The Courier-Mail’s second Galaxy poll of the campaign shows the LNP opening up a 51-49 lead, following last week’s dead heat. The LNP leads Labor on the primary vote 44 per cent to 41 per cent (compared with 43 per cent to 42 per cent last week), with the Greens steady on 8 per cent. The Courier-Mail promises tomorrow will bring “a raft of other surprise results, including voter views on the important issues and their opinions of the leaders”.
UPDATE: (Possum)
Galaxy polls usually have a sample size of 800 which gives an MoE of around 3.5% – so these movements are all within the sampling error. In fact, it would take a sample size of over 9500 for these movements to be significant.
UPDATE 2: (William)
A further smattering of info from the Courier-Mail:
But the Galaxy Poll of 800 voters on Tuesday and Wednesday contained positives for Ms Bligh and some worrying trends for Mr Springborg. Ms Bligh holds a commanding lead as preferred premier at 48 per cent to 37 per cent, however she has dropped two percentage points during the campaign … According to voters, Ms Bligh is also well ahead of Mr Springborg when it comes to listening, trustworthiness and in touch with their concerns. Most also prefer Ms Bligh’s vision for the state.
74 Comments
Pages: « 1 [2] Show All
What? The LNP’s mental health policy requires volunteers to act as some of the staff?
http://www.theaustralian.news.com.au/story/0,25197,25142939-601,00.html
Isn’t mental health a more serious issue than that, that actually requires tertiary qualified professionals to provide help?
Ryan, what happens once the Nationals have to come up with their own ideas?
http://www.theaustralian.news.com.au/story/0,25197,25144149-25090,00.html
ShowsOn, it’s the Nationals version of workchoices, choose to work for no pay.
These Labor seats will turn BLUE:
Chatsworth ALP 0.1%
Clayfield (*) ALP 0.2%
Whitsunday ALP 0.4%
Burdekin (*) ALP 0.9%
Mirani (*) ALP 1.2%
Cleveland ALP 1.3%
Hervey Bay ALP 2.1%
Mudgeeraba ALP 2.7%
Aspley ALP 3.1%
Gaven ALP 3.2%
Barron River ALP 4.8%
Pumicestone ALP 5.4%
Springwood ALP 5.7%
Redcliffe ALP 6.0%
Broadwater ALP 6.6%
Redlands ALP 6.8%
Toowoomba North ALP 7.6%
Cairns ALP 8.0%
Southport ALP 8.0%
Keppel ALP 8.1%
Coomera ALP 8.3%
Mansfield ALP 8.4%
Ashgrove ALP 8.5%
Mount Isa ALP 8.5%
Burleigh ALP 8.8%
Townsville ALP 9.4%
Mulgrave ALP 9.8%
Mount Coot-tha ALP 10.4%
The crux of the problem for Labor is summed up well here:
http://www.theaustralian.news.com.au/story/0,25197,25145646-5006786,00.html
The line that this poll is within the MoE of the last poll and that therefore “nothing has changed” surely misses the point. Why would anything have changed?
What this show poll is that the previous poll was no rogue. If anything, it was too favourable to Labor. This poll confirms that this is a lineball election.
“show poll” = poll shows
LNP will romp in the regions and have enough in Brisbane to form government. This will happen.
A further smattering of info from the Courier-Mail:
Well, I was wrong about the swing to Labor – I don’t know enough about the electorates to say who will vote what where (that’s why I read here) but it looks interesting.
Don’t know if LNP will win Ryan, but maybe a decent change of seats. We’ll see soon enough…
David @ 56 the poll does not show it will be a line ball election – as Antony’s calculator shows the LNP need more than 52% to win.
They may well get that but the poll does indicate that they will.
They might need more than 52% to win. They can win with 51% depending on where the swing is, though it is less likely than if they get 52%, or 53%. These numbers are nice approximations of what is needed, but there’s no reason to be dogmatic about them. If Galaxy is right and the LNP 2PP is 51%, then the result will be interesting. If the result in LNP 51%, then we won’t be calling a winner till after 9pm. In the 2001 election, I was able to call Labor back at 6:40. I’m not sayin 51% is a line ball election, but it certainly means you’ll have to watch where the seats fall.
Antony & others, can I take you back to our discussion of the 2PP vote at 29, 31 and 32? Winston & Antony both said it had been found that allocating 2nd prefs on the basis of previous elections had been found more reliable than relying on the pollees’ stated intentions. Well, you should know so if there is experimental evidence of that I’ll believe it – I’m a great believer in experimental evidence.
