates that they number every square. (If they just give Greens a '1' their vote is informal.)

Numbering every square means eventually choosing between Labor and Coalition, so they go for the lesser of two evils which, in practical terms - assuming the Green candidate doesn't actually win the seat - is the same as giving the ALP candidate a full vote.

So what are the Libs doing? Well, back in 1996-98, the Coalition initially quite liked One Nation. It too gathered in disaffected Nationals, plus pilfered some of Labor's primary vote and leaked it to the Coalition. One Nation was initially a net plus for the Coalition.

Directing Preferences

However, they got into all sorts of trouble with "directing preferences" (advising voters who to put where on 'how to vote' cards). Urban Liberals, appalled at Pauline Hanson - and perceived damage to Australia's standing in the region - advocated putting One Nation last. (In 1998 Peter Costello unilaterally declared that would happen in his seat of Higgins - and earned a public rebuff from the Prime Minister.)

But rural Libs and Nationals found the idea of putting Labor ahead of One Nation even more intolerable. 

The 1998 Queensland election decided the matter. The state Coalition government swapped preferences with One Nation and suffered an urban backlash that lost them government. (The 2001 Northern Territory election saw something similar.)

That jolted the feds into action, and One Nation promptly went last on all Coalition ballot slips in the October 1998 federal election.

Fast forward to today. If the Coalition can paint the Greens up as such a danger to society that no politician with a scrap of decency should have anything to do with them they can turn the tactic back onthe ALP. George Brandis was keen to make the link between the Greens and One Nation last week. 

It seems a big ask, but the idea that the Greens may win a few seats and hold the balance of power might get traction.

Anyway, if this thesis is correct, watch out for Coalition boasting of putting Greens last - and demanding Labor does the same.

November 1

Louise Dodson in The Age repeats the old furphy (apropos of Greg Barns' book) that John Howard "has successfully redrawn the electoral map in Australia, with the Liberals now holding many blue-collar working-class seats previously held by Labor".

"In the 1996 federal election, the Liberal Party moved into Labor seats and now holds most of the 50 lowest income seats in Australia as well as the 50 highest income seats, Barns says."

Absolute bollocks. Read also the blurbs for Lindsay and Parramatta on pendulum.

October 26: See the newest pendulum. A thing of beauty.

October 24. Morgan says 50 50

October 21.

Return of the dreaded worm. Gary Morgan dusts off his beloved for the Oakes-Bush interview.

Newspoll today has Labor ahead after preferences, 51 to 49.

Dennis Shanahan writes "Labor is now in its best position to win the next federal poll since the last election campaign."

Even within just Newspoll parameters, that's not true. This table shows Labor further ahead (rounding to the same) several times last year. But back then, Newspoll wasn't distributing preferences (see this article, after which they changed their ways. See also the seventh par - "There is a small if ... " - in this, after which they calibrated their ways further).

In today's Newspoll, the Coalition's primary vote has dropped to 39 percent, which is its lowest since the last election. Unlike the ALP, the government has few parties from which to get preferences. One Nation was its biggest source last time, but it now only registers one (down from 4 at the last election).

In the notional distribution process, most of the government's 10 percent (to take them from 39 to 49 percent) come from the 12% "other", which are distributed as they flowed at the last election: 52 to the Coalition, 48 to Labor.

But back then there were only 4.4 percent "others". So who is this lot, and where will their preferences go? It's anyone's guess. If they're rural independents, they'll probably heavily favour the government. If community-based lefties, they'll help Labor. 

It'd be nice if Newspoll started measuring preferences - rather than notionally distributing them. ACNielsen and Morgan do.

October 20. The federal ALP's continuing quest to keep losing, part xxxivvv.

Nationals deputy leader John Anderson has all but spelt it out: party polling shows Coalition support is soft in the regions. See pendulum. This is a worry; thirteen rural seats would fall to Labor with swings of less than five percent, more than enough to put Crean in the Lodge.

"Peter and Lindy" are missing out on the growth and crazy house prices. Their votes are up for grabs. Hence the refocus, with suggestions that infrastructure and services spending should come before tax cuts.

And the opposition's response? Shadow Treasurer Mark Latham lambastes the government for fiscal slackness. Labor's razor gang promises to cut regional projects. They don't really want Peter and Lindy's vote.

Latham keeps talking about free markets. That's dangerous talk for an opposition. He spends his time chatting to four wheel drivers in outer western Sydney. He thinks they're the key to victory, but they're not.

