April 13th: Mid-campaign projections

We just had the two debates and some people think the real campaign starts now. It might be true. So I decided to do mid-campaign projections. What you have displayed here is the 95% confidence interval for the seats projections. To calculate them, I used an average of most recent polls conducted since April 1st. I then calculated the margins of error for each province and apply them to each party (one at a time, not simultaneously). As I've said previously, my intervals are much more narrow than on other projections websites. It is because they are made using an average of polls, instead of cherry-picking the best and worst poll for every party. If I was to do that, the intervals would be much wider. Just today, we had one poll from Ekos showing a short Conservative lead of 5-points, and a Compass poll with the Tories at 45% and the Liberals at 24%!

The big difference with the start of the election is that the Conservatives have lost some ground, especially in the Atlantic and Ontario. The results being that instead of projecting a majority in average, these mid-campaign projections put the Tories at a maximum of a majority, therefore less likely. We'll see if the good performance of Harper during the English debate (he was really weak in French) will help him and his party climb back to majority territory.

For the Liberals, Michael Ignatieff is doing a relatively good campaign, but his lackluster performance in the English debate might put a stop to his small rise. Anyway, even in the best case scenario, the Liberals would be slightly ahead of 2008 and down from Paul Martin's results of 2006. One of the problems is that, so far, the Liberals have only been able to steal some votes to the NDP, especially in Ontario. It's ok to prevent a Tories majority, but not for winning the election. On a side note, and I don't want to start a debate about coalitions, but I'm getting more and more convinced that Ignatieff's real objectives is to force Harper to another minority and then defeat him quickly. Then, he could get to TRY to form the government. Ignatieff is probably betting that even without a formal coalition, the NDP and the Bloc wouldn't bring down a Liberal government. Again, not trying to start a debate, just food for thoughts.

For the NDP, the best and worst case scenarios aren't that different. We are talking more symbolism here, with being able to cross the 40-seats thresholdor getting more than 1 seat in Quebec. However, Layton performed very well in the debates (both, and he's the only leader who got two good performances), so I wouldn't be surprised to see this party climb to around 20% nationally.

The Green are not projected to win one seat even in the best case scenario. But in their case, the battleground is really the riding of Elizabeth May. So this party might get one seat the same way Quebec Solidaire got its in Quebec: no increase in the polls, but a lot of hard work in one riding.

As for the Bloc, the 95% CI is a little bit too narrow to my taste. But with as many as 6-8 polls used for the average, it means the MOE are only 2.02%. And for the Bloc, being at 35% or 39% doesn't change so many things as the federalist parties are splitting the remaining votes almost evenly. If you ask me, looking at the close races, I would say the Bloc should be projected between 43-49 seats.

Here are the riding-by-riding projections. There are for the average of course, not the best and/or worst case scenario.