Modeling tomorrow’s Australian federal election
This fascinating election is just around the corner. Not only have psephologists modeled how the aggregate “two-party” vote plays out in individual districts; the media actually pay attention! (If your average US newspaper paid attention to district-level margins and two-party votes, people would take presidential and congressional elections for the uninteresting, predictable “contests” they really are.)
What’s more, Australia uses IRV (alternative vote, etc.) to elect the House of Representatives. So the best models have to take into account which of the two major coalitions/parties will receive preference transfers from small party candidates. Malcolm Mackerras - I met him in 2006 while working in electoral reform - developed just such a model, which is the basis for this calculator at the ABC. His “electoral pendulum” doesn’t work so well in the US, where the big challenge is controlling for incumbency, not vote transfers.
Fruits and Votes, who has numerous Australian readers, has been following the election. He reports a tightening projected margin, but Labor is projected to win the “two-party preferred” vote 52-48 over PM Howard’s Coalition.
After he was written off as a ‘has-been’, John Howard’s last week of campaigning will give the Coalition hope of pulling off a miraculous victory tomorrow.
Labor is still in front and favourite to win the 2007 election but the latest Newspoll survey is showing a late surge to the Howard Government, particularly in Queensland and Western Australia.
Newspoll’s two-party preferred figure, based on preference flows at the 2004 election, has the Labor Party in front by 52 per cent to the Coalition’s 48 per cent.
The sample size was larger than average - 2614.
Last week the two-party preferred figure was 54 for Labor and 46 per cent for the Coalition.
Yes, there’s been some tightening of the margin, and I’m a fairly new observer of Australian electoral politics, but this looks like a Labor win. By American standards, 52-48 is a big spread. In fact, that’s how well the Republicans did in 2002, just 14 months after 9/11. (That says something about what the Democrats’ 54% in 2006 meant - and what any GOP nominee faces in 2008.)
Bob Richard on 25 Nov 2007 at 12:40 pm #
Labor did indeed win as predicted here. There’s a fascinating follow-up discussion at the Canadian site Liberals for Electoral Reform about whether AV/IRV would lead to proportional representation in Canada, or whether it would actually be a detour from the road to PR. I doubt that this question is answerable with any certainty, but it’s clearly important here in the U.S. as well.
Jack on 25 Nov 2007 at 2:04 pm #
Some people might say, “Didn’t we just try this IRV thing? And now our voting system is still broken?” Then again, I suspect those people are likely “no” voters in any referendum.
I doubt there’s a “road” to PR in any teleological sense. What IRV-first gives you are (1) familiarity with ranking candidates and (2) STV-compatible voting equipment in theory. The lack of equipment, as you know, has kept many willing jurisdictions from implementing either.
It would be nice if machines were designed to run the actual STV algorithm on the basis of a few simple variables: district magnitude and total valid poll, for instance. Then you’d have machines to use with both systems.
The Doctor on 17 Dec 2007 at 4:55 am #
Jack,
you do realise that there are multiple versions of STV, for example both the ACT and Tasmania HoA use Hare-Clarke with Robson rotation, which means any thing used count IRV votes will not straight forwardly translate into a Hare-Clarke algorithm.
But any ranked voting system would at least stop those computerised abacuses the US uses from being so easily tampered as all votes would have to stored before being totalled.
Jack on 17 Dec 2007 at 2:38 pm #
Good point about algorithm variants. I did not know that ACT and Tasmania use Robson rotation.