Oz Politics Blog

Regular visitors to Ozpolitics.info will know that I am more interested in the data trends over the medium term than the vagaries of poll-on-poll movements. One of the techniques I use to discern the polling trend is a monthly weighted aggregation. I take all of the polls for a month — Morgan face-to-face, Newspoll, ACNielsen and Galaxy — and recombine them into a single poll result for the month.

From March through July, a clear trend was emerging. That trend, if it continued, would have seen the Coalition in a winning position by mid December 2007.

This month’s aggregation is almost complete. It is only missing the last Morgan poll for the month.

The news? Well it looks like the trend back to the Coalition may have encountered an obstacle. Yes, it may simply be a perturbation in the data. But it is more likely to be something real.

To get a sense of the underlying movements in the polls, I aggregate the four main polls out each month: the Morgan face-to-face series, ACNielsen, Newspoll and Galaxy. Around 9300 people were polled in July.

The weighted aggregation yielded primary vote predictions for July of 47 per cent for Labor, 40.1 per cent for the Coalition, 6.4 per cent for the Greens and 6.5 per cent for others. The aggregated two-party preferred vote predictions for July were 55.5 per cent for Labor and 44.5 per cent for the Coalition. You may need to hit the refresh or reload button on your browser to see the latest graphs.

Aggregated, the polling data tells three stories.

First, Labor is currently well ahead and would win in a landslide if an election were called now.

Second, the gap is closing as the Coalition claws its way back for a possibly tight election contest in November or December 2007. Since Howard’s nadir in March 2007, the Coalition’s primary vote has improved 5.2 percentage points, the Labor primary vote is down 3.0 points and the Coalition’s TPP prediction is up 4.5 points. However, a couple of caveats are needed on this second story. There is no guarantee that the Coalition’s March to July trend will maintain its current pace until the end of the year. And for Howard to continue the trend, he will need to wind back the Labor primary vote, which has been remarkably resilient to date.

Third, the Greens and the minor parties are suffering a big squeeze. Since November 2006, the “anyone else” primary vote has dropped from almost 20 per cent to 13 per cent. If it continues to drop at the current rate, it will be around 12 per cent come the election. This drift in the “anyone else” primary vote means the Coalition and Labor require higher primary vote thresholds if they are to win government. The Coalition would need at least 45 per cent of the primary vote and Labor at least 42 per cent if it were to win government (assuming a 60-40 split in preference vote flows in Labor’s favour).

Note: for aggregation purposes I typically place a poll in the month for the last day of polling. However, because Morgan typically polls over two weekends, I seek to place its results in the month with the most days polled. I included the Morgan poll for 23, 24 and 30 June and 1 July in the June period.

The usual opinion poll graphs are here.

I am expecting two more polls in the last weekend of July (a Galaxy and a Morgan face-to-face over this weekend). Nonetheless, I found the July results to date so fascinating I decided to share them.

The shock news: John Winston Howard appears to be on-track for a late December election win. The trend-line since March 2007 is heading exactly where John Howard would want.

A fly in the ointment for John Howard is Labor’s sticky high primary vote. As you can see in the next graph, the Coalition has been improving its primary and two-party-preferred vote through a decline in the Green, other minor party and independent vote. However, Howard’s capacity to mine that vein appears to be diminishing (as shown by the trend-line in the next graph).

If Howard is to win the 2007 Federal election, he will need to start winding back the Labor primary vote in the coming months.

As they say in the Opera, it ain’t over until the fat lady sings. The 2007 election could end up being a lot closer than I had previously thought.

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.

[Note: I included the Morgan poll for 23, 24 and 30 June and 1 July in the June period as three of the four polling days where in June. I have not included the Morgan telephone polls].

So far this June, there have been five opinion polls, with a combined sample of over 6000 people. Collectively, these polls suggest a significant decline in Labor’s primary vote for the first time since Kevin Rudd’s appointment as Labor leader.

I looked at excluding the Galaxy poll from the first weekend in June as it appeared anomalous. Galaxy predicted a two-party preferred vote share for the Coalition of 47 per cent. The weighted average prediction from the four other polls for June was 42.9 per cent. However, even if you exclude the Galaxy poll from the first weekend in June, there appears to be an early June trend back towards the Government (see below).

Notwithstanding this movement, the June polls to date suggest that were an election held in the first weeks of June, it would have been a Labor landslide win.

I have not produced an aggregated poll report since March. This latest look at the monthly aggregations is revealing.

It appears that Labor reached its peak in March, with 50.1 per cent of the primary vote and 60.0 per cent of the two party preferred vote. The Coalition’s primary vote reached its nadir in March: 34.9 per cent.

Since March, the Coalition has regained some ground in its primary vote. In May it was 37.2 per cent. But this has been largely at the expense of the Greens and the other minor parties.

What should be worrying for the Coalition is the resilience of Labor’s primary vote. From its March peak of 50.1 per cent it has fallen to 49.7 per cent. That is a negligible fall of 0.2 per cent per month.

The trend in the two-party preferred vote share reflects both the modest gains the Coalition has made in its primary vote and the resilience of Labor’s primary vote. From a nadir of 40.0 per cent in March, the Coalition’s two-party preferred vote share has climbed to 41.8 per cent. If it continued to climb at this rate — ie. 0.9 percentage points per month — it would reach 47.2 per cent in time for a hypothetical mid-November election. According to the election calculator, if the Coalition secured 47.2 per cent of the TPP vote, Labor would win the 2007 election with around 83 seats.

However, the Coalition’s TPP vote share can only continue to climb at 0.9 per cent per month if the Coalition becomes more effective at eating into Labor’s primary vote share. Otherwise the Coalition’s TPP vote share will plateau and Labor will win the election in a landslide.

In summary, Kevin Rudd increased Labor’s primary vote by 10 percentage points and so far he has managed to keep them. He moved the TPP vote eight points, but that has since moved back by just under two points. And, he moved the Coalition primary vote by six points, which has since returned by a little over two points.

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.