Making sense of the polls

Making sense of the polls

Federal Poll Analysis - D-day Minus Eleven

This Week’s Polls

As before, the table below shows the poll, two party result, change from last issue, fieldwork dates and sample size. Newspoll and Morgan both decreased their sample sizes to their usual levels.

poll table.

Newspoll recorded a 3% rise in the Labor primary to 37%, but the Coalition’s primary was unchanged at 47%, and so Labor failed to make up much ground after preferences.

The online polls such as Essential are behaving very differently from the other polls at this election. At the 2010 election, Essential’s final poll was virtually spot-on on primary votes. However, a good rule in analysing polling is not to believe a poll that is far from the overall trend, as this poll certainly is. Newspoll has a strong track record when considering its final pre-election polls against the actual results, while Essential has only been tested at the 2010 election. All other polling shows that the Coalition would have won an election held last weekend easily.

Update 28 August: A national ReachTEL robopoll of over 3,500 respondents conducted Monday night 26 August has the Coalition’s lead at 53-47, unchanged from the last ReachTEL on 10 August. Other than Essential, this is the best polling news Labor has had all campaign. It indicates that the trend against them has stalled. Also, ReachTEL has about a 1% lean to the Coalition compared with other polls, so this poll might be equivalent to a 52-48 from another pollster.

There have been many individual seat polls, all showing dismal results for Labor. Confusingly, it appears that Galaxy is using robopolling for its individual seat polls, and the results in Victorian and Western Sydney marginals agreed well with ReachTEL robopolling in showing a disaster for Labor. ReachTEL did more polling in Tasmania, and showed a Labor wipe-out there.

The most surprising individual seat results came from Newspoll’s live interviewing, which showed Rudd losing his seat of Griffith 52-48, concurring with a Lonergan robopoll which I had heavily criticised. Newspoll also did a survey of the eight most marginal Coalition held Queensland seats, and found a 7% swing to the Coalition in those seats. These seat polls are suggesting at least 60-40 to the Coalition in Queensland, yet the Queensland breakdowns of national polling are about 53-47 to the Coalition according to the Poll Bludger’s Bludgertrack. I cannot explain the discrepancy.

Notes on these Polls

  • Nielsen had Rudd’s approval down 5% to 43% and his disapproval up 4% to 51% for a net approval of -8. Nielsen’s approvals tend to be kinder than Newspoll’s. The Coalition’s lead on the economy was a massive 21 points at 58-37, up 3% from two weeks ago. Labor led 50-38 on broadband and 47-39 on paid parental leave. However, it is the economy that matters. Nielsen still does respondent-allocated preferences, though these are not used as the headline figure. Using respondent-allocated preferences, the Coalition’s lead is 52-48, rather than 53-47.

  • Newspoll had Rudd’s approval up 1 to 36% and his disapproval down 2 to 52% for a net approval of -16. For Labor to have a chance, Rudd needs to get his Newspoll approval rating into at least the 40’s.

  • Morgan’s respondent-allocated preferences were 51.5-48.5 to the Coalition, compared with 52.5-47.5 using the previous election’s results.

  • Essential had Rudd’s approval down 4% to 41% and his disapproval up 2% to 45% for a net approval of -4. Labor’s paid parental leave scheme was favoured over the Coalition’s by 35-24. A majority of voters say that it is at least somewhat unlikely that either party can pay for its promises without cutting spending. 60% of Coalition supporters and 46% of Labor supporters say they will definitely not change their minds on who they will vote for.

Why is the Economy such a Negative for Labor?

I think there are two main reasons why Labor is getting comprehensively thrashed on best party to manage the economy. First, Labor won government at the 2007 election, just before the financial crisis hit. Although other countries have suffered from this crisis far more than Australia, voters look back at the Howard years as a time when the economy was booming, Australia had a huge surplus, and most people felt more prosperous than they do now. Economic data is currently soft, with the unemployment rate trending up. In these circumstances, it is difficult for a government to retain office.

The second reason for Labor’s economic troubles is poor communication. Peter Brent has articles on how former Treasurer Wayne Swan was a poor communicator, and on mistakes Labor made in their first year in office. The economy is the most important issue for voters, and the government is vulnerable on the debt and deficit attack lines. The government has not defended itself adequately on these Coalition messages, and this is why they appear headed for an election defeat.

Does Labor have any Chance?

A week before the election was called, the polls were at about a 50-50 tie; now, they are at 53-47 to the Coalition, a 3% movement against Labor in four weeks. However, trends can, and often do, go into reverse. Other sources of hope for Labor are the Essential poll, and respondent-allocated preferences from both Nielsen and Morgan, which show a closer race than the previous election preferences. Many voters do not decide until the final week, so if Labor can perform well in the next two weeks, it could still be close. However, I would now say that the chance of a Labor win is less than 10%.