|
| |
May
29 Roy Morgan Research reckons the Mind and Mood of Australia is just a
little unplugged.
3
May 27 An Australian Republic? 3
May 26 Saturday's Lies
and Statistics column on last week's Tuesday polls. 4
May 23 Latest Morgan poll puts
it at 50 50.
4
May 20 Blast from the past. Newspoll leadership survey, Jan '94: 4
May 18 Big poll next week 4
May 17 Hogg Pt II 5
May 15 Bob Hogg Pt I 5
May 9 Latest Morgan has Coalition
on 44 and Labor 38 percent; 50-50 two party preferred. 6
May 6 Another Newspoll another
Bring Back Beazley story.
6
May 3 Whopping Newspoll graph 6
May 2 A PM from Queensland? 7
April 28. Has the world gone
mad? 8
April 25. Latest Morgan poll
has Coalition ahead 51 to 49. Morgan may be in the doghouse after last election
but they do ask respondents for full preferences, unlike Newspoll which extrapolates
from primary votes - and does so a little strangely. 8
April 15. Newspoll has 55 to
45 to the government. (My calculations, based on their primary figures, are
54 to 46, but still, it's a move towards the government). 8
April 14 (i) Pedant's corner.
Glenn Milne in the Oz reckons that 8
April 13. Those crazy guys at
Morgan have it 54 to 46 - in Labor's favour! (Measured, they're quick to point
out, before the fall of Baghdad.) 9
April 11. A new game! 9
March
30 Poll Shock Morgan and Newspoll agree! 11
March 12 Latest NSW Newspoll 12
March 11 Newspoll preferences
- the quibbling continues
12
March 10 Latest NSW opinion poll 13
March 7 NSW election coverage
updated 13
March 4: The saga continues 13
Hopeful souls have suggested the Peter Hollingworth farrago should boost
the republican cause.
They're dreamin. People should realise it ain't gonna happen. Ever. Not
by referendum, anyway, which means we'll only accept it when we have no choice,
when the British Royal family no longer exists. Even then ... what if we refuse
to recognise it by referendum?
It's not widely appreciated how out of the box our referendum process is.
It no doubt seemed a good idea at the time, but it has proved an irredeemable
dud. We're lucky we've never had to decide anything really crucial, and that
the High Court has, undemocratically, taken up the slack.
Only one other country has a constitution that must be changed by referendum,
and that's Switzerland. (Most nations amend their constitution in parliament,
say by two thirds or three quarters majority of elected representatives. You
know, those people we choose and pay good money to make important decisions
for us.)
First thing to note is that Switzerland has 26 cantons (or states). A majority
out of 26 is an easier proposition than one of 6, which is two thirds.
But, more importantly, the Swiss are better equipped than we are for the
process. That might sound un-strayn but it's true. They behave like
grownups, we are children.
Australian voters are highly partisan and our politicians probably the most
combative in the world. We ain't much interested in issues. It makes for fun
election nights, but ill-equips us to take important decisions.
Imagine (just off the top of my head)
a question ten years ago proposing that Australia connects to the world wide
web.
A few articulate demagogues, as always, would jump on the bandwagon.
Don't forget, the 'no' and 'yes' cases receive equal funding and airplay.
As the republic vote showed, an argument can be mounted against the specifics.
It can all start to look scary. If it starts to look iffy, expect at least
one political party to take the easy route.
Lost in every state, would be my prediction. See also this and this.
May 20
Blast from the past. Newspoll
leadership survey, Jan '94:
Which one of the following do you think is the best person to lead the Liberal
Party?
Answer: Bronwyn Bishop 48; John Hewson 34.
Same question in October '94 with Alex Downer
innovatively heading her maj's opposition: John Howard 20; Downer 18; Nick
Greiner 14; Bishop 13; Peter Costello 10; Hewson 9
Today's Newspoll gives no two party
preferred change. A notional distribution of their primary numbers is still
more likely to give 53 to 47 than 54 to 46, but it makes little difference in this case.
(Read more on preference
in discussion of a previous Newspoll, here.)
(I've deleted a portion
of the May 18 entry. It's my site and I'm only human. Am giving the predicting
game away.)
