The latest Ipsos poll (via ThreeHundredEight.com) strengthens the suspicion that the Liberal collapse in the latest week of the EKOS poll was due to statistical noise. The interesting numbers in this poll come from:
- Atlantic Canada, where the Grits lead by 12, compared to the NDP being miles ahead in the latest EKOS and the Tories crushing everyone else in the previous Ipsos;
- Ontario, where the virtual tie contrasts with the large Tory lead in the latest EKOS and the Liberal lead in the previous Ipsos;
- BC, where Ipsos has numbers broadly in line with the latest EKOS and my poll average, while the previous Ipsos poll showed a whopping Conservative lead.
Including this poll causes 3 seats in Atlantic Canada to move from the Tories back to the Liberals:
CON - 133
LIB - 80
BQ - 55
NDP - 40
Nice round numbers, aren't they? Well, here are some more for you: my national poll averages are: CON 35.0, LIB 28.0, NDP 16.5, BQ 10.5, Other 10.0; the Tory advantage is therefore 7.0%.
Note: Despite the slower summer depreciation for polls, only the 3 most recent EKOS releases, this Ipsos Reid survey and the latest (somewhat dated) Harris-Decima poll are included. This is partially because Angus Reid, which usually has a poll at the end of every month, went AWOL at the end of June. Also, Nanos and Harris-Decima, who publish their results irregularly, don't have recent polls. The relative paucity of data means that projections may temporarily exhibit increased volatility.