Well, the polls were always going to be right, but they were also completely wrong.
The Coalition have seen in a comfortable victory, but with a seat swing not as great as had been predicted, for which Labor can supposedly be grateful to Kevin Rudd.
But where does this sense that the victory for the LNP and the loss for the ALP is not as great as it could have been, come from? Well, from the polls of course, not from the campaigns themselves.
So let’s look at how accurate the polls were in relation to how the reportage of them may have influenced the actual outcome of the election. To do this, I will look at the polls taken in the marginal seats where the tabloid reporting of those polls was most concentration. And what we find is the polls that News Ltd reported were substantially inaccurate, but that reporting them clearly had an impact.
In a previous post, I had suggested that Abbott had a more focussed campaign on marginal seats in the largest cities and that it was in these cities that the News Ltd tabloid offensive had been most strident. A big part of the offensive, which has been quite overt, with editorials and advertorials supporting the government elect, was the reporting of the polls.
The election we have just had has been characterized by more obsessive reporting of the polls than has been seen in previous elections. Reporting a one-sided result of micro-polling arguably has had more influence over voting trends than political advertising. When the polls become a dog-whistle to marginal electorates that suggest changing winds, they can potentially inaugurate a bandwagon effect. Journalists from all media outlets, are quick to jump on this wagon, interviewing each other about the polls, and speculating on how great the loss might be for the side that looks like it is losing.
So let’s see what happened.
Case study one: Marginal seats in western Sydney
On August 23 The Daily Telegraph in Sydney reported that Labor was facing a western Sydney election wipeout’ with only the seat of Barton nominated as one that it would easily retain. Telephone polling was conducted of 550 voters in each of 7 seats, with a wipeout expected in Werriwa, Lindsay, Greenway and Banks and Reid, a close running in Parramatta. But in fact, of the 5 wipeout seats, Werriwa and Greenway were retained by Labor, Lindsay and Banks were won by the LNP with very small swings, and Reid remains undecided. If Reid does go to the LNP, it won’t be a landslide. Barton which was nominated as an easy ALP win, ended up in doubt with a big swing to the LNP.
Whilst these polls were well wide of the mark in their ability to support an ‘exclusive’ ‘wipeout’ story, they did correlate with one trend that gave them some validity … every seat polled except for Greenway, displayed a definite swing away from Labor, even if it did not always convert to an LNP seat.
Case Study 2: Marginal seats in Brisbane
Brisbane is the other marginal-seat city that had seen some of the heaviest campaigning by both Abbott and Rudd.
On August 24, the Courier Mail ran a front page story:
‘Poll Shock: Voters in Rudd’s own seat say: Time to Zip’ claiming he was trailing Bill Glasson for the seat of Griffith, 37% to 48% on first preferences.
On page 2 of the paper Madonna King went on to explain that the weekend Newspoll had surveyed 1382 voters across eight marginal LNP held seats in Brisbane to find that support for Labor across the seats had gone down 4% and the LNP up 8.3%.
Of course, Rudd won his seat easily, and there was a small swing to each of the other marginals, but as of today, this has averaged 3.28% across the 8 seats, well short of the up to 10% swing consolidation of mooted for the LNP .
Again however, in all these seats, there was a swing to the LNP across the board.
But it is difficult to say just how much these small but consistent swings across the board, may have been influenced by the reporting on polls in which much larger swings were forecast.
When a poll is reported, its influence is far more substantial that political advertising, because, whereas the ads are just spruiking a party line, a bundle of spin and opinion, a poll presents itself as an inert fact. However, the influence of these polls also depends on their presentation, and the editorial discourse that surrounds them.
For example, the article on the front page of the Daily Telegraph doesn’t even name the ‘exclusive poll’ which is later generalized to ‘polls’, adding a sense of authority to a Labor loss being inevitable. The news becomes: ‘the polls (plural) are saying this and saying that’. The audience can even become addicted to ‘officially’ hearing the latest about the polls, in the same way as we might obsess about checking our twitter account or email.
The ability for publishers and broadcasters to publish polls, stands at the apex of their influence. A trend emerges that looks to be unstoppable and can take on a life of its own. The polls quickly become trans-media stories. They are mainly commissioned by newspapers but get first billing in the television and radio news bulletins.
When they take on a life of their own, and our view of them converts from understanding to belief, we do not question how they came about in the first place. But they are based very small samples and their methodologies can vary so widely, even though they seek to legitimate themselves by publishing a ‘scientific’ margin for error. They have inbuilt biases, with many only surveying people with landlines, but the questions can be so selective and biased also. You can ask all kinds of questions about voting attention, but which results get published?
The range of questions that polling companies asks leaves room to accent one outcome over another. For example: ‘Which party are you currently leaning toward’, ‘how likely is it that you will change your mind’, ‘has your opinion (of Labor or the LNP) gone up or down since the election was called, ’ Many of the polls only give voters an option between ALP or LNP and not the Greens or the minor parties, which doesn’t pick up the complexity of preferential voting, or minor party affiliation.
Journalists will inevitably report the results that create the best story for selling papers or fit with the newspapers political editorializing. It is impossible to measure how poll mania can mobilize votes that would not have occurred if the polls were taken out of the election campaigns.
But given our electoral system embraces the idea of a media blackout in the last days of the campaign, and given the lessons we have learned about the flaws in polling itself, and given the concentration of media ownership in this country, it is worth reiterating that a poll blackout during election periods would do wonders for Australian democracy.