On the other hand, ACNielson, in the SMH and
Age,
gives Labor a bounce - to 51 to 49 in the government's favour. Headlines and tones couldn't
be more different from The Oz's.
Rightly or wrongly, Newspoll gets almost exclusive attention these days
- after Morgan's own goal at the last election. (Neither can really boast:
Morgan was out by 5.5 percent, erring in Labor's favour, Newspoll gave 2 points
too many to the Coalition).
Has there ever been a single survey more anticipated by the Labor leadership
than the one being conducted right now, this weekend? The budget reply was
seen as Crean's last chance, and next Tuesday the Oz will deliver a
verdict. That's modern politics for you.
They should forget the satisfaction and preferred PM ratings and look at
voting intentions.
Further to Hoggy, below. He followed up the column with a post budget reply
appearance on Lateline, where he twice mentioned the Victorian ALP's change
from John Brumby to Steve Bracks as an example of one from a decent, hard-working
guy who just wasn't getting through to voters, to someone who had the goods.
But Hogg wrote in The Age on 26 August 1999 that it "stretches
credibility too far for Bracks to look voters in the eye and say we can/will
win this election".
It wasn't obvious that Bracks - who was getting trounced in the polls on
every measure you can think of - had the goods until polling day the following
month. Then, of course, it always was.
This week's award for Least Meaningful Comment From Someone You'd Expect
to Say Meaningful Things goes to former Labor national secretary Bob Hogg
in today's Financial Review (subscription, no link): "The ALP's
history shows that it wins government only when it pushes forward".
(Hogg's advocating a generational leadership change.)
Well, yes. It wins when it pushes forward ... in votes in the right marginal
seats to form government. Did it push forward in replacing Hayden with Hawke
in 1983? Gough Whitlam pushed forward in '69 and '72. Before him was Scullin
in 1929. Not sure if he pushed forward.
Did they push forward with Evatt or Calwell? Obviously not, because they
didn't win.
Had Beazley won in 1998, as he nearly did, then we could look back and wisely
pronounce the party to have "pushed forward". Same with 2001.
For anyone old enough to remember the 1993 election campaign (which Hogg
ran for Labor), the consensus until polling day was that the Liberals had
run a superior campaign - finally, they'd learnt how to do it - successfully
creating the momentum for change. The day after the surprise result everyone
agreed that the Liberal campaign had always been pretty dreadful.
Hindsight is always selective.
May 9
Latest
Morgan has Coalition on 44 and Labor 38
percent; 50-50
two party preferred.
Newspoll's most
recent was 46 to 36,
which they say is 54 to 46 two party preferred but really is more like 53 to 47 (see reasoning,
in relation to the previous Newspoll, here), and
ACNielson
has primary support 47 to 35, also 53 to 47 two party preferred.
So Morgan is still out of step with the other two.
May 8 Don't forget this about preferred PM ratings.
The two best performing opposition leaders on this measure, since Newspoll
started tracking it in 1987, were John Hewson and Kim Beazley.
If I can be pedantic, those two party preferred numbers should be 53 to 47, not 54 to 46. See my reasoning,
in relation to the previous Newspoll, here.
The Oz piece shows how silly "preferred Prime Minister"
can be. Of course Peter Costello gets a low rating, just as Keating did from
the Treasurership. Of course the incumbent comes out on top.
Still, Kim's much preferred to Crean. And Beazley doubtless would have equipped
himself better during the war. But there is no reason to think that he would
be delivering better voting intention numbers.
(ACNielson in the SMH has
similar story. They (presumably) do ask for preferences - and give 53 to 47.)
Voting intentions are what matters. Not unrelatedly, latest Fin Review Lies & Stats
is now up. Warning: it's number-heavy.
Beady eyed AFR readers may scratch heads at my quoting the latest
Newspoll result as 54 to 46 instead of 55 to 45 given in The Australian. I reckon my calculation
is more correct than Newspoll's.
It might seem presumptuous to second guess Newspoll - they conducted the
survey - but they don't measure preferences. They don't ask Democrat/Green/One
Nation/"other" voters who they will give preferences to; they just
ask which party will receive the respondent's primary vote.