Queensland pendulum up.

The latest addition to the post-distribution works in progress is the Queensland portion of federal pendulum. Eventually there will be pendulums for all states and geographic classifications, with links to seat descriptions. See all pendulums so far.

October 15. Bob Carr does John Howard.

This is about Carr flirting with federal colleagues by showing he can meet and equal the PM at his own game. That game: play the news cycle by (a) saying something that appeals mildly to xenophobia; (b) the next morning, deny you meant anything at all ("I don't care if they came over on the Sirius"); and (c) turn on the radio, sit back and enjoy the high octane talk back.

He's telling everyone he knows what to do and would have no qualms doing it.

More on Carr.

October 14. Oh. No Newspoll today. One next week, apparently.

October 10. Post distribution pendulums. Work in progress.

Morgan has 50.5 to 49.5

October 8

Coming soon (this week), brand new, all singing and dancing post distribution pendulum. With mini state and socio demographic pendulums. Lots of nice colours. Should be up by Friday - time permitting. 

October 4

I got this email yesterday

Dear Mumble

I don't know if you heard Gerard Henderson's comments on Newspoll this morning.

His argument was that despite the 50/50 result, the Coalition is comfortably
in front because it has a 6% edge on the primary vote compared to the ALP.

Henderson therefore stated that the ALP vote is 'soft' and it is more likely
to change over the next few months before the election.  He quoted that Green
preferences flow about 70-80% to the ALP, the Democrats a bit less but still
favour the ALP.  But the rest can go anywhere.  Preferences intentions can
change especially for independent candidates and One Nation.  He also said
that there were only about 6 seats that were won on preferences on the last
election of which 4 went to the ALP.

Henderson stated that for the ALP to have any chance it needs to get its
primary vote above 40% which is about what Beazley got in 1998.

Regards,

I didn't catch the segment, and ABC online doesn't tell us much. My take on recent polls is in Fin Review today and posting below. But while a low primary vote is of course never as good as a high one, it's not the problem per se. Rather, 50 percent after preferences wouldn't have been enough for Labor at the last election, and at the one before 51 percent wasn't enough, so 50 percent probably won't be enough next time.

The point about few seats being decided on preferences at the last election is strictly correct but open to misinterpretation. Usually at federal elections you get a few seats going both ways on preferences. That is, several have Labor ahead on primary votes but are won by the Coalition and in several the Coalition leads on primaries but Labor wins.

In 2001, they all went one way. Apart from three cornered contests - with both Liberal and National candidate - only Labor came from behind on primaries to win. The total number of seats decided this way was historically small, but that was because almost a million Green and Democrat preferences flowed to the ALP and prevented the Coalition coming from behind in any seats.

No mistake, Green and Democrat preferences saw off a landslide in 2001.

And on Hendo's last point, the ALP won with under 40 percent of the primary vote in 1990. Their two party preferred vote was 49.9. That's the only time on record Labor has lost vote but won election. Very difficult for them to do it this time.

A Lies and Statistics piece today deals with opinion polls. Will post it here next week.

Five hundred words is not enough to include everything I want to. A few points that couldn't be made were:

• Gary Morgan reckons that his 2001 polling was accurate but voters changed their minds in the last few days. This is not beyond the realm of possibility. Their polling still tends to favour Labor more than the others, and it will indeed take an election to sort it out.

• It is ludicrous to interpret a poll that shows 51 to 49 in the government's favour as saying that, because that was the result in 2001, the position is the same as it was at the last election. It's also wrong to take from the  next fortnight's showing 50 50 as evidence that Labor has had a good two weeks.

        Instead, we should take them both together to say that it's pretty close but the Coalition
        is probably ahead.

• In any event (as noted above) 50 percent is not usually enough for Labor to win. In single member electorates there is no particular reason why 50 percent plus one of the two party preferred vote should give you victory. (If votes were perfectly evenly distributed, 50 percent plus 150 votes - 150 being the number of seats - would give you 100 percent of seats.)
  
  Historically, parties getting between 50 and 51 percent two party have emerged victorious about half the time. That is, when the two parties are at 49 to 51 or closer, it's pretty much a toss up that can go either way, and being ahead in votes doesn't give you a better chance of winning.
  
  (A close vote doesn't preclude a big win either. In 1987, Labor's 50.8 percent two party preferred gave them a 24 seat majority. In 1998, 51.1 percent gave them a 13 seat loss)
  
  Another factor is that the Coalition nearly always does better in these close contests. Only once (in 1990) since WWII has the ALP won with a minority of the two party preferred vote; the Coalition has done it four times.
  
