The last week gave us a big bag full of polls containing a substantial amount of variation in the results – so today we’ll update our Pollytrend estimates to try and look through that noise to pick up any underlying movements in the true value of public opinion, and also giving us an estimate of the state of play at the beginning of the campaign.
Starting off with the primary votes – this is what the raw major party primaries look like (click to expand):
As we can see, there’s been a substantial overlap of the party results during the last few months, where Labor’s clear lead has pretty much evaporated. Running our Pollytrend estimates through this (our locally weighted polynomial regression trend line using a pooled all-pollster series weighted by both time and sample size as our underlying data) we get:
What looked initially like a big noisy mess actually contains some quite robust medium term trends in action, with Labor continuing to increase their primary vote while the Coalition is experiencing a slight contraction. Doing the same for the Greens, but just covering the 2010 period we get:
The Greens are the hardest to model properly, because they experience some quite volatile short term fluctuations fairly regularly – such as what we saw over the period two weeks either side of the Gillard ascension – but we’ll struggle on and do our best. Thankfully, at the moment, the Greens vote has clustered fairly tightly around the 12-13 point mark, making our trend estimates far easier.
If we look at the current values of these trend estimates for the primary vote and compare how they’ve changed since Gillard’s first polling cycle as PM at the end of June, this is how they pan out:
What we’ve seen over the last 20 days is the ALP vote continuing to increase from the trough Rudd had it in before he was replaced. We also see the Greens getting a few points of growth and the Coalition slightly edging back. The big losers have been the broad “Others” with their vote nearly halving in the lead up to the election campaign and over its first few days.
Moving along to the two party preferred, looking at the raw data we get:
This time, the trends in the data are much clearer to the naked eye. Running our Pollytrend line through it, we get:
If we do the same for our two party preferred trend estimates as we did for the primary vote trend estimates earlier and look at how the value has changed since the end of June, we get:
Over the last 20 days, the ALP two party preferred has increased by 1.5 points on the phone pollster trend and 1.8 points on the all pollster trend. Regardless of which measure you use, the trend is positive for the ALP by over a point – so the incumbent government, at this stage, is slowly pulling away from the opposition.
If we plug those phone poll trend estimate numbers into Antony’s spiffy election calculator, we end up with Labor on 83 seats, the Coalition on 64 seats and 3 independents. You might notice that the current estimate is exactly what the current Parliament is – but because of the electoral boundary redistribution, that actually makes the ALP in a position of being 5 or 6 seats behind their 2007 performance (since they notionally gained 5 or 6 seats in that redistribution).
So they ALP is currently ahead of the Coalition – but are behind their 2007 election result both in terms of their current two party preferred result and their net seat position. However, their vote is trending up and has been for around 50 days now, starting at the back end of the Rudd leadership and continuing through the Gillard leadership to today.
In other news, the Geelong Advertiser ran a cute little phone poll of the seat of Corangamite – currently held by Labor’s Darren Cheeseman with a margin of 0.9%. Just keep in mind that a sample of 227 has a margin of error of 6.5% and probably wasn’t demographically weighted to reflect the local population. So it’s probably completely worthless, but some of the non voting details are still marginally interesting.
28 Comments
An upward trend in the Labour primary through July to October/November 2009 – it all looked rosy then. Who could have forseen what happened after that?
The steady drop in the ALP primary from November 2009 to March/April 2010 and then a big drop after that.
With hindsight, it is easy to say Rudd should have done this or that back in November 2009 – but of course these things are much clearer after the event.
Thanks Poss. The trend looks good for the ALP at the moment, and that trend has been on even though there has been much “discussion” about boats and their unfortunate passengers of late.
Makes me wonder what will happen to that trend line once real issues like the Economy,Jobs, Health and Education start getting more of a run in the campaign?
Hmm, I didn’t realise Julia Gillard was running in Corangamite… (“Julia Gillard ahead in marginal Corangamite”)
Interesting wording that Possum..
If you back up the Pollytrend to the week before Rudd’s downfall, there was a uptick in the PollyTrend and I think you pointed it out too.
Personally I think the shape of the trend has everything to do with Abbott having exhausted the tactic of attacking without any clear or sane policy. Julia might have added a point or so, but we can never run the experiment again.
One other observation is that the PollyTrend has a limited slope and that it takes months for an upward trend to dissolve. So in essence, its unlikely that it will end up lower than it is now on Aug 21. One niggling doubt I have is this – are those parameters in part an artifact of the time weighting and how much real week on week volatility is there that can’t be seen because there isn’t enough data? (Real as in would have an effect if you could move the election a few days either way.)
Btw, I don’t think Rudd could have done much come last November. Where he needed to act, to lead public opinion, was back in 2008 and early 2009 where there was time to head off the climate denialists for one thing. What I am curious about is the xmas/new year period where he went absent and Abbott got the coverage. Some have said that was a deliberate tactic to allow Abbott enough rope. I personally think Rudd was just plain burnt out, but now that Abbott really is starting to look dodgy you’ve gotta wonder whether the tactic (if it did exist) has actually worked.
p.s. I still want to see the Possum Comitatus PollyTrend PollerCoaster DeathRide!
Any state by state breakdowns?
Also, I know I ask this all the time but any chance of a post on the senate?
The best post of yours I have ever read was on the greens in the senate and it seems that the most important factor in this election with less and less splitting the two major parties.
Sure, you might have to make some quite big assumptions to get it to run (preference flows, translation of lower house votes into upper house votes) – but why not give it a try prospecively and see how it goes as a predictor?