Newspoll then distributes the non-major party votes as they flowed at the
last election to get a two party preferred figure. (The Oz only started
quoting two party preferred after I wrote this
piece in January; until then they just served up primary support.)
However
... Newspoll just distributes the non-major votes as a lump. For example,
the latest poll had 33 percent for Labor and
46 for government and 21 percent to minor parties and
others. On November 10 2001, 58.3 percent of the latter eventually flowed
to Labor, 41.7 to the Coalition, and distributing the 21 accordingly gives
you, after rounding, 55 to 45.
But since
the election the Green vote has almost doubled and One Nation's has collapsed.
The "lump" approach doesn't recognise this nor does it distinguish
between different types of non-major party votes - very misleading.
So a better approach is to take Newspoll's numbers - Coalition 46, Labor
33, Democrat 2, Greens 8 One Nation 1 and 10 for "others", and distribute
the last four as each flowed at the election, namely (to Coalition: Labor)
36:64, 25:75, 56:44, 51:49 respectively. If
you do that you get 54 to 46 two party preferred.
It makes little difference in this case, but when the result is close it
means a 50:50
is presented as 51:49
- this happened earlier this year - which is the same as the election result
and so can be reported as the government being ahead - as it was.
I also addressed this earlier here.
On the Labor leadership, at last count there were two, maybe three Queensland
hopefuls: Kevin Rudd, Wayne Swan and Craig Emerson (the maybe). You could mount an argument for a
Queensland Federal Labor leader as follows ...
Queensland is the third largest state, with 27 out of 150 seats. (NSW has
50, Victoria 37, WA 15, SA 12 and Tasmania 5 (two each for the territories).
However, it has a small state mentality; it is parochial. So a Queensland
Labor leader might reap vote dividends with little negative effects in other
states; people in NSW and Victoria are less state-centric.
Queensland has a bucket of vaguely winnable seats. The last time Labor got
a two party preferred majority there - in 1990, a bare one of 50.2 percent
- it bagged 15 out of 24 seats. They currently have seven (with 45.3 percent
2pp), and are nine or ten seats away from victory.
The last time Labor got a two party Queensland majority was in 1961, when
Menzies won by one seat. Queenslanders don't like federal Labor.
A month of good polling and Glenn
Milne's talking about "entrenching the Coalition in power for at
least a decade". He's not alone, and not for the first time I'm reminded
of Paul Keating's last term. Incidentally, half way through that he was on
57 to 43 in Newspoll, which dwarfs Howard's current 54 to 46.
Morgan may be in the doghouse
after last election but they do ask respondents for full preferences, unlike
Newspoll which extrapolates from primary votes - and does so a
little strangely.
This graph, of Newspoll results since
1984, shows that the main feature of the current term is the closeness of
the contest.
April
15. Newspoll
has 55 to 45 to the government.
(My calculations, based on their primary figures, are 54 to 46, but still,
it's a move towards the government).
After such a shocking start I might give this predicting game away. There's
a lesson there.
"according to the latest Newspoll .. [an election would see] the
loss of 18 Labor seats", including "[Wayne] Swan, who sits on a
4.9 per cent margin"
Glenn's over-reaching. The last Newspoll showed the equal biggest lead for
the government since the last election, 54
to 46. That's a swing of three percent, which from the pendulum in uniform terms would be a gain
of 11 seats for the government, not including Swan's seat of Lilley. Cold
comfort for Labor, but it's nice to get these things right.
Last Monday, the Oz published a whopping quarterly
analysis of Newspoll data with tables that showed a bad result for the
ALP in marginal seats in the first quarter of this year - Jan - March.
Today there's a repeat
performance, though now for all seats across the country.
The theme is the same: Crean's doing woefully and Howard holds all trumps.
They don't notionally distribute; my calculations put it at 51 to 49 for the first
quarter of the year, the same as at the last federal election.
April
13. Those crazy guys at Morgan
have it 54 to 46
- in Labor's favour! (Measured, they're quick to point out, before the fall
of Baghdad.)
Incidentally, by now (Sunday 6:30pm) the weekend's Newspolling should be
over. The numbers are locked in. (see April 11, below.) I've stepped outside
and sniffed the breeze, and I'm now going for a Newspoll result (assuming
one will be published in the Oz on Tuesday) of, um, 52
to 48 to the gummint.
