Party insider poll tips are for mugs

Mugs are what most of my colleagues in the media are. Most media are  reporting internal ALP polling which shows that the government has 52 percent of the two-party vote, but is doing poorly in marginals in Queensland, New South Wales and West Australia. They give this polling credence and say that the election will be close.

At the same time they are also quoting a Newspoll, two-thirds completed, which shows the parties on 50 percent each, and a Galaxy poll which is 52/48 in the government's favour.

Out of all of this they mostly seem to decide to run with the Newspoll figure for the country, even though it is the odd-one out, and the ALP poll for marginals, even though there are only 80 responses from each marginal electorate polled.

Why do they do this? Because it suits the narrative they have already seized on, which is that the government will lose. I guess it should be the preferred story, given it has the most suspense, so readers and listeners will have to tune-in tomorrow, and an archetypal theme - the under-dog triumphs.

The Newspoll figure is an honest attempt to measure sentiment, but could easily be out by 2.45% which is the maximum margin of error on the likely sample size of 1600 voters. It is also likely to under-represent the younger vote, because they are the hardest to find, and the ones that you typically get later in the polling after you have filled the easier categories. Pollsters will compensate for this by weighting the under-represented groups, but that inflates the margin of error.

The ALP figure is obviously trumped-up and should be discarded. The ALP would be doing tracking polling in each of the 20 electorates that they sampled up until at least last week. Those samples would probably be rolling samples of at least 250 each.

There is no reason for them to do a poll of 20 marginals and only sample 80 from each electorate unless it was designed to be "leaked", a word which in this case really means "released". A poll of that size would be absolutely no use in determining where to direct resources and what messages to use, which are the major reasons that parties poll seats at all.

The ALP are terrified that their lead will slip away with enough voters deciding to lodge a protest vote because they think Labor will win and can do so safely.

There is a huge protest vote out there at the moment, of course, which is why the Greens vote is so high, but most of that will come back to Labor.

Great media management from the ALP, but very poor form of my colleagues to allow themselves to be used like this.

It may be a close result, but none of the polls to date show it being much closer than a margin of around 6 seats, which is not that unusual in Australian election results.


Share this article on your favourite social bookmarking sites:
Digg! Reddit!! Google! Facebook! StumbleUpon! Twitter!