But then you both also said that at the last minute people tend to follow the HTV cards. Voters for major parties do that, but in fact voters for the minors (and those are the ones who matter when 2PPs are being counted) do not. Look at the 2pp figures for 2006 on the ECQ site. They’re hard to find, first cos there is no link to the 2006 results from the “archived results” list but with the help of google I found them at . Then the 2pp counts are on the results/booth pages not the results/district pages where you’d expect them. But they’re there. I looked at Sandgate – safe ALP seat where the Possum and I both live (not together) – and Clayfield for comparison.
In Sandgate they (oddly) distributed the Greens before FF though the Greens had a few more votes. The Greens went 28.5% to ALP 11% to Lib 13% FF and 47.5% exhausted. Then FF went 25% ALP, 37% Lib and 38% exhausted. In Clayfield the Greens went 47% ALP, 21% Libs and 32% exhausted. I don’t know what the Green or FF HTVs said (you’ll know, Antony, good old walking encyclopaedia) but clearly the majority was not following them.
The government and Parliament went out of their way to help the pebble-counting and pendulum-making business some years ago by including s 124 in the Qld Electoral Act to ensure that prefs are counted for information even when not necessary for the result. People like MM and AG are using the results for constructing pendulums (yes, English plural) but you don’t seem to have looked deeper into the results -otherwise you couldn’t say things like “the voters follow the HTVs”.
I think there’s a BA Hons or MA thesis in waiting for someone to study (i) how the distribution of Green prefs varies across the socioeconomic range and (ii) whether the distrib of minors’ prefs varies in sync with the swings in first prefs from ALP to Libs/LNP. More importantly for present purposes, these figures could (should) be considered by anyone trying to predict 2pp figures when all they’ve asked the voters for is their first pref. (Is the guy from Galaxy listening?)
OK, putting a URL in angle-brackets doesn’t work – html thinks it’s a tag! The links I was trying to refer to above, without bothering to make them proper links, are http://www.ecq.qld.gov.au/elections/state/state2006/results/summary.html http://www.ecq.qld.gov.au/elections/state/state2006/results/district72.html and http://www.ecq.qld.gov.au/elections/state/state2006/results/district21.html
…. and the booth details where the 2pp counts are hidden are http://www.ecq.qld.gov.au/elections/state/state2006/results/booth72.html (Sandgate) and http://www.ecq.qld.gov.au/elections/state/state2006/results/booth21.html (Clayfield).
I think you’re being rather too dogmatic over the concept of uniform swing. If one party wins more than half the vote, then there’s every chance they’ll win more than half the seats.
Jack a randa, plenty of work has been done on how people are influenced by how-to-votes, though there is plenty of further work that could be done using AEC data and tying it to how actives candidate are in distributing how-to-votes. People are clearly ‘influenced’ by how-to-votes, every bit of research points to that.
The ECQ has published ballot paper surveys at every election since 1992. Before you slag me off for not looking closely enough at the data with your one reference to Sandgate, I’d like to point I have looked at everyone of of those reports, which is something I presume you have not done. I have summary tables of preference estimates in every Queensland electorate since 1986 which I consult to produce preference estimates for election night. We use estimates until the point when the Electoral Commission releases actual preference counts. And yes, I do collect how-to-vote material which I store away and keep with the preference tables so that the tables can be interpreted.
I do think the ECQ ends up with an overestimate of the influence of HTVs because they always get a high rate of following ’1 only’ HTVs, but it is still the best data available on the subject because it actually looks at the ballot papers.
Another approach is to look at preference tally sheets, which are done on election night to produce indicative preference counts. The tally sheet method summarises preference destinations, where preferences ended up rather than the exact copying of a sequence. But these can be used to measure the effectveness of preference direction.
The AEC has published this data at every election since 1996, and I can assure you I have lots of table of preference flows for every party in every electorate at every election. The SA Parliament has done something similar using preference tally sheets for SA elections, and I’ve done the same research for the NSW and WA Parliamentary Libraries.
In the case of my NSW election research, it was the only work of its type done on optional preferential voting. You can view the summaries at http://www.parliament.nsw.gov.au/prod/parlment/publications.nsf/0/399440F051AE6096CA2573F00018CA2C/$File/2007ELection%20Final.pdf or the full detail of preferences booth by booth at http://www.parliament.nsw.gov.au/prod/parlment/publications.nsf/0/4ABFD34F024071FACA2573F0001EA4BF/$File/408878.pdf
I seriously think before you slag people off for not knowing what they talk about, you should do a little more research than looking at a few pages on the web.
