Monday, September 13, 2010

What if Australia's election system was in Canada?

Now that the business over who gets to form what government in Australia is over, I thought it interesting to take a peek at what would happen if Canada employed the same system that Australia uses to elect MPs to Parliament, which is a form of alternative voting called instant-runoff voting (IRV).

The situation goes like this: the "first preference" votes are much like what we see here in Canada, with parties often getting under 50% of the vote. An example is the electorate of Melbourne, which for first-preference voting had the Labor candidate Cath Botwell gain 38.1% of the vote, and the Green Party's Adam Brandt on 36.3%.

Under Canada's first-past-the-post system, this would be a near-miss for the Green Party, with Labour retaining the seat with only 38% of the vote. We have many ridings like this in Canada, none so stark as the riding of Gatineau, which has a member elected on only 29% of the vote.

But under Australia's IRV system, something different happens. Back at the polling station, the voters marked down their preferred candidates. The first-preference votes are those which people chose that are their first preference to represent the constituency, ending up with what we had in Melbourne. Given the results of the first-preference votes, starting from the last candidate, their votes are distributed among their second-preference votes, eliminated, and so on until a candidate has a majority.

This is when the second-preference votes come in; those candidates marked down on the ballots that had their first-preference votes for candidates who ended up eliminated marked down their "second preference," which in this case ended up being either Labor or Green. Those votes are then allocated by the second-preference vote to the two remaining candidates, until one is declared a winner when they get over 50% of the votes.

What we end up with is Cath Botwell, who was leading before with 38% of the vote, loses 44-56 against Adam Brandt, who attained more second-preference votes than Botwell. Brandt then goes on to be Melbourne's MP, with a majority vote in favour of him.

So, what if this were applied to Canadian electorates? I hope to demonstrate a small example of what may actually occur, using my own riding of Burlington.

Now, obviously I don't have the second-preference of all voters in Burlington, but the miracle of Ekos' second-preference polling gives me a chance to see what it may look like. The way I do this is simple: using the second-preference by party, the eliminated candidate's votes will be allocated to the top-two candidates based on those numbers. Those that don't indicate any other preference don't get counted.

Let's apply this to Burlington:

First Preference

Mike Wallace - Conservative - 28,469 - 48.45%
Paddy Torsney - Liberal - 19,524 - 33.22%
David Laird - New Democrat - 6,600 - 11.23%
Marnie Mellish - Green - 4,170 - 7.10%

After eliminating Laird and Mellish, we use Ekos' tables:

Second Preference

Mike Wallace - Conservative - 29,454 - 57.1%
Paddy Torsney - Liberal - 22,133 - 42.9%

And Wallace is elected MP, as sure as under FPTP, but this time with a majority of voters electing him, instead of only 48%. Interesting, no?

Let's try the riding I mentioned earlier, Gatineau:

First Preference

Richard Nadeau - Bloc Quebecois - 15,050 - 29.13%
Francoise Boivin - New Democratic - 13,467 - 26.07%
Michel Simard - Liberal - 13,111 - 25.38%
Denis Tasse - Conservative - 8,700 - 16.84%
David Inglis - Green - 1,334 - 2.58%

Second Preference

Michel Simard - Liberal - 18,006 - 53.6%
Richard Nadeau - Bloc Quebecois - 15,581 - 46.4%

Michel Simard becomes the duly elected MP for Gatineau - but wait, wasn't Francoise Boivin of the NDP in second?

Well, yes, she was - but while recounting both the Conservative and Green votes, their second-preferences broke more in favour of the Liberals and the NDP, an it was relatively close to begin with. Simard had already moved ahead by the time the other two were eliminated, making Simard in the top-two, ahead of Boivin. This is what happens in these types of vote counts. Again - interesting, no?

And doing this on a national scale, there's an even more interesting make-up using the 2008 results:

CON - 134
LIB - 87
BQ - 45
NDP - 39
IND - 2

Some close races include Edmonton-Stratchona (51.6 for NDP Linda Duncan, to 48.4 for Con Rahim Jaffer); Sudbury (50.3 for NDP Glenn Thibeault, to 49.7 for Lib Diane Marleau); Egmont, PEI (50.8 for Lib Keith Milligan, to 49.2 for Con Gail Shea); and South Sore-St. Margaret's (50.6 for NDP Gordon Earle, to 49.4 for Con Gerald Keddy).

Though the make-up isn't too different, the result of the change is clear: the regionalist Bloc, who often win by slim margins in certain seats, have their caucus reduced; the Liberals and the NDP both benefit from more proportional voting, attaining some seats that they lost to the Conservatives and the Bloc by razor-thing margins

I don't claim this is perfect - far from it, actually. However, I can only use the information at my disposal. And unlike the Australian electorate, we have quite quite a few parties (though if you broke apart of the Coalition, you'd get something similar to ours), so the difference isn't as exacerbated. Nevertheless, it does introduce a level playing field in our constituencies, requiring candidates to reach a 50% threshold to represent the district, something I consider important for the fairness of our system.

No comments:

Post a Comment