Jan 30th 2014: Abacus and Ekos polls show Liberals favorite to win, but regional numbers very different

With the new session that just started, we've got two new polls. One from Abacus and one from Ekos. Both have a relatively large sample and agree on the general numbers. However, the provincial breakdowns are quite different. More about this later.

Doing an average of the two polls, we get the following projections (remember that since Election Canada still hasn't published a transposition of the 2011 results onto the new electoral map, I'm still using the old map. I contacted Election Canada and they say it's coming soon).

Voting intentions; Seat projections with 95% confidence intervals; Chances of winning the election

The Liberals of Justin Trudeau would naturally be the favorites to win the election. Still, Harper and the Conservatives have 13% chances of winning. While this is low, it's still a lot better than in most recent polls of the last couple of months. Also, please notice that Ekos actually has the Tories ahead if they use their "likely voter" model. I use the "regular" numbers as Ekos' model has been disappointing to say the least. While I command this firm for being very open and at least trying to create such a model, I just can't see myself using their numbers as is. After all, in the last BC election, the NDP's lead was actually increasing when using the likely voter model... Nevertheless, the idea of an underestimation of the CPC in the polls is not unlikely. It happened in 2008 and especially in 2011. Remember though that the simulations should take care of that.

While the Liberals have a pretty comfortable lead, a majority seems out of reach. Out of the 5000 simulations, there are 12 cases where Trudeau would get a majority. The only way that would happen would be if both polls clearly underestimated the Grits and the electoral system was working in their favor.

Let's talk about the provincial numbers. I understand the samples are small (well, except for Quebec and Ontario), but some of the differences between the two polls are just too important. For instance in BC, Abacus has the Conservatives first with 36%, while Ekos has them 3rd (almost 4th!) at only 16.9%! In the Prairies, Abacus has the Tories 8 points ahead of the Liberals while Ekos has them only 1 point ahead. Finally in Ontario, Abacus shows the Liberals enjoying a 7 points lead, Ekos only has a 1-2 points lead. Given the importance of Ontario, such a difference could ultimately be the factor between a Harper or a Trudeau government.
Of course, outside of BC, the variations are not completely out of the ordinary and can certainly be explained by the normal statistical variation (even in Ontario, the difference for the Tories between the two polls is not significant, you can see that yourself here). Still, given that my model uses provincial numbers, we need to remember there is quite a lot of uncertainty. But again, the simulations are there to take of that, to some extent.

In Quebec, the Bloc keeps doing its yo-yo. I talked about this before, but this party has a sort of threshold at 20% (use the Canada simulator to see convicne yourself). Below it, the Bloc wins almost no seat. Above it and it's game on in a lot more ridings. Just look at the 95% confidence interval for the seats, it goes from 1 to 13! My projections thus keep alternating between a Bloc with 3-5 seats or 15-20. This phenomenon could well be (one of) key of the next election. Indeed, if the Bloc colapses, we now have a two ways race between the NDP and the LPC (note: it's crazy how the Quebec political landscape evolves so quickly. Over the last 10 years, we had LPC vs Bloc, Bloc only, Bloc vs CPC and now possibly LPC vs NDP). But if this party somehow manages to climb back up a little bit, then both federal parties would probably win a lot less seats. For Trudeau and his hope of a majority, a Bloc under 20% could be essential (although the NDP is clearly more affected).

In conclusion, the numbers for Trudeau look very good, although less than 6 months ago. Their lead in term of seats is also highly dependent on the crazy high numbers in Atlantic. At the same time, I'm sure the Conservatives are satisfied with these polls as well. Keeping power is definitely not out of the question (good campaign, a good vote targeting, etc). As for the NDP, the fact is thet they are still domnating in Quebec (and BC depending on the poll) and aren't that far behind should be a source of satisfaction for them. Of course, Trudeau is really hurting them, especially it seems in the Prairies. Next election could well be a true three way races where everything can happen. Another minority government seems very likely at this point.