Antony, I wasn’t slagging you off for “not knowing what you’re talking about” – I was pointing out that something you had said seemed a trifle too dogmatic. I guess that’s the occupational hazard of being an expert – you can tend to sum up a lot of knowledge in a sweeping generalisation.
Oh, and to get back to the original point – I still don’t see why I should have much confidence in pollsters’ predictions of the 2pp vote in a State with OPV. The first prefs, yes, but the 2pp – nuh-huh.
Yes, I know. We all tend to use words like ‘follow’ when ‘influence’ is the more correct phrase. And I agree you that we shouldn’t have complete confidence in state wide 2PP from a poll, but there is nothing a pollster can do about that.
I’ve just finished a publication for the WA Parliament and the final 2PP comes out at Labor 48.1%. The primary votes were all exactly the same as Newspoll, but the 2PP was 2% out, for a variety of reasons.
Even then, the 48.1% 2PP is on one possible method of counting the 2PP. I reckon it would be nearer 47.5% if more account is taken of the National Party’s position in the election.
I’m am, as usual, about to have an argument with my producers about whether we can use a 2PP swing graphic in Queensland. We couldn’t in WA because it was wrong. The more accurate you make your method of picking a winner in an individual seat, the less accurate you make your statewide 2PP. In trying to ensure we get correct 2-candidatye preferred results in Charters Towers, Gladstone, Maryborough, Nanango and Nicklin, I lose the ability to get an accurate state-wide 2PP.
And before someone asks why don’t we produce a 2PP estimate and add the numbers up, it doesn’t work like that on election night. We don’t add 2PP figures, as all 2PP figures are predicted results, and the statewide figure comes from weighting the electorate swings. And this gets stuffed up by non-2PP seats.
Up-to-the-minute bookmakers’ odds for the Queensland election
http://way2bet.com.au/odds_comparison/21362/21363/46155/Queensland+State+Election
Antony, seems to me you need to modify your computer program so that in seats like that it keeps 2 computed, or imputed, records – one of the 2-leading-candidates vote (once you’ve worked out who they are likely to be – not always easy I know), and another of the 2pp, as in 2 major parties preferred, vote. Then you’d use the 2-candidate-preferred for your beautiful rising-and-falling bar chart for the specific seat, but factor the 2-major-parties-preferred count for those seats into the statewide 2pp total.
Of course there would be definitional boundaries to be drawn for the 2pp for these tricky seats – clearly all votes for Stuart Copeland would be counted as non-Labor, but 2pp purists might suggest that some people who vote for Peter Wellington would vote ALP if he weren’t there, so his vote should be notionally split. But if you think of 2pp as Labor vs non-Labor rather than Labor vs Really Conservative then all major independents who have a chance of stealing a seat, whether conservative or vaguely-centrist, can be counted as non-Labor. (Certainly some of us would say that people who vote for independents who might win, as opposed to voting 1 Green 2 ALP just to ginger-up the ALP, aren’t real Labor voters or even Labor-preferrers!) I think that would produce a number with some meaning, and if ALP campaign managers complained that you were undercounting their 2pp vote, well, it would only be a significant undercount in 1 or 2 seats out of 89. Anyway, ask them to suggest anything better!
But, if you think this is a useable idea, how quickly can you modify the program? Can you dash off line after line of flawless C++, as I rather imagine you being able to do, or do you have to hire a contract programmer? If you can’t get it going for March 21, perhaps it’s worth considering for later elections? (But don’t ask me how to apply it to South Australian State elections – the 2pp concept really collapses there!)
I know all that Jack, but you don’t do it in a 17 year-old system. Not a hope of getting it done. It’s in for the re-write on a more modern platform. A statewide 2PP is of no use in picking the result anyway, so I’m happier to operate on getting each seat right. I’ve operated without it for 17 years and it only became a problem when someone suddenly decided they wanted a graphic with statewide 2PP and swing.
Fair enough Antony – and yes of course the statewide 2pp is no use in picking the result, and it loses its meaning when elections become multi-dimensional contests rather than the simple left-right contests that we (well, I, anyway) grew up with. But as long as there remains a substantial element of left vs right in politics, (i) pre-election, it’s a nice simple measure of the popularity of parties, and people like nice simple measures, and (ii) post-election, some people use it as a measure of the “fairness” of the result (though we know that an “unfair” result could arise from deliberate fixing or simply from the flukiness produced by single-member electorates). So I’m sure we’ll all be interested in what your new platform can produce, whenever it’s up and running.
Pages: « 1 [2] Show All