  That's the electoral geography. Labor wastes votes in safe seats. So when it's close, the Coalition usually wins.

Sept 30 Newspoll has 50 50.

Naturally, no sign of it in online Oz. (Previously noted it seems to take a six point government lead for a Newspoll to make it online.)

Sept 28. Gratuitous ode to Beazo

At least Kim Beazley understands that an opposition has to present an overarching story - interesting, complementary and cross-pollinating. Only that way can you drag the conversation onto your preferred territory.

Well, Latham seems to get that too, but his story is underbaked, corny and unattractive.

Crean seems to believe that it's about hammering points: if you just bring the conversation back to health and education at every opportunity, the message will sink through to the electorate and, eventually, they'll vote for you.

In reality you just bore them stupid and make them wish they'd never mentioned health and education (in surveys) in the first place. 

Sept 26. Morgan's latest, still 48 to 52.

Sept 25 Three Labor leadership scenarios

Scenario 1   Simon Crean remains ALP leader and surprises many by leading party to a respectable defeat - on a similar scale to 2001. 

Scenario 2   Crean gives way to Mark Latham late this year. He gets a quick though modest poll fillip of - like a couple of recent published hypothetical ones - one or two points after preferences, which mainly come from the Greens. Journos love him, and huge store is put  in his ability to muscle up to Howard on his own turf.

Gradually, however, Latham morphs into Labor's Alexander Downer: a cure worse than the disease, lagging by amounts his predecessor only sweated about in nightmares. The question then becomes whether the government calls a snap poll.

Scenario 3   They Bring Back Beazley, who gets a  jump in opinion polls to clearly lead the PM in voting intentions. Beazley more or less stays comfortably ahead until the election campaign, but many assume Howard the master politician has his measure and will easily flay him on the day. Howard tries but fails, and Beazley wins easily.

Scenario 4   A combination of 2 & 3 above sees Latham replaced by Beazley. That is, like the Downer experiment, a much trumpeted generational change falls in a heap and baton goes to recycled seeming has-been, who goes on to win. 

Sept 23. During the 2001 election campaign, ACNielsen's numbers were closest of all pollsters to the actual result. They also, unlike Newspoll, always measure full preferences.

But Fairfax only commissions a Nielsen survey once every six months or so. One in SMH today gives 52 to 48 (to Coalition). With the hypothetical generic Anyone But Crean ALP leadership, Labor's primary vote jumped from 34 to 39, two party preferred about 50 50 or 49 51.

That's similar to the one percent improvement Newspoll gave a Carr leadership (see Sept 21, below). Most of the would-be vote changers are, alas for Labor, current Green or Democrat voters who are giving their preferences to Labor anyway. But two percent would also come off the Coalition primary vote.

A bit meaningless because people don't really know how they'll feel. 

Over at the Oz, Phillip Adams' rant on the hopelessness of the ALP and consequent desertion of people like him to Bob Brown, includes this. "And don't be too confident, you party-machine men, about second preferences." Actually, they can be, and Phillip is a case in point. He's going to vote Green, but is there really any possibility of his putting the Coalition ahead of Labor further down the ballot slip? Led by Peter Costello, maybe - just maybe - but John Howard? 

The only way the Phillipes of this world can avoid giving Labor a preference is by voting informal. That means not voting Green either.

Sept 21 Newspoll in the Sunday Tele  shows the NSW Premier as Federal Labor leader would deliver an extra ... wait for it ...  one percent, two party preferred, to the ALP.

Hence the headline "Poll Backs Carr for Canberra".

Primary votes 43 to Coalition, 41 to Labor, two party preferred 50 50. (Last weekend's Newspoll had 51 to 49 from primary support of 42 to 36). Telegraph doesn't mention minor parties, but Labor's primary is up by five while  its two party preferred only improved by one, plus Coalition is actually up too, so Carr's extra support is nearly all at expense of smaller parties, in particular Greens.

Perhaps not surprising in light of Carr's environmental credentials. (He is very enthusiastic about border protection, however.)

But no evidence of his taking votes down the middle from the Coalition - his supposed trump card.

Of course, it's just one survey.

Still on polls, Mark Latham's Light on the Hill speech tells of a party of principal, not spin or public opinion. "We are following our beliefs, not the opinion polls".

(This is the guy who, spooked by polling in his seat in the last election campaign, unilaterally called for the reintroduction of corporal punishment in schools.)

