Australian Financial Review

14 October 2006

Lies & Statistics, pg63    see also accompanying notes

Voting intentions seem to be more sustainable when satisfaction ratings are low, Peter Brent writes.

Labor MP Bob McMullan recently observed that although published opinion polls over the past six months show the federal opposition ahead, most Australians wouldn't know it because of the tone of the press coverage.

As if to prove his point, the Sydney Morning Herald this past week reported a survey showing 54 to 46 in Labor's favour - thumping landslide territory - under the headline: Labor vote creeping up on Howard. (The graph was titled Labor catching up).

Many journalists can't reconcile their current political perceptions with the survey data. A few lay the specific charge that the ALP is doing well but leader Kim Beazley isn't, with satisfaction ratings worse than Mark Latham's and not much better than Simon Crean's. Voting intentions coupled with low satisfaction ratings are not sustainable, they say.

Are they right? Newspoll has conducted 450-odd surveys since 1985. I averaged the pollsters' satisfaction ratings and voting intentions (called here ASR and AVI) for each of the opposition leaders who made it to an election, and compared them with the election results.

Latham, from December 2003 to October 2004, generated easily the highest ASR of 52, while the lowest, 26, belonged to Andrew Peacock, 1989-90. Latham's AVI was 50.6, but his eventual vote was a disappointing 47.3. Conversely, Peacock's AVI was a low 48.7 but on election day he got 50.1. John Howard (1985-87) clocked a low 31 ASR and a doldrums-level AVI of 47.1, but achieved a respectable actual vote of 49.2. John Hewson (1990-March 1993) was more like Latham, highly approved at 41 and an AVI of 52.4, but just 48.6 at the ballot box.

The above numbers run counter to the thesis. Does any evidence support it? Yes, Beazley in 1996-98 got a high ASR of 44, a low AVI of 47.9 and a vote of 51.1. And Howard in 1995-96 gives perhaps neutral evidence, with high numbers all round: ASR 46, AVI 53.1 and eventual vote 53.6.

One problem with satisfaction ratings is that it takes in all voters, not just those who might change their vote. Probably a larger issue is respondents putting on their commentators hats - is Beazley making political inroads? McMullan's point becomes relevant here.

But if the Newspoll data points anywhere, it is in a surprising direction, suggesting that survey voting intentions are more sustainable if accompanied by low - rather than high - satisfaction ratings. It might be that high approval can artificially boost voting intentions - possibly applicable to the Latham bubble.

Beazley (2005-) is interesting. Over the first seven months, his ASR was a respectable 41 and his AVI a low 48.3. But over the past year his ASR dropped to 33 while his AVI rose to 50.6. And for the eight most recent Newspolls his AVI was 51.0. This consistent coupling of low satisfaction and high vote is unprecedented, so we will have to wait and see. But we should remember that votes - not "satisfaction" - decide elections.

Peter Brent is publisher of mumble.com.au, a website that looks at electoral behaviour.

Back to main page