Far be it from the Poll Bludger to tell Rupert Murdoch how to do business, but one can only wonder if News Limited is getting value for money by having three different titles commission three different agencies to conduct three different sets of opinion polls. For better or worse, The Australian’s Newspoll, the Courier-Mail’s Galaxy Research and now the Sunday Mail’s TNS all point in the same direction, although TNS predicts a somewhat more modest Labor victory than the other two. After distribution of the undecided and unresponsive, Labor is on 45 per cent of the vote compared with 20 per cent for the Liberals and 16 per cent for the Nationals, with Labor leading 55-45 on two-party preferred.
I have never seen a News Ltd poll that overestimates the eventual Green vote. It seems curious that the polls always sample Greens results on the negative side. The only explanation is that the samples are not stratified correctly. The curious part about this is that all of the polling companies commisionerd by News Ltd show the same type of sampling error. Why is this?
TNS says Labor is on 55
Galaxy says Labor is on 54
Newspoll says Labor is on 58
So lets split the diff and say Labor is on about 55.6, which does actually sound about right.
In the Sunday Mail today, Prof. Paul Williams said “I think Labor is probably feeling quite rosy about snatching Caloundra off the Liberals as well.”
In other words he is predicting that the Liberals will have between 4 and 5 seats. Interesting prediction.
I don’t see Caloundra happening for Labor. The party doesn’t seem to be travelling especially well on the Sunshine Coast, plus Mark McArdle is now the incumbent and should poll better than he did as a new candidate because of the benefits of incumbency (Americans call it ‘sophomore surge’). If Labor couldn’t win Caloundra last time despite a change of Liberal candidate, they’re not going to win it now.
Back to the topic and 55 or so for Labor does sound right. Though again this poll seems to give people the option of selecting between Liberal and National, when obviously only one will be on the ballot paper. And interestingly, they don’t exclude don’t knows and refused when calculating primary vote percentages.
I’ve seen undecideds at 20% in Newspoll. How does a Newspoll determine how these people will vote. Can you really assume they will vote similiar to the other poll respondents?? Is this level 20% historically normal for polling??
Just looked at the Weekend Australian in HK airport, and noticed a basic error of fact in reporting on the Qld election – how much care does it show they take when they can’t get basic facts right?
The thing I noticed, and it’s little but any political junkie would pick it immediately, was in discussing the Greens’ preferences for the seat of Barron River in the current Qld election, and the story read that the Greens’ preferences in the 1995 election turned out to be important as Labor lost the seat and then their majority after the Mundingburra by-election.
The newspaper said that the National Party won Barron River in 1995, except that in fact the Libs won it (see http://www.parliament.qld.gov.au/view/historical/documents/memberBio/Warwick.htm). You might say that, well, a Coalition member won it, what does it really matter? It probably doesn’t matter very much except in showing the care with which the story was written.