One might suggest that good spin is exactly what they need more of.

Sept 16

In parliament today, Alexander Downer made  fun of Crean's record low preferred PM ratings. He said something along the lines of: he (himself, Downer) was pretty unpopular as opposition leader, but he never reached Crean's depths.

That's true as far as the preferred PM rating goes, but not two party preferred voting intentions. The two graphs below show Downer's eight months in the job compared with Crean's 22. (Both are Newspoll.)

Scales for both graphs are 40 to 60 percent You can see Downer (at left) had a glorious start (debuting at 54 percent) but after four polls went downhill. Crean, on the other hand, below, had a very poor start (standard post election leap for the government), improved to have a stretch of leading as often as he trailed - both by small amounts. Then came Bali bombings - that big jump in Coalition support half way through.

As always, Coalition is blue and Labor red.

Downer's (above) highs were more dramatic, but his lows were very low. Crean (at left) has nearly always trailed, but has delivered a closer contest. Downer's support leapt around between 43 and 55; Crean's has been in the range of 46 to 51.

anyway ...

Newspoll again has 51 to 49. Looking at primary votes, it does seem Newspoll is distributing preferences by individual party these days, rather than in bulk. Full marks.

Don't forget, 51 to 49 doesn't mean the same seat result as last election, even though that also saw 51 to 49. 51 to 49 could double the government's majority to about thirty, it could give us a hung parliament, it could even see a Labor win.

(The last scenario becomes more likely if, as everyone seems to think, blue collar workers are flocking to Howard. They tend to be in safe Labor seats and make little difference to seat outcomes. The larger the "battler" component of the government's 51 percent, the more likely a Labor win.)

The polls show it's very close, with the Coalition favoured to win, because when the two party preferred vote is close, the Coalition usually wins.

Mumble's odds!

Here are odds for the next election, with four leadership scenarios.

1. Simon Crean is still leader: Labor 2 to 1 to win

2. Bob Carr is leader: even money (1 to 1)

3. Kim Beazley is leader: Labor 1 to 2 to win

4. Mark Latham is leader: Labor 5 to 1 to win

(I think my understanding of odds stacks up. If it doesn't, I'm saying Beazley is most likely to win an election and Latham least.)

Sept 12

Beattie's performance is even more impressive because Qld, the least urbanised state, rarely votes Labor at state or federal level.

And if you ignored the hype and tried to estimate each premier's inherent "it" factors and political appeal and nous, Beattie might top that too.

Sept 10. Michelle Grattan writes "Crean critics reject the argument that Newspoll's Labor-Coalition two-party vote of 49 to 51 per cent suggests the Opposition isn't doing too badly. This tells little, they say, because it distributes preferences according to the last election, and is at odds with private polling and anecdotal electoral evidence."

These off the record tete a tetes are a worry. Grattan's right that distributing preferences according to the last election is inferior to measuring them, but Labor's leaked polling never seems to bother with preferences distributed at all. They just talk about dire primary support levels. As noted below vis a vis Qld, it's not nearly as bad as they make out. (Although still dreadful.)

Sept 8 Crean's new poll shocker in Queensland

Leaked ALP polling in the Oz shows Labor's primary at 32 percent in Queensland, the Coalition 52 percent. As always with this internal stuff, they don't give us two party preferred or minor party votes. Primary support tells a more dire story, which is the name of the game.

At the 2001 election, the ALP got 34.7 percent in Queensland, the Coalition 45.6. One Nation's 7.1 was the biggest in the country;  Democrats got 4.3, Greens 3.5 and "other" 4.8.

All washed through to 45.1 to 54.9 two party preferred.

Steve Lewis writes: "all of Labor's Queensland representatives ... would lose their seats, based on the results."

That's hard to justify. The pendulum shows the safest Queensland Labor seat, Oxley, has a margin over 8 percent (based on last election; there's a redistribution going on in Qld now.)

So for a uniform swing to sweep up Oxley for the government, the two party gap would have to be bigger than 37 to 63. (Add eight to 54.9, take it away from 45.1). That means the government gets  eleven out of the sixteen points of non-major party preferences.

That can't happen. They're more likely to get about five or six. As I said, we don't know the minor party votes or preference flows, but One Nation is unlikely to be greater than one or two percent (polling done before Hanson verdict).

Those numbers would probably see a swing of about three percent to the government to sit at 58 to 42. In uniform terms - again based on the last election - that's a gain of one Queensland seat for the government.

