Tuesday, January 24, 2006

Those seat projections

My seat projector, a modification of a spreadsheet developed by talkCanada, is surprisingly accurate I reckon. Some of you may be interested in how it preformed under actual regional numbers from tonight's election.

I plugged in the final unverified results from this 2006 Canadian election into the projector and came out with these seat projections:

Conservatives 122
Liberals 96
NDP 26

When in fact, the actual result is something like

Conservatives 124
Liberals 103
NDP 29

Not bad! The projector was most wrong in Quebec, where it overestimated Bloc support for the percentage of vote the Tories gained. In other words, it underestimated that Liberal and Tory support in Quebec would translate into seats for each respective party. And what projector could have predicted an independent in Quebec?

  • In Ontario, the projector was off by a very small number of seats. Conservatives gained 39 seats compared with 40 seats in the actual election (although Tony Clement won his seat by only 20-odd votes) whereas the projector underestimated Liberal seats by only 2 (56 actual vs. 54 projected). The NDP also got 1 more seat than was projected.
  • The projector was exactly right in Alberta and Sask, Newfoundland, PEI, and the Norther Territories.
  • In Manitoba, the projector underestimated Tory support by only 1 seat and thus overestimated NDP support by one seat. Ditto Nova Scotia and New Brunswick, except it overestimated a Tory seat at the expense of the Liberals.
  • The projector was quite off in BC (by about 4 seats). But what can you expect given a near exact popular vote outcome of last election?

No comments: