I followed the same methodology that I did before, but my results this time should be questioned even more. The Environics article suggests that they consider the GTA portion of the 905 area to include 22 seats, of which the Conservatives won 6. When I looked at the ridings listed on the Election Prediction Project, and compared to maps, I wasn't quite able to figure out which 22 ridings they were thinking of.
Anyway, the Environics poll results for the GTA-905 were:
The number in brackets is the vote share in the 2004 election for the 22 ridings that I judged to be in GTA-905. In those 22 ridings, the Liberals won 18 seats and hold 18 now, the Conservatives won 4 and hold 3 now, the NDP have none and there is one independent.
Following the same process I used last night, I produced these results: Liberals 15, Conservatives 6, NDP 1. That's a pick-up of only 3 for the CPC. I'd say that's disappointing given their poll gains. I'd attribute it to a lot of very close races all going Liberal. I'd also say that the NDP gain could be noise caused by the low precision of the poll data. (I.e., no decimal points... not that Environics should have included any since the sample size was small with a 4.5% margin of error.)
As I mentioned a major flaw in my approach is that I was looking at a different set of 22 seats than what Environics considers to be the GTA-905. Perhaps I was mistaken to include Oshawa and Whitby-Oshawa, and instead should have included Clarington--Scugog--Uxbridge and York-Simcoe. However, even if I made that adjustment, I would still have 17 Liberal and 5 CPC wins (in 2004) compared to the 16-6 in the Environics set of ridings. I can't find another Liberal win to drop.
However, given all that, I also don't know how rigorous Environics was about restricting its 905 polling to just those 22 ridings and making sure that they had a good distribution of respondents across each one. So, this is all just fun and games.