The funny thing about these internal poll leaks is that each one on a particular state paints a dire position for Labor, but the published national ones show support at similar levels to the last election.

September 6

Belated Lies and Stats piece on Crean's Badgerys Creek decision in today's Financial Review.

Not unrelatedly, it turns out Bob Carr wants to go to Canberra. That explains this week's endorsement - apropos of absolutely nothing - of mandatory detention.

It seems people in Sydney suburbs like Penrith and Ryde vote for John Howard federally and Bob Carr at state level. Explanation mark! Logical solution: Carr goes to Canberra to take on Howard. Carr knows how to relate to voters.

Sydney voters, anyway.

Some context: something is going on in the four eastern states, with Labor governments achieving whopping, unprecedented (in Vic and Qld) majorities. Within that something, Carr is the most ordinary performer. He gets very low primary votes and Steve Bracks, Peter Beattie and Jim Bacon get higher primary votes, two party preferred ones and seat majorities (percentage-wise, and excepting Tasmania which has proportional representation).

Queensland and Victoria are (unlike NSW) traditional conservative states, which makes the Labor performances more staggering, and these guys manage to win without banging on about law and order and immigration.

Oh, but Carr understands Sydney.

Hang on, weren't we talking about winning a federal election? Here we go again: win Sydney and the rest falls into place.

There's probably a simple reason successful state premiers don't do well in Canberra: by the time they arrive, their goodwill and capital in their home state is depleted, and the luck and circumstances that got them to the state's top job stubbornly refuse to come round again. Stripped of the accoutrements of incumbency and power, they lose that certain something and become mere mortals again.

Try to imagine John Howard as, say, Victorian opposition leader. That's how Carr might look as federal one.

Still on the subject of Sydney ....

Parramatta

What's in a name? In the case of the seat of Parramatta - plenty. Parramatta is the best known western Sydney suburb to most non-westies - ugh boot territory, that sort of thing.

The seat of Parramatta was won by John Howard's Liberals in 1996 and retained ever since. This of course is seen as evidence of "Howard's battlers". 

It's nothing of the sort, for a couple of reasons.

First, Parramatta has been held by the Liberal Party (and its conservative forebears) for 80 of its 102 years in existence. It has never been Labor heartland.

More importantly, Parramatta the seat includes Parramatta the suburb, but one ain't typical of the other. Parramatta the suburb is still strongly Labor - the booth of Parramatta voted 67 percent two party preferred for Labor in 2001. Parramatta is a diverse electorate, which the Liberals usually win because of strong Liberal booths in places with names like - fittingly enough for today's PM - Winston Heights, Winston Hills and Northmead.

Speaking of Howard, his seat of Bennelong snuggles up to Parramatta. That gives you an idea of its socio-demographics.

It's harmless for the media to get carried away with the myth of  "Howard's Battlers" in Sydney's west. They have to write about something. But the really bizarre thing is that the ALP seems to believe it.

And if the electorate was called, say, "Winston" it might be seen  for what it is: Liberal territory that occasionally strays into Labor hands. See also this.

September 6 

Preferred PM

Dennis Shanahan in The Australian takes Simon Crean to task claiming this week that "The preferred prime minister is always the incumbent, always has been, always will be."

Shanahan reckons this is a "desperate act of denial", because "since 1987, Opposition leaders have regularly been in front of the incumbent: John Hewson (over Bob Hawke and Paul Keating in 1991-93), Alexander Downer (over Keating in 1994), John Howard (over Keating in 1995) and Kim Beazley (over Howard in 1998 and 2001)."

They're both wrong. Crean's "always" rules his strictly incorrect, because the opposition leader has sometimes been preferred. But Shanahan's "regularly" is more wrong in spirit.

It is unusual for opposition leaders to be ahead as preferred PM. Why Shanahan prefaces his numbers with "since 1987" is a bit of a mystery, because he then ignores John Howard and Andrew Peacock, who never got close to Bob Hawke as preferred PM; he jumps straight to John Hewson in 1991, who did.

What is true is that there have been periods within which the opposition leader was ahead of the incumbent. One of those was during the 1993 election campaign, which the government won. One of them wasn't the 1996 election campaign, which the government lost.

Anyway, here's a graph. And this old AFR article.

About this week's opinion polls.

Friday's Morgan poll got a decent run in the Sun Herald and elsewhere, neatly supporting a theme of damage to he government from a succession of stuffups - ethanol, Tuckey and now Abbott.

