Note: there'll be a quick projections update later this morning and then nothing until Sunday night. If you wanna know, my current projections have the Liberals at 137 seats and the CPC at 127.
You win elections by convincing people to vote for you and then actually making sure they go out and cast a ballot. While I'm not one who believes "machines" to get the vote out are the most important factor -I worked for campaigns that were clearly disorganized and yet the vote got out- I have to acknowledge that it can make a small difference.
We now got multiple polls surveying people who already voted. That is to say sampling the 4.7 mio Canadians who voted during the advance polls, a record. The 5 polls are from Mainstreet, Angus-Reid (check the detailed pdf), Ekos, Insight West (BC only) and Innovative. Firms with different methodologies and various results for the voting intentions.
In order to remove as much "house effect" (i.e the bias introduced by the specific methodology of a pollster), I took the difference between the reported votes and the voting intentions from the polls from these firms. For instance for Ekos, while 30% declared wanting to vote for the Conservatives, 37.5% declared having done so during the advance vote, thus a differential of 7.5%.
The table below presents all the results as well as a simple average.
It seems the Tories are the ones who most consistently score higher among people who have already voted. The Liberals do particularly well according to Mainstreet but it's a disaster according to Ekos (interesting that the two extremes for the LPC are from the two firms using IVR technology).
The average is influenced by the BC only numbers from Insight West of course. But even removing this poll wouldn't change the overall picture. Also, Campaign Research shows that BC is the province -with Alberta- with the highest ratio of people who already voted over the number of people who intend to do so (40% versus 50%; remember polls always overestimate turnout because people who don't vote don't admit or straight up don't answer polls). This could matter because during the advance poll, the NDP was only starting its rise and was clearly 3rd in this province. That means it's possible many people who voted back then could have changed their mind. I don't think it's an issue major enough to cause a massive polling failure -mostly because people who vote in advance tend to be more committed anyway- but it could have a small influence and help the Tories.
As a reminder, in 2015, data shows the Liberals beat their polling numbers during the advance vote (and then went on to do the same on election day thanks to a late surge). It doesn't seem to be the case this year, unless Mainstreet is right (but if Mainstreet's numbers are right, the Liberals are also easily ahead in Ontario and could well be close to a majority).
I really hope Election Canada will release the advance turnout by riding today so that we can see where the increase is. In Quebec last year, we could clearly see the Liberal vote not coming out for instance. In BC 2 years ago, we could tell Surrey was going to flip - and it did. Surge in turnout is usually bad for incumbents.
So all in all, I don't see anything that would suggest the Liberals to beat their polling numbers. If anything, it seems the Conservatives are more likely to do so. Although it's also likely that the CPC has more very committed voters but will pick up a lot less of undecided on Monday (or NPD/Green voting strategically at the last minute).
