Friday, September 2, 2011

Blunt Object's Provincial Projections

Over to the right, you'll see a nice little picture of all three major party logos in Ontario. Click it, and you'll be introduced to the current draft of my projections page, which will be updated when necessary.

At the current moment, I'm projecting a PC minority - 48-39-20, to be exact. I may post riding-by-riding projections soon, but it's a bit of a hassle.

However, I'd like to note now that the projection system isn't mine, but my blog partner's, Teddy Boragina. I've just modified his Electomatic system to work as an analog for the Ontario election, with full ridings included. It works!


  1. A PC minority implies the NDP will back them. Keep in mind that, contrary to what Harper tried to convince us these past few years, the governing party REGARDLESS OFF ITS SEAT COUNT technically has first stab at forming a government. If the NDP were to back the Liberals the PC's would be kept out.

    And if they were dumb enough to back the PC's, what a credibility nightmare for them when the cuts come (see Britain).

  2. The ElectoMatic was designed to work in the "Canadian 2.5" party system, and thus is just as accurate for Federal projections as it is for Ontario projections. It has a remarkable record in BC as well. The "problems" come when you get smaller and smaller. The Yukon for example, and even Prince Edward Island, have ridings that are so small, they local factors make any ElectoMatic projections partially useless. Being off by 6 seats might not be much in Ontario, but in the Yukon, that is a third of the legislature.

  3. My "expected" error ratio was always 2% plus 5 seats. This would mean 11 Federal seats (96.4% accurate), 7 in Ontario, 7 in BC, 6 in PEI, and 5 in the Yukon (73.7% accurate)

  4. ...Remember too that these are flip seats. IE if all of them were for the same party, the worst expected case would see a 10-5-4 result in the Yukon turn into a 5-5-9 result. A party would need the error number, above the majority line, to be assured of a win. In the Yukon this means 16 seats, which, has never happened. In Ontario, this is 61 seats, which is certainly possible.

    The other problem is with regional, or, swing-heavy changes. IE the GTA and the Bloc vote. In fact, if you over-swing the GTA and the Bloc-to-NDP my federal projections are actually pretty accurate, even showing properly which seats within the GTA the Tories won. The problem was the improper vote distribution within the GTA and outside of it.

    I'll be working with Volkov to ensure that any similar split is accounted for in the coming provincial election.