Although Ontario — where the election race is as close as can be — provides an interesting campaign, the result there is not likely to be the defining factor in this year’s election.
As far as influencing the outcome, Quebec is more likely to be the key, which could give Thomas Mulcair the edge he needs.
There were fewer polls this week than last, and voting intentions remained mostly stable. We got a new Abacus poll as well as a Forum. There was also a Quebec-only Crop and the usual Nanos update (with the usual four-week average, which makes it impossible for us to use their numbers).
Aggregating these new polls (along with the Forum, Mainstreet and Leger of last week, all conducted during the last two weeks) to obtain the most accurate picture of the situation, we end up with the following projections.
They use past election results as well as the current polls in order to predict the winner in the 338 ridings. They include regional and incumbency effects. The confidence intervals and the chances of winning are obtained through the use of 5,000 simulations that account for the uncertainty of the polls as well as for the distribution of the vote and the electoral system. In other words, these simulations try to include every possible scenario, given the information we currently have.
Seat projections, as of Aug. 22:
Remember, we do not simply average polls. We adjust them slightly to account for their tendency to overestimate smaller parties and underestimate incumbents. Specifically, we do not allocate undecideds proportionally. Rather, we give more to the incumbent Conservatives and then an equal share to the Liberals and the New Democrats (in Quebec, however, the NDP and Liberals receive more).
It may look like tinkering with the data, but this method has worked well every time we have used it, including in the recent Alberta election. Moreover, polls greatly underestimated the Tories in 2008 and 2011 (by about three percentage points). If your objective is to predict the outcome, you can’t simply take the polls at face value.
As you can see, the situation is still very much like what we had last week, with the Conservatives and NDP fighting for first place and the Liberals behind. Chances are now in Mulcair’s favour, thanks to the increase in Quebec where two polls, Crop and Abacus, placed the NDP as high as 47 per cent (numbers high enough for the NDP to almost completely sweep the province).
The drop of the Bloc Québécois means the party is now projected not to win any seats. If that were to happen, one might wonder if it wouldn’t be the end for the party. Especially since the Parti Québécois isn’t doing much better, according to the poll from Crop. The return of Gilles Duceppe is definitely not working as well as the Bloc expected.
Ontario remains incredibly divided, with all three parties statistically tied in the low 30s. Unlike Quebec where the New Democrats have a 100 per cent chance of finishing first, Ontario sees the Conservatives winning the most seats 73 per cent of the time, the Liberals 25 per cent and the NDP about two per cent. This is one of the few provinces where any of the three parties can theoretically finish first.
However, just because Ontario provides the most seats (121 out of 338) doesn’t mean the province will necessarily determine the colour of the next government. Specifically, if we define “determining the election” as a situation where the winning party is different if we exclude a specific province from this election’s calculations, we get the following table:
Excluding Quebec changes the outcome most often by far. This is obviously because of the NDP, which usually wins about half its seat total in la Belle Province (in our simulations). So, if we remove Quebec’s numbers and add only the seats in the rest of Canada, the NDP doesn’t finish first very often.
Interestingly, while Quebec could well determine who becomes prime minister, it is also one of the least competitive provinces in this year’s election.
What that means is that both Harper and Trudeau should be very worried by the lead Mulcair currently enjoys — it will be hard to overcome.
What may be more surprising is that excluding British Columbia changes the outcome almost as often as Ontario. The chance that it could very well give the NDP the edge over the Conservatives is important. If voting intentions remain stable until election day, media and pundits will most likely have to wait longer before calling the election. Indeed, British Columbia could be crucial.
ALSO ON HUFFPOST:
Information about the new polls mentioned in this article:
1. Forum poll conducted between August 17th and 19th. 1473 respondents contacted by phone (IVR). Margins of error are 3 percent 19 times out of 20. Question asked: “A federal election has been called for October 19. Which party are you most likely to vote for in this election?”.
2. Abacus poll conducted between August 14th and 17th. 1439 respondents from the Ipsos online panel. A probabilistic sample of this size has margins of error of 3% 19 times out of 20. Question asked: “If a federal election was held tomorrow, which one of the following parties would you vote for in your constituency?”
3. Crop poll conducted between August 12th and 17th in Quebec. 1000 respondents contacted online. A probabilistic sample of this size has margins of error of 3.1% 19 times out of 20. Question asked (translated from french): If a federal election was held today in Canada, for which of the following party would you vote for?”
Bryan Breguet has a B.Sc in economics of politics and a M.sc in economics from the University of Montreal. He founded TooCloseToCall.ca in 2010 where he provides electoral analysis and projections. He has collaborated with the National Post, Journal de Montreal and l’Actualité.
He will provide analysis and updates for The Huffington Post Canada throughout the federal election campaign. For riding by riding projections, visit his interactive simulator.