Oz Politics Blog

The most recent ACNielsen poll, of some 1400 people between Thursday and Saturday, predicted a national two-party preferred vote of 52 per cent for Labor and 42 per cent for the Coalition (here, here and here).

The primary vote predictions were 42 per cent for the Coalition, 39 per cent for Labor and 10 per cent for the greens.

It is difficult to make more of the same sound interesting. But with these results, ACNielsen has been gently sloping in Labor’s direction for most of 2006.

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.

Commentators on this blog often make claims about the accuracy of various polling companies. I decided to take a look at the opinion polls within the final weeks of the last three Federal elections to see how the polling companies stacked up. I compared the predicted Coalition and Labor primary vote shares and the predicted two-party preferred (TPP) vote share with the final outcome. I also considered the predicted preference flows to the Coalition.

Where a prediction was within two percentage points of the final outcome, I scored it as a pass (shaded green in the table below). Where it was within three points I scored it as a possible pass (shaded yellow in the table below). Where it was off by three percentage points or more I scored it is a fail (shaded red in the table below).

Taking an average for the three elections, at a two per cent margin of error, ACNielsen was correct 74 per cent of the time. Newspoll was right 48 per cent of the time. And Morgan was correct 22 per cent of the time. If we take a three per cent margin, the scores improve to 96, 78 and 50 per cent respectively.

Is this conclusive? No. It is not conclusive. It was a small sample. And it was possible that three weeks out from an election actual voting behaviour may have been different to that which occurred on the election day. In addition, polling companies can change their methodology at any moment, and the methodologies used previously may not be used for the next election. Furthermore, there is always the risk that I have made a transcription or adding error. Still, it was interesting.

Update: as I had feared, there was a transcription error in my initial set of calculations. I had swapped the election primary votes for Labor and the Coalition in 1998. It has now been fixed. If there are any other errors, please let me know.

Over the past three weeks, there was little movement to report in the betting market for the next Federal election. The average of the four bookmakers suggested a 57.3 per cent probability of a Coalition government following the 2007 Federal election (down from 57.6 per cent on 16 August).

• For a Coalition win Centrebet would pay $1.57 and for a Labor win it would pay $2.25. The implied probability of a Coalition win at the next election is 58.9 per cent (unchanged)
• For a Coalition win IASBet would pay $1.67 and for a Labor win it would pay $2.20. The implied probability of a Coalition win at the next election is 56.8 per cent (unchanged)
• SportingBet is paying $1.67 for a Coalition win and $2.15 for a Labor win. The implied probability of a Coalition win at the next election is 56.3 per cent (unchanged)
• SportsBet is paying $1.65 for a Coalition win and $2.20 for a Labor win (was $1.60 and $2.25). The implied probability of a Coalition win at the next election is 57.1 per cent (down from 58.4 per cent)

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

The Parliamentary Library has released a research note on preference flows at Federal elections.

It provides good evidence to ignore some of the more excessive polling preference flows, which can sometimes go as high as 80 per cent for Labor. For example, with the 2004 election Morgan was close on the predicted primary votes, but it got its two-party preferred prediction wrong because it had determined that the Coalition would only get 22 per cent of preference flows. Personally, I reckon 60-62 per cent for Labor is a much more compelling assignment of the non-Labor and non-Coalition preferences when examining polling data.

The Parliamentary Library has also released an e-brief on the annual allowance for senators and members.

Just as the medium-term moving-averages from the individual pollsters are pretty much a flat line one or two points into Labor territory, so are the combined monthly polls.

(Note: this analysis is generous to the Coalition. For ACNielsen, I recalculate the TPP score using the preference flows at the last election. If I had not done this, and if I had used Morgan’s TPP score based on respondent preferences, the overall result would have been far more negative for the Coalition. Nonetheless, I think this is a fairer assessment of the current state of play).

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.