Oz Politics Blog

Yesterday was the 26th full day in the formal election campaign. There are only fourteen more days (13 sleeps) until the tally for the 2007 election.

The Coalition had no new policies listed on its website.

Labor announced GP super clinics for Cairns and Onkaparinka, and a commitment on Adelaide’s South Road.

ACNielsen on-line: Today’s Age reported ACNielsen’s third on-line poll. If this poll comes from the same panel as the previous polls, it is arguably the most significant poll so far this campaign. The use of a panel for ACNielsen’s on-line polls was reported with the first on-line poll. The advantage of panel based polling is that the movement between polls represents genuine movement within the panel group. Poll-on-poll differences are always significant and not subject to margins of error (although the application of the poll result as a population-wide prediction is subject to both sampling and systemic errors).

The most compelling observation from the latest ACNielsen poll is the absence of movement between polls. If today’s ACNielsen poll is from the same panel as the earlier on-line polls (see table below), it would suggest the election campaign to date has been a fizzer for the Coalition. There has been no narrowing of the gap between Labor and the Coalition during the campaign.

Update at 10am: I was wrong. I have just received this email, “I thought I better let you know that the Nielsen online poll actually draws a sample from their 90,000 Your Voice panel each poll, so the people doing the poll aren’t the same every poll … As nice as it would be to have a true longitudinal poll, unfortunately this aint it ;-)” Oh well, what can I say? I should have re-read the first report more closely before blogging. Anyway, maybe next election one of the polling companies could organise a longitudinal study (just for the period of the election campaign) of (say) 2000 randomly selected voters so that we can follow the impact of the campaign from week to week.

Finally, while I was wrong about the ACNielsen on-line polls using longitudinal panel data, I was probably not wrong on the flat line observation.

Morgan: Morgan released two poll results yesterday. Last weekend’s face-to-face poll of 990 voters yielded the (implausible in my view) headline prediction of 62 to 38 in Labor’s favour on a two-party preferred basis. I suspect Morgan agrees with me on the plausibility of this poll result, as Morgan did not include this data point in its graph at the top of its report. The chart of the longer-term Morgan face-to-face polling follows. You can see that this series is often quite volatile. I reckon the latest result is more likely to be noise associated with that volatility, than a significant population wide change in voting intentions.

Morgan’s Wednesday to Thursday phone poll of 552 voters yielded a prediction of 56 to 44 in Labor’s favour (which showed some statistically insignificant movement on the previous phone polls - see below).

The usual opinion poll graphs are here. You may need to hit the refresh or reload button on your browser to see the latest graphs.