Mr Shanahan will write something like this (once he's finished with the
next to useless and very boring preferred PM and satisfaction ratings): "Support
for the Coalition dropped by two points, although the government would have
won an election held on the weekend with an increased majority ..."
Incidentally, apropos of little, a
trip down memory lane. A month out from the 2001 federal election,
Newspoll had the Coalition ahead 56.5 to 43.5. The final Newspoll, taken on the Thursday before
election, had 53 to 47.
Election result was 51 to 49.
And: I've said it before and I'll
say it again: commentators think that Simon Crean's ordinary presentation
skills (let's face it, he's not crash hot at the moment) and even less inherent
oomph make him unelectable.
Au contraire. The flipside is that he isn't scary or particularly
loathed. Remember opposition leader John Howard. And see this AFR piece on preferred
PM ratings.
Each fortnight I'm going to play a new game. It's silly, pointless and foolhardy,
and it's called "guess the Newspoll result".
It occurs to me that writing things like "this [jump in government's
lead] is to be expected in a time of war" sounds glib and smartypants
after the fact. So I'm going to try my hand at predicting.
This, the first attempt, is written on Friday April 11. Anticipating that
images of celebrating Iraqis will continue to trump those of looters and perhaps
suicide bombers (whose opportunities are now rife) over the weekend, but guessing
community relief that it's almost over may work against the government - when
Newspoll is phoning their one thousand and something people - throw in a few
magic ingredients, and I reckon .. (drumroll) ... that next Tuesday The
Australian will report (more drumroll) ... something like ... 53
to 47 two party preferred in the government's
favour, a one point shift to Labor from the last Newspoll.
Which might be counter-intuitive, given unanimity among pundits that the
fall of Baghdad has given Howard yet another "resounding win". The
unanimity encourages me to go the other way.
Guessing the outcome of a measurement that in itself is imprecise might
be of limited use, but I for one will be interested to see how I go in the
longer term.
(Here's the possible primary vote breakdown: Coalition: 46, Labor: 36, Greens:
9, Democrats: 1, One Nation: 0 and "other": 8)
May revisit this one after the weekend (but before Tuesday's Australian).
April 7 The
Australian has some federal Newspoll tables. Not new data, but three
monthly consolidations in marginal and non-marginal seats that give a mixed
and unexciting bag. However, no self-respecting journalist would write such
a piece today that doesn't conclude that John Howard conquers all and Simon
Crean can't take a trick, so that's the theme.
Actually, the numbers, unless "preferred PM" and "satisfaction/dissatisfaction"
ratings do it for you, say little other than that the ALP is about where it
was at the last election, or a little behind. Strangely, the Oz has
reverted to not bothering with two party preferred - presumably because it
wouldn't fit the story.
Anyway, in these marginal seats, at which two party preferred votes were
50:50 at the November
2001 election, my calculations have the Coalition leading 51 to 49 in Jan-Mar 2003.
Same numbers for the last quarter last year, but down from Labor's 51 to 49 lead for July
- Sept 2002. Yawn.
As I've said before, the main feature of opinion polls since the last election
is the absence of a big lead by anyone. Remember Beazley's 58 to 42 in early 2001?
I've been slack updating the Newspoll preference table lately. Will get
to it ... soon.
April 3 Morgan polls in Qld,
SA
and WA
have state Labor governments a mile ahead in the first two but 50-50 in Western
Australia.
April 1 Newspoll
in
The Oz has
government ahead 54 to 46,
the same gap as immediately after Bali bombings. Labor's primary vote is just
34 percent (ouch).
March
30 Poll Shock
Morgan and Newspoll agree!
Latest
Morgan poll, like this week's Newspoll, shows a swing to the federal government
with the first dropping of bombs. Morgan says 52.5 to 47.5.
Newspoll in The Oz on Tuesday had government ahead 53 to 47 (no link).
Like post-Bali jump, to be expected during war.
See also Morgan
poll in US, UK, Australia and New Zealand with question: "Should UN have sanctioned US military involvement in Iraq?"