But as Morgan's graph shows, they've had the ALP in front more often than not since the 2001 election; it's just that Morgan's currency has been devalued since its performance then, when it consistently predicted a comfortable Labor win, and commentators generally don't pay them much attention.

The Newspoll shows, yet again, 51 to 49. What that really means is that if an election was held last weekend, and if the way they were held was that the AEC rang people up and said "hey, you didn't know this, but there's an election today - who are you voting for?", the two party support would have been between 49 to 51 and 53 to 47.

Probably. There's one chance in twenty that it wouldn't have been. (That's a two percent error margin with 95 percent confidence.)

That means the government probably would have won an election last weekend. But if you consider that Newspoll's final survey in  the last election campaign showed 53 to 47, and the result came in at 51 to 49 ... well, these numbers could mean anything. And 51 to 49 could easily result in a Labor government, if it was distributed properly.

It's unhealthy that Newspoll gets all the attention, especially as The Australian, which reports them, has a tendency to ... draw conclusions that aren't there.

Saulwick's preference distribution is no good.

In the interests of diversity, Fairfax should commission regular ACNielsen surveys. ACNielsen and Morgan are the only two who give preferences the respect they deserve. ACNielson was closer to correct than Newspoll at the last election.

Sept 2. Today's Newspoll in the Oz has 51 to 49. Like Morgan, it has One Nation up to 2 points. Democrats down to 1 and Greens at a relatively low 6.

August 30. Morgan has 48 to 52. One Nation has gone from 0.5 to 2 percent, a whopping 300% increase, but still nothing like its 8 percent vote in October 1998, or even 4 percent at 2001 poll.

As we speak, Newspollers are polling, to be published in Oz on Tuesday. Wonder what One Nation registers there? Where will any jump come from - Morgan's numbers suggest it hurts Coalition more than Labor.

August 19. Newspoll out today, in The Australian. There seems to be a Coalition lead threshold of six percent required for web publication; like last week the 52 to 48 doesn't make it online. (Ie no link.)

August 16. Morgan's latest: 50.5 to 49.5.

Gary Morgan says that means that "if a Federal election had been held in early August the L-NP would have won".

In reality, numbers that close, in fact anything in the range of 51 to 49 and 49 to 51, are too close to call, although probably pointing to a Coalition win. 51 percent of the two party preferred vote can give you a 24 seat majority (ALP in 1987); it can give you a 13 seat loss (1998).

The Coalition usually gets the better of the vote-seat equation, because the ALP wastes votes in safe seats. However, if what many seem to believe - that the "Labor heartland" is turning towards John Howard - is true, that wastage is minimised.

If Labor's safe seats are swinging to Howard, then for any given total vote, someone must be swinging the other way. If ALP support is up just a little in, say, marginal rural seats - where votes really count - then 50.5 to 49.5 could well translate to an ALP victory.

That's if you believe Morgan's numbers. There's a Newspoll being conducted this weekend. Will Hambali trump Honan? 

August 10. Sunday Age says Howard will go full term.

Three years exactly from the last poll will take us  into early November 2004. Early November 2004 is when the next US Presidential election is.

Them's big coat-tails.

August 5. Newspoll has 52 to 48 (no link).

We can compare with Morgan's latest, which has 50 50.

Newspoll has primary support for Coalition at 44 and Labor on 36. Morgan gives 43.5 and 38.5.

So for Newspoll, an eight point gap on primaries shrinks to four after preferences. And Morgan's five point lead disappears to zero.

This is typical of differences between the two over many surveys. The major one is that Morgan gives a higher primary vote to the ALP than Newspoll does. A secondary one is that it distributes the preferences a little more favourably to the ALP. 

Morgan, remember, measures preferences while Newspoll calculates from last election.

August 2. Morgan has 50 50.

July 24. Additions to median income table. (Relates to this column.)

Now has seat swings in '98 and '01. At both elections, swings generally biggest in lower income seats. Best to start with Sydney and work outwards.

July 22. Newspoll in The Oz has 53 to 47. 

Beat through the template of "Howard the wiliest politician to ever inhabit the earth", and comparing apples here with oranges overseas, it sees our PM as weathering the "uranium from Niger" storm far better than his US and UK counterparts.

July 19 In Australian Financial Review, refuting "Howard's Battlers" furphy with combo of APH and AEC data. 

Associated data is in three tables: a national one ranked on median household income, that same table excluding rural seats, and one for the 27 Sydney seats. 

Morgan's latest: 50.5 to 49.5