This negative ad seems to work because it is not targeted at personalities, and it cites real figures with footnotes. Research in the US suggests that voters actually like negative ads such as these because they see them are providing information that they need to make their case. Not all voters though.

“This is true. All statements of fact with tangible $ to show how Labor is spending our money. It's a complete disgrace...so many unfinished or postponed projects.” Female, 25-34, Consultant

If voters have decided that you are not performing, then running on your record won’t convince them of anything. This is really what this ad shows. Those who want to re-elect Labor are convinced, but those who are unsure or who have made-up their minds are incensed.

“At least this ad is more positive. But positive on what? That state ALP has stuffed up, and so therefore they will work with Kevin Rudd? Surely they have some ideas of their own that they can employ? Anna Bligh also conveys no warmth in the ad. She doesn't smile, and every time I look at her I think of all the government spin I've seen since I moved to Queensland.” Male, 25-34, Teacher

What makes this ad fail is the way it caricatures Springborg. Labor has a credibility problem with respondents who see it as driven by spin rather than substance. It needs to be careful about what it says. Exaggeration is not and option, because even when it tells the truth many voters disbelieve it.

Comparisons with Barack Obama are also part of the exaggeration. Would Anna Bligh stand-up to a comparison with him?

There is also a sense that the government has been in too long and that it is repeating itself. The “Whingeing Wendy” style of ad first debuted in 1987 when it was used by Bob Hawke against the John Howard opposition. The ad exemplifies this.

“Too contrived and so obviously a fake” Male, 55-64, Tourism Operator

That’s Labor’s problem this Queensland election. We market-tested two ALP and two LNP ads with 665 Queenslanders from our regular panel, and both the ALP ads had a strongly negative effect.

They tended to reinforce the views of those who have already swung to the opposition, and energized traditional Liberal and National voters more than they affected those who are still uncommitted.

You can see commentary based on the responses in my article from today’s Australian.

This analysis will give you a little more depth in some areas.

The Australian reports today on Newspoll research which puts the Liberal National Party in a small lead over the ALP. It's not up on Newspoll's site yet, but it confirms Galaxy and our own findings.

Most notable movement in the figures appears to be the deterioration in Premier Anna Bligh's satisfaction rating plunging from 52% to 47%. Springborg satisfaction is 44%, almost the same figure as Bligh, although his disapproval rating has been rising, and a larger number of people are undecided about him than Bligh.

We uncovered strong dissatisfaction with Bligh in June last year. What we detected 9 months ago appears to have leaked into the general population. This suggests that the LNP lead is likely to increase.

While our respondents haven't moved much, the general populace has. My suspicion is that this is because our sample regularly thinks more deeply about politics than the average, but that when the average turns its mind to the same issues as an election approaches, it thinks similarly.

We are a week and a half out from the election and it is about now that most voters seriously start thinking about how they will vote. That means, assuming my theory is true, that the quantitative polls should start to capture more defections at the same time that the undecided vote collapses and tends to favour the LNP which should push the LNP vote relatively higher.

All of this assumes that nothing happens from outside the system between now and election day to radically alter perceptions. Cyclone Hamish making a landfall is just one of those potential factors; or one or other of the leaders making a major gaffe.

(Re-reading my earlier research, the only thing which doesn't appear accurate is my prediction that there would be no early election. Perhaps an example of a rogue population rather than rogue sampling, as I was essentially predicting the behaviour of a single person - the Premier.)