"Yes", seems to be the answer, degrees of enthusiasm in the order
listed above; a Newspoll in the Weekend Australian showed that same
relative enthusiasm on support for the war.
Pundits have to ponder on something, and rivers of ink have been devoted
to the domestic political ramifications of this war.
They really shouldn't bother. The last gulf war, much more popular than
this one, didn't stop US President Bush losing the next election nor Bob Hawke
the Labor leadership.
One might hazard a guess that, in several months when the war has disappeared
from our screens, irrespective of how bloody it was, most Australians will
retrospectively rationalise our involvement and think it was a good idea.
They'll give John Howard points for "doing what's right", but domestic
politics will go back to its mundane domestic self.
On the other hand, in the unlikely
event that we're still fighting - either in Iraq or on our next excellent
adventure in Iran, Syria, North Korea, Libya or wherever - well, that's a
different story, and the word "quagmire" comes to mind.
March 18 Newspoll
again 51 to 49. No quibbles with preference distribution this
time.
Although, with numbers like that - too close to call given margin of error
in a survey like that - the tone of the article, represented in the headline
"Howard winning war of opinion" overdoes it just a little.
But we're used to that.
March 15 If you're here via the AFR
column today: Here are graphs of all
states' absolute and relative votes.
March 13
Another anniversary. Ten years since the sweetest you know what of all
The quibbler is me. Regular visitors to this site will know that I go on
and on and on about preferences. The reason I go on about preferences is that not enough other people
do.
In February I had this piece in the
Australian Financial Review. If you want to understand the rant
I'm embarking on here, you'll need to read that first. (Less than 500 words).
I then set up this table and
declared that as The Australian wasn't going to distribute their preferences,
I would - each fortnight, as they were published. Lo, the very next Oz
Newspoll of voting intentions indeed had a notional two party preferred vote
based on preference distribution at the November 2001 election (51 to 49 in the government's
favour). All was well.
However, the
Oz report after that again had two party preferred votes but they appeared
to be wrong. My calculation, based on the primary votes, was 50 50, the Oz again had
51 to 49.
Now I know what you're thinking;
you're thinking: so what?, and you're right. What's a percentage point in
a survey like this? The answer is: not much, but there's a widespread belief
that the Howard government is "way ahead in the polls" which is
starkly at odds with reality. Actually seeing, in black and white, the numbers
50 50 in a Newspoll report might give everyone a reality
check.
Anyway, it turned out that rounding was partly responsible; Labor's published
primary vote was rounded up, the Coalition's was rounded down, and I was calculating
from the rounded figures. But rounding wasn't the only reason for the discrepancy.
We're half way through today's
rant! If you made it this far it's all down hill from here
It seems that Newspoll's notional preference distribution takes all intended
primary votes for minor parties and independents and distributes them as they
flowed in aggregate at the last election. That is, 58% of all votes
for someone other than a major party candidate eventually flowed to the ALP,
while 42% flowed to the Coalition, so they distribute the whole lot in those
proportions.
I had wrongly assumed they do as I do, which is distribute them as per each
minor party's flow. The Greens flowed 75% to Labor and 25% to the Coalition,
One Nation favoured the Coalition over Labor 56 to 44 and so on. (It's all
here.)
When you consider that since the last election the One Nation vote has gone
from 4.3% to something that usually rounds down to zero but sometimes rounds
to 1, and the Greens have gone from 5 percent to usually sevens or eights,
then treating them as one great big lump is highly misleading.
Look, in the end it will make a difference of one percentage point at most.
But it's all in the presentation. A poll result of 51
to 49 given inherent error, is too close to
call. But 51 to 49
was the result at last federal election result, which in the fevered imagination
of some constitutes a landslide (see, instead, the Victorian result of last
year 58 to 42
- now that's a landslide).
Our major daily broadsheet quoting the latest Newspoll showing 50 50 two party preferred
might drop a great big penny that - oh my God, it's too close
to call!
So I'll be keeping the table
going.
End of rave. I think there's an election campaign in NSW
Newspoll preferences:
Just when you thought it was safe ...
Tuesday
afternoon. Update. Apparently something along the lines of the scenario suggested
here actually occurred. It's all
in the rounding. So apologies to Newspoll and Mr Shanahan. (Read on anyway.)
