Polls matter. Much of the flesh of an election year grows on a skeleton made up of public opinion polls. The whole political community hangs on the ups and downs of polls as they are published from week to week. In many ways, polls are markers on the campaign journey.
The major newspapers put great store on their own polls because of the guaranteed news content they provide. The Australian has its Newspoll and the Melbourne Age and the Sydney Morning Herald have the ACNielsen poll.
There are other polls that get coverage too, including the Galaxy and Morgan polls. Peter Brent’s chapter on the polls in The Crikey Guide to the 2007 Federal Election, shortly to hit the bookstores, is well worth a look for a fuller account.
Explaining the movement of the polls always has its traps for commentators. They can jump around. One useful guide is the Reuters Poll Trend, which tries to even out volatility and provide a moving average of Newspoll, ACNielsen and Morgan. Reuters confirms that Labor remains well ahead. In fact the government has trailed by more than 10% in each Poll Trend since Rudd became Labor leader.
The commercial nature of the polls and the investment in them by the major media outlets mean that their political journalists can suffer from tunnel vision. There is little problem with this when the message of the different polls seems similar, as appears to be the case at the moment with the voting attention of respondents for the forthcoming election.
But when there is an apparent difference, as there has been over the last fortnight with the results on preferred prime minister, there is not enough comparative analysis. Newspoll showed Howard drawing almost level with Rudd two weeks ago and The Australian made a big fuss about this trend.
A week later ACNielsen reported that Rudd was still well ahead on this same question. Yet on the following day The Australian’s team of journalists continued to discuss its poll from the previous week without a mention of the competitor poll that was showing something different. This is misleading and may be explained by commercial imperatives, which get in the way of more informative analysis.
For all their technical traps — margins of error and design questions first of all — the second most important feature of the polls is that they are only as good as