We are more than two weeks into this federal campaign, one week away from the debates, and we must admit it: it's a boring election. From a polling perspective at least (what I do on this site) but I believe we could make the argument that this is the case when looking at the campaigns and policies as well. And I'm not the only one saying this:
The Globe and Mail's @RobertFife says the campaign has been marked by a lack of vision. #ctvqp #cdnpoli #elxn43 pic.twitter.com/P8mBagVFgl— CTV'S QUESTION PERIOD (@ctvqp) September 29, 2019
But this blog is about polling, so let's stick to that. And before any smartass tells me that elections aren't there to be exciting and I'm missing the point of democracy, etc. Yes, you are right.
Nevertheless, here are the most recent projections. I haven't updated with the Mainstreet and Nanos numbers of Sunday, but they pretty much cancelled out (Nanos was good for the Tories while Mainstreet was good for the Liberals). Honestly though, I almost waste my time every day entering the new numbers as barely anything has changed. See the analysis after the infographics.
Proj 28.09.2019 by bryanbreguet on Scribd
A boring election
Let's look at the voting intentions of the main two parties, the Conservatives and Liberals. The graph below includes every poll published. For the daily trackers of Nanos and Mainstreet, I only included one data point every 3 days.
Don't be fooled, there is almost no variation. It looks more volatile than it is because I cut off the axis at 25% (and I used a polynomial trend). The CPC has, in particular, been remarkably constant. The standard deviation is only 1.3%. Just to put this in perspective, a party at 35% has margins of error of 2.2% for a sample size of 1748 (the average sample size this election), that corresponds to a standard deviation of 1.1%. In other words: the volatility observed is perfectly in line with the CPC being actually constant and the variations are just caused by sampling (realistically, it's also probably caused by variations in sampling methods, but let's forget about it).
Just to be super clear: it is quite likely that the Conservatives were at 34-35% 2 weeks ago and haven't moved. People like me (and others) obsess about looking at polls every day, but the truth is: most likely nothing has happened so far.
For the Liberals, the standard deviation is 1.7%. That's higher than what random sampling would predict, so the LPC did seem to have moved. I have already shown in a previous post that the blackface incident probably caused the Liberals to drop by 2-3 points, an impact that was likely short lived.
Let's look now at the NDP and Green, the two parties fighting for 3rd place (in the popular vote at least).
First observation: for an election that was supposed to be about climate change, there sure is a lack of a green wave. Sure the party of Elizabeth May would likely get its best result ever if the election were tomorrow, but this party is still around 10%. More importantly, it appears that Jagmeet Singh has managed to give his party some momentum. While pre-election polls had the Green passing the NDP, the gap between the two has now grown in favour of the Neo-democrats.
It should be noted however that pollsters do not agree on the NDP. IVR polls (automatic phone calls) have the NDP about 4 points lower than online polls. This explains the bigger volatility observed. Given that the NDP is around 11-13% in average, the standard deviation should be 0.8%. In reality it has been 2.2%, by far the most volatile series between all the parties. This is potentially problematic as there could be a pretty major difference between the NDP below 10% or at 13-14%. That could well be the few seats the Liberals need to win in order to get another majority. We'll see if polls start converging later on.
What about Quebec? After all, it seems this is the one province where things are moving. Has the Bloc, in particular, really been rising since the start of this election (Graph is in French, sorry)?
We can indeed see this rise. It's not a wave but it's there. This is especially important because the Bloc starts winning many seats (the "payoff zone") above 22%. The Bloc at 25% would likely prevent Trudeau from winning much more than 40 seats in Quebec, something that would make his objective of a second majority really at risk (since he's pretty much guaranteed, at this point at least, to lose many seats elsewhere).
The Liberals are remarkably constant while the Tories seem to be declining. That might not matter much for them however as their seats are fairly safe and the possible gains are limited anyway (I've said it before but almost-insurance-broker Andrew Scheer will win or lose in Ontario, not in Quebec. Even though he absolutely needs the help from the Bloc to prevent the Liberals from winning 50+ seats).
So, boring election so far. What's next? Next week is full of debates (that even Trudeau is attending, yay!), so things might change. If that doesn't cut it, this election will become a game of turnout. Whoever will be able to motivate its voters the most will win. The 2015 election was pretty much tied three ways until the Liberals started taking the lead in early October. Will it be the same this year? Do people really only start paying attention with about 2 weeks to go? We'll see.
There still are quite a lot of undecided (the exact number varies based on the polling method, with IVR having much lower rates), especially in Quebec. So I'd be surprised if we really didn't see any movement until October 21st.
