The Adventures of QueerPenguin

Sitting on a poll

If all else fails, play the puppy-dog eyes. Sorry J-Ho, ain't gonna work. If you lose this election it's your own fault for not 'graciously' (I'm being generous) handing the leadership over to Costello last year and putting yourself out to pasture when you should have done.

However, I'm still far from convinced you will, in fact, lose this election.

Why? It's all about the psephology. What a nice word is psephology. Psephology. Say it out loud a couple of times. Psephology. Now throw in the silent p, if you haven't already been doing so, just for fun. Psephology.

I digress.

Opinion polls should be neither instantly dismissed nor taken as gospel. The pollies take them seriously even though they pretend not to. A good way to determine their accuracy is to compare what was the final opinion poll recorded on the last day before the election, with the actual results. Very rough summary for 2004 (rounding up): Actual result was Coalition 53, Labor 47 on 2pp. Nielsen called it 52/48 in their final poll, Newspoll 50/50. Not bad, but not spot-on.

I recently consulted with psephologist extraordinaire William Bowe over at The Poll Bludger to get an unbiased reality check on opinion polls that currently have Howard judged, juried and executed.

(M)ost poll-watchers will tell you that a 52-48 result is the measure of what Labor will need for a clear win...

Pollsters generally say they have a 3 per cent margin of error within a 95% confidence interval, which means they will be within 3% of accuracy 95% of the
time.

Now, there needs to be uniform swing of 3.3% for the Coalition to lose its majority, and 4.6% for Labor to win a clear majority. Or, in simpler terms, using this fantastic 2007 election calculator, Labor will need to win at least 52.5% of the 2pp vote this year to form majority government. This equates to Labor winning 18 seats - 6 in NSW, 3 in SA, 2 in Victoria, Queensland, WA and Tasmania and one in the NT - as well as not losing any seats they currently hold.

Were this to actually happen, Coalition casualties would include Howard, Malcolm Turnbull and Jackie Kelly, among others.

Not to mention, this would also mean Labor holds onto two WA seats, Swan and Cowan, to which it's desperately clinging with a margin of less than 1%. This in a state where all the kids are into AWAs but potential Labor Mum isn't so keen to let them play with them.

You can see how this is a really, really, really big ask.

Given the 3% margin of error, the major opinion polls will still need to have Labor on at least 55% of the 2pp at their final pre-election polls before I start to believe any 'annihilation' will take place.

Sure, there are a bunch of Howard voters right now who are giving serious thought to offering the Kruddster a shot, but 6 months is still a very long way to go. Howard will still have plenty of bribes and other rabbits, no matter what he otherwise claims, and don't forget that around 30% of voters don't make up their mind with absolutely certainty about who to vote for until the last week - and the overwhelming majority opt for the incumbent government, particularly when the economy is performing strongly (1949 and 1996 elections being exeptions to the rule).

So let's all keep a calm head for the moment. Re-read that card you carry around in your pocket to remind yourself, 'Always say/do the opposite of what John Howard says/does'. And we'll talk on election day.

Labels: 2007 federal election, John Howard, opinion polls, psephology

posted by Sam @ Wednesday, May 23, 2007