Oz Politics Blog

“Carbonsink” emailed me to say tomorrow’s headline ACNielsen figure was reported on the ABC News.

You may need to hit the refresh or reload button on your browser to see the latest graphs.

Today’s SMH and Age reported on an analysis of ACNielsen polling aggregated from November 2006 until May 2007.

The headline message: “Labor would be returned with 110 seats to the Coalition’s 38 — the most lopsided outcome since World War II.” “[T]he swing to Labor in Queensland alone could be enough to make Kevin Rudd the prime minister by Christmas.”

The state by state swings were as follows.

And there was a final note of caution.

ACNielsen research director John Stirton warns that poll findings and the election outcome are quite different things.

“Polls are not predictions of voting at the next election, which is still months away,” he says. “Polls measure current voting intentions, so changes can be tracked over time.”

Mr Stirton says the Coalition’s primary vote has risen by 2 percentage points in each of the past two polls. “It is possible that we will look back in several months and see this period as the start of the Government’s recovery,” he adds, predicting that the next month or two should make it clear whether any “bounce” back to the Government is under way.

At first blush, it is a bad poll for the government. Labor is unchanged on 58 per cent of the national two-party preferred vote share. Were an election held last Thursday to Saturday, the Coalition would have got 42 per cent.

However, the figures belie the Coalition’s primary vote which has improved for the second ACNielsen poll running. The Coalition’s primary vote prediction was 39 per cent (up two points). Labor’s primary vote prediction was 48 per cent (down two points).

Furthermore, the two-party preferred (TPP) prediction had the Coalition capturing only 23 per cent of the preference votes (down on the more typical 35 to 40 per cent). As a consequence, I suspect a more accurate TPP prediction would have been 56 to 44 per cent in Labor’s favour (using the preference flows from the last election).

In terms of the ACNielsen series, the 2007 election year is not out of kilter with 2001. A Coalition win from here remains plausible (but by no means guaranteed).

All of the polls suggest there is some movement back to the government since Rudd’s peak in March. What is more difficult to discern is whether the pace of that movement looks sufficient (should it be sustained) for Howard to win a fifth term. For me, the Coalition’s biggest concern is Labor’s extraordinarily high primary vote. That will take some beating, and my graphs suggest it is the most resilient against erosion.

I should also note that the previous three graphs are highly speculative. However, as we approach the election, they provide a way of testing the hypothesis that Howard will from come behind to win in 2007 as he had in previous years.

At this stage I would give Labor a 60 per cent chance and the Coalition a 40 per cent chance of winning in November-December 2007. The betting market has it closer to 50-50.

The usual opinion poll graphs are here. You may need to hit the refresh or reload button on your browser to see the latest graphs.

Update:

• Quiggan suggested there has been a sea change with the commentariat coming to the view Howard could well lose this time. He also made the following observation:
  

  Interestingly, the betting markets don’t seem to have moved too much away from even money, while the polls have been giving a consistent message all year. The election will be a big test for the relative predictive powers of polls, pundits and punters.

• Lalor asked his readers for their experiences with Brand X workplace reform — while nine out of ten responses are little more than partisan babble, there are some quite compelling stories of disadvantage, perhaps real vote changing stories

Galaxy and ACNielsen both have polls out today. Both have the same headline prediction: a national two-party preferred vote of 42 per cent for the Coalition and 58 per cent for Labor, were an election held last weekend.

In terms of the primary vote, both predict the Coalition would get 37, and Labor would get 50 (ACNielsen) or 49 (Galaxy) per cent.

The ACNielsen poll played a little mischief. It found that Labour would do better in a Rudd v Costello contest (with a prediction of 61 to 39) than it would in a Rudd v Howard contest (58 to 42). It’s a finding that should Araldite the Prime Minister to his seat.

It was honest mischief in the Galaxy poll:

When voters were asked who was more honest, 44 per cent opted for Mr Rudd, while only 23 per cent chose Mr Howard.

Even among Coalition supporters, only 52 per cent backed Mr Howard when asked about honesty. “Perceived honesty is not John Howard’s strong suit and even his own supporters recognise he can be economical with the truth,” Galaxy’s principal David Briggs said.

What happens next

I had a software upgrade go horribly wrong over the weekend, and I lost the last six months of betting and polling data. A real pain! But it forced me to re-examine the previous polls. From an ACNielsen perspective, Labor cannot rest on its laurels. In previous election years, it was roughly at the same place it is now at this point in the cycle. In the last three elections, Labor crashed in the first or second week of September. As they say in the classics, it ain’t over until the fat lady sings.

It is too early to tell whether today’s TPP movement heralds the beginning of another election year decline for Labor. It may be random noise (with Labor bouncing around in the 58-61 per cent range), or it may be the start of something significant. As always with these things, you want to see a few subsequent data points before coming to a view.

The usual opinion poll graphs are here. You may need to hit the refresh or reload button on your browser to see the latest graphs.

Were an election held between last Thursday and Saturday, Labor would have received 61 per cent of the national two-party preferred vote. The ACNielsen poll of some 1400 people also predicted that the Coalition would have received 39 per cent. [SMH, Age]

In respect of the primary votes, the poll predicted 35 per cent for the Coalition and 50 per cent for Labor.

I was critical of the recent Morgan poll that predicted 61.5 per cent for Labor. Then I wrote, “Bollocks! There is no way that Labor would get 61.5 per cent of the national two party preferred vote, either last weekend — were an election held then — or in October/November this year when the next election is likely to be held.” I have not changed my view. A 61 per cent two-party preferred vote is implausible come the election.

With today’s ACNielsen poll, it would appear that at least some of the result can be attributed to unusually low preference flows to the Coalition. Typically, the Coalition receives around around 40 per cent of the non-Labor-non-Coalition vote in preferences at an election. In today’s poll it got just 27 per cent. If we adjust preference flows in line with recent elections, today’s result looks more like 59 to 41 per cent in Labor’s favour. Still significantly above any election result since the second world war.

Will the 2007 election be the ‘Ruddslide’ that today’s papers predict? I have answered this question a number of times this year, but I will say it again. It is simply too early to tell. I cannot predict from the polling to date whether 2007 will play out like 2001 or 2004 or not (see below). In Rudd’s favour he is starting the year in a better polling position than either Beazley or Latham. It makes the challenge for Howard in 2007 all the more difficult — but not impossible.

Headlines proclaiming that Howard’s spoiling strategy had failed are a little premature. Howard is playing a medium term game. It is the standard two pronged game: pander to the punters and slam the opposition at every possible turn. The effectiveness of Howard’s medium term strategy cannot be judged after a few short weeks. If previous election years are any guide, It was not until the middle of the year that a recovery trend (from Howard’s perspective) was evident. Howard only achieved positive polling territory from the middle to late in the third quarter of the year.

While I think it too early to call the 2007 election, I would rather be in Rudd’s shoes than Howard’s with polls like these. If Rudd is stll polling high 50s to low 60s in June-July it may well indeed be a Ruddslide.

The usual opinion poll graphs are here. You may need to hit the refresh or reload button on your browser to see the latest graphs.