I’ve written before that I question Ekos’ methodology. Traditionally, they always seem to favour the “underdog” in an election poll. In 2008, they missed the results of the Conservative Party by just over 3%, which is a significant amount, especially since they consider their margin of error to be roughly +/- 1.8%.
Obviously things happen in the course of 24 hours, and, since publishing poll results is banned on election day, this is the last poll they will release.
The advantage to using Ekos as a source for seat projections is Ekos breaks down their results showing support and subsample size in urban areas (Vancouver, Calgary, Toronto, Ottawa, and Montreal). This is very useful, because these areas have large concentrations of seats, and also can have very different levels of support than extra-urban areas.
So, after plugging the Ekos results into my election prediction algorithm, here is what we see:
Conservative: 150 (149-150)
New Democrat: 116 (114-116)
Liberal: 37 (37-40)
Bloc Quebecois: 5 (5-5)
Ekos also asks people how certain they are to vote. This is crucial because just because someone says they would vote a certain way does not mean they are actually going to show up; so I plugged those numbers from Ekos into my prediction algorithm as well.
Conservative: 149 (145-152)
NDP: 118 (114-122)
Liberal: 36 (36-37)
Bloc: 5 (5-5)
That’s the math, based upon raw data. At this point, intuition has to come into the mix, because we have a number of un-measurable factors at play as well as, as I have posted previously, there are huge assumptions build into any prediction algorithm.
The various talking heads are mentioning that Quebec’s NDP slate of candidates is, well, dubious at best, and the NDP ground organization is, well, virtually non-existant. This will affect their Get Out The Vote (GOTV) campaign; simply because they do not have the personnel at the grassroots level. Clearly they are going to put various GOTV plans in place, however without local personnel, they can only do so much. Therefore, for the NDP, it will come down to how motivated the voters are to get out and actually show up at the polling station.
Ekos also has a tendency to poll low numbers for the Conservatives. They published 34% in 2008 as well, and the Conservatives ended up coming in at 37. If the Conservatives are adjusted up by 3% with each of the other three parties dropped down by just 1%, we see this:
Conservative: 159
New Democrat: 114 (111-114)
Liberal: 29 (29-32)
Bloc: 6
I am still not convinced that the NDP will take 60-ish seats in Quebec, especially outside of Montreal where the population density is higher. I think the NDP will take about 13 seats in Montreal, and a few seats outside of there, possibly as many as 10. That puts the NDP at taking 23 seats in Quebec, and, assuming the numbers hold elsewhere, at 76. The NDP non-pickups in Quebec will be someone else’s gain, of course. The Bloc overwhelmingly the source of the NDP votes, so we have to speculate a little bit on what is going to happen with the Bloc’s GOTV campaign. It is quite possible that the Bloc actually ends up inadvertently doing the NDP GOTV work for them, as the NDP surge happened fairly late in the campaign. Who knows? Assuming the Bloc gets out Bloc vote only, and assuming that the NDP is not able to get out their vote as well, this puts a very large number of Quebec seats into play. If the NDP takes 23, the split could just happen such that the Conservatives hold on to their existing Quebec seats, and maybe pick up a couple of extras, with the Liberals losing a chunk of Quebec seats.
My prediction for Quebec: Bloc: 30, NDP: 23, Liberal: 10, CPC 12
So here is my national prediction:
Conservative 156 (155-160)
NDP: 76 (70-80)
Liberal: 46 (40-50)
Bloc: 30 (25-35)
By the barest of margins, yes, I am indeed predicting a Conservative majority.