I got a bit nerdy with a prediction.
See spreadsheet with charts at bottom:
https://spreadsheets.google.com/ccc?key=0AvpufhYn0jaXdG13OUdsWDlnLTNtQzZ5N1VaUTRMMHc&hl=en
This is based on:
1. The greens returning all their current seats and gaining 1 in QLD, Victoria and South Australia.
2. Steve Fielding’s seat will go to the LNP.
3. Seats to Greens come off party that was lowest in last election (via Anthony Green)
Of course the above could be wrong but it’s a good baseline, right?
From this the senate makeup would give the Greens the outright balance of power…
Comments?
Fielding’s seat will go the ALP or the Greens.
At present the probability of an ALP victory seems quite high. I am curious as to how high the probability of an outright coalition victory would compare with the probability of a hung parliament.
cud chewer said:
He might simply have announced in August 2009 that he was presenting legilation based on Garnaut in fulfilment of his promise and challenged the Libs to put up or shut up. At that stage, Abbott was saying the legislation should be passed to get it out of the way.
If the Libs had refused, Rudd says — OK … the Coalition says we don’t have a mandate to act. This is critical to sort out right here right now. Also, it would be best if the government could ensure that it was able to follow a coherent program to meet the challenges of the GFC and plainly, the Coalition wants a different course. Business must have certainty and so we will be going to an election to allow the people top have their say.
No boat people, no insulation scheme, no mining tax and no BER rorts for the Coalition to beat up.
Can there be any doubt that he would have won a smashing victory over the disunited rabble he faced? He would be still PM today and 2 years out from the next election. He’d have gone to Copenhagen with legislation in his back pocket. Abbott would have been nowhere.
He played the game badly.
Rob Hoffman is quite right as the 2004 split was 3 Coalition – 2 ALP – 1 FF.
I have a feeling unless it is a downright rout that the Greens may be squeezed yet again in Victoria (and possibly NSW too) unless either the ALP get a lot more than 3 quotas and can pass on preferences to the Greens OR the Greens get a quota (14.4%) in their own right in which case it would be hard for the ALP to get the 3rd seat (unless the Coalition were absolutely routed).
I have never contributed to Possums blog before… I do like my new avatar!!
Whoa!!! Check that Pollytics Green vote table. 18-34s from 9 to 19! Bob will have to get a bigger hat for sure. I hope from this we are seeing a recognition that Labor & LNP do not have enough respect for the electorate, so the electorate disconnects?
I think Rob is right too…
And it actually what I put in. Had the greens picking up a seat in Vic and Stevie the way of the Dodo. Obviously not sure if it’s right… just a starting point.
https://spreadsheets.google.com/ccc?key=0AvpufhYn0jaXdG13OUdsWDlnLTNtQzZ5N1VaUTRMMHc&hl=en
TRI$TAN: Perhaps it would be easier to go along to Fielding’s cult church (is it Hillsong? Some crap religion) and check to see if the attendances have soared, remained static, or fallen. While there you can see who his preferences are going to.
No?
Venise
From memory [cos I checked once] Family First usually [ in most if not all states] put Fred Nile, DLP and Christian Fringe Group [whatever] in their first few preferences with the Libs sometimes getting a look in there somewhere in there and the ALP near last and the Greens dead last.
In return Fred and the DLP usually put FF in their top preferred few and return the compliment to each other.
A cosy little ‘family’ group’ looking after each other.
Possum, you say Labor is BEHIND their 2PP from the election but they are at 53.6 now compared to 52.7 then
Poss,
In the table at the top you have Greens at 12.5% and “Others” at 5.4% and their combined second prefs added to the ALP primary of 42.1% to get a 2PP of 53.1%.
What values are you assigning to the rate of second prefs flowing to the ALP from each of these 2 groups?
If you would please.
Oops.
Actually 53.6% 2PP.
Either way I’m wondering what the estimated flow of second prefs is from the Greens and the ‘others’.
Andrew and Fredex – I’m using the phone pollster trend estimate, rather than the all pollster one.
Fred, I didnt need to get into prefs, I’m using the pollsters own allocations for their TPP which is a mix of respondent allocated (e.g. Nielsen) and allocated on the basis of the 2007 election flows (e.g. Newspoll).
So it looks like Labor will basically get the same 2PP results as last time. But that a lot of votes will come via Green preferences.
Of course on the day there is no guarantee that is what will happen. Pollsters always seem to have the Green primary vote higher than election outcomes.
The only other possible hiccup is if lots of the Green primary votes are Turnbull supporters who preference Liberal second. Could potentially result in a more even flow of greens preferences than previously seen?
I don’t know how much to believe in these polls but it seems like the Labour messages are not getting across.
I just read an article on how Labour wants to increase the age of compulsory pay out to workers to 75 and Abbott want to scrap it on Yahoo; and there are so much nasty antagonistic rant against Labour as wanting to force people to wait until they reach 75 to be able to get their super when the policy is not that but requiring employers to pay super to employees over 70 who do not get super at the moment.
And despite the best effort of some commentariat to explain what the policy is about the rants against Labour’s policy still flow on the same, it seems like the pro-Libs do not bother to know what the truth is.
Long way to go to election day, but Abbott and co are looking more like a desperate rabble than an alternative government. Gillard and co are cool, calm, collected, disciplined and on message.
My tip is Labor to increase their majority.
CC comes in ninth in order of importance according to this from Essential
http://blogs.crikey.com.au/pollytics/files/2010/07/issueimportanejuly19.PNG