Background
The story so far. Last month I unilaterally appointed myself preference
distributor of fortnightly federal Newspolls. The Oz, who publishes
them, had been reporting only primary support levels which don't show a realistic
electoral picture. This AFR piece
explained why.
I made this table of Newspoll results
since the last election with calculated two party preferred ones (based on
preference flows back then, which is how Newspoll advocates doing it) and
promised to update it with each published survey.
And then ... the next Oz Newspoll (published
February 18) actually gave notional two party results (51 to 49 in the government's
favour). So I was out of business, but the world was a better place.
Today
However ... today (March 4) the latest one
is out. It again has two party preferred figures (which is good), but ... they appear to be wrong
(not good).
Dennis Shanahan reckons it's still 51
to 49 in the Coalition's favour ("the
Coalition would still win an election held now"). My calculations (which
should be the same) give it to Labor 50.1 to
49.9.
This of course rounds to fifty fifty. The difference between this and 51 to 49 is not great
in a survey like this, but in a presentational sense it matters. The reporting
of one set of figures leads readers to believe that the vote situation hasn't
changed since the November 2001 election; the other indicates it's too close
to call.
The latter conclusion is the correct one.
As I never tire of repeating, I'm not here to burnish poor old Simon Crean's
leadership credentials, I just desire a world in which people have a true
understanding of our major parties' electoral chances.
The latest. Apparently it was all in the rounding, something like this really did happen.
Shaun
Carney in The Age worth a read apropos of stuff below.
In defence of Simon: watch what they do, not
what they say
The knives are out yet again after yesterday's Newspoll. As usual, Alan Ramsey
gives the anti-Crean case.
Says Alan: "Based on yesterday's Newspoll in The Australian,
more voters (53 per cent) are now dissatisfied with Crean's leadership than
for any opposition leader since the Liberal Party's Alexander Downer (58 per
cent) the weekend he resigned in January 1995."
And he's just warming up; it's not pretty.
But ... a big but ... if you're old fashioned enough to believe that what
really counts are voting intentions then there is one huge difference.
That final Downer poll showed the Keating government ahead by 9 Newspoll
points after preferences. That's called unelectability.
Yesterday's had the parties still stuck on 51
to 49 in the government's favour. That's too close to call.
Voting intentions indicate electability. Votes get you elected. The rest
is mostly puffery.
Put crudely, of course Green voters can't stand Crean. But come the next
election, they'll give him their preferences ahead of Howard.
And on the much hyped preferred PM numbers, see this from early
January.
As I've said before, it's not my role in life to talk up Crean's chances.
Just don't put your house on a fourth Howard victory quite yet.
A week and a half ago in the Fin Review
I complained that The Australian often misleads its readers on the
federal electoral situation by reporting the primary, but never two
party preferred (except during election campaigns), Newspoll voting intentions.
See also here.
Well, the very next poll they make a
liar of me, giving two party preferred figures, for the first time outside
an election campaign that I'm aware of.
We warned the Czar and the Czar took heed!
Anyway, I'll keep the table going,
although my new self-appointed role as Newspoll preference distributor has
been redunded .... for now.
NSW pendulum has more bells and whistles.
NSW pendulum is getting more detail.
A wee clarification for today's AFR Lies & Stats piece: Hewson
more often than not recorded higher preferred PM ratings to Keating in his
time before the March '93 election, but not for the year and a bit
that he remained opposition leader after.
NSW election is moving along. Initial pendulum is up. Have a look here.
I've got a piece in the Canberra
Times today anticipating John Howard's retirement later this year. In
it I touch on the record of electability and "preferred PM" rating.
I've been through the Newspoll archive and it's pretty interesting; if anything
there may be a negative correlation between scoring well in this from
opposition and winning elections.
The two recent opposition leaders to perform best as "preferred PM"
were Hewson and Beazley! Say no more.
I'll be expanding on this in a Fin Review Lies & Stats piece
this coming Saturday.
It looks like I'm getting around to reorganising this site. This is an initial
attempt; it's still rather messy but I shall be working on it. Please stay
tuned. I'm constructing pendulums for the NSW election as we speak
| |
|