Oz Politics Blog

Yesterday, Morgan released two poll results: a face-to-face poll of 850 electors on 13-14 October 2007, and a telephone poll of 598 electors on 17-18 October 2007. Last weekend’s face-to-face result was 57 per cent to 43 per cent in Labor’s favour. The Wednesday-Thursday phone poll result was 55.5 per cent to 44.5 per cent in Labor’s favour.

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.

The latest Morgan is out.

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.

Yesterday saw the first Morgan phone poll for some 6 weeks.

According to the pollster …

“This latest telephone Morgan Poll is a similar result to last Friday’s ‘face-to-face’ Morgan Poll — both would result in a massive ALP landslide.

“The 50% who said who said they would vote for Labor comprise of 22.5% ‘Soft ALP’ voters (electors who say Australia is ‘heading in the right direction’ yet say they would vote Labor if an election were held today) and 27.5% ‘Strong ALP’ voters (electors who say Australia is ‘heading in the wrong direction’ and would vote Labor if an election were held today).  The Morgan Poll considers ‘Soft ALP’ electors to be the key to the upcoming Federal election.

“The Morgan Polls are showing higher levels of support for the ALP than most other polls.  As this is of interest to all poll watchers it is worth noting that the Morgan Poll conducted ‘face-to-face’ includes those who don’t have telephones and those who only have mobile phones — generally an ALP skewed segment. 

“Results for telephone polls that exclude these people have an inherent L-NP bias.  We believe this explains the difference between the Morgan telephone poll and the Morgan ‘face-to-face’ poll.

“Morgan Polls are weighted by age within sex within area.  We don’t know how other public opinion polls are weighted.  If they are weighted by past vote (the Morgan Poll hasn’t done this since the late 1960s) then the vote would be closer, but would still result in an ALP landslide.”

This is the second poll in a row where Morgan has completely wiped any Coalition gain since Rudd’s previous peak in March 2007. The TPP Coalition-slump is largely driven by Labor’s primary vote figures, which are now stratospheric (according to Morgan). The Coalition primary vote is down, but not to the extent that the Labor vote is up. The other losers are the Greens and minor parties.

Just look at the Labor primary vote trend in the next graph.

It is worth speculating why Morgan is finding such a Labor-spike and Coalition-slump, when Newspoll and ACNielsen are not. I have been wondering whether Morgan’s face-to-face methodology may have resulted in an urban bias in the sample frame. The last Newspoll Quarterly suggested the bush was coming back to the Coalition. If rural and regional electorates are shoring up for the Coalition and the polls are flat-lining, it is possible the urban electorates might be slipping further away for the Coalition. Some of the recent individual-seat polling (such as North Sydney) would be consistent with an urban drift hypothesis. It is also possible that the differences are just random noise and in a couple of polls it will all wash away.

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.

Update: Simon Jackman makes some observations on the plausibility of the last two Morgan polls.

Morgan underestimated Coalition 2PP support in 2001 by 5.5 percentage points (poll on Nov 3/4, election on Nov 10). In 2004, I estimated the Morgan 2PP bias to be 4.9 percentage points. This would suggest subtracting 5pp from Morgan’s estimate of ALP 2PP vote intention would put you in the right ballpark (i.e., right around the last Newspoll).

…

Morgan is face-to-face, and so one suspicion is that Morgan skews towards Labor for reasons to do with geographic coverage and an odd patten of non-response, say, contra phone. That is, my conjecture is that face-to-face works well in urban areas, but tougher to do it well in rural and regional with a two day field period. Also, face-to-face may well generate a refusal pattern that sees younger people more willing to talk to the (generally young) interviewer (indeed, invite them into the home), inducing a Labor bias: again, just a conjecture.

The latest